17 TH Symposium

Lecture Notes in Civil Engineering
Manish Shrikhande
Pankaj Agarwal
P. C. Ashwin Kumar Editors
Proceedings of
17th Symposium
on Earthquake
Engineering
(Vol. 4)
Volume 332
Series Editors
Marco di Prisco, Politecnico di Milano, Milano, Italy
Sheng-Hong Chen, School of Water Resources and Hydropower Engineering,
Wuhan University, Wuhan, China
Ioannis Vayas, Institute of Steel Structures, National Technical University of
Athens, Athens, Greece
Sanjay Kumar Shukla, School of Engineering, Edith Cowan University, Joondalup,
WA, Australia
Anuj Sharma, Iowa State University, Ames, IA, USA
Nagesh Kumar, Department of Civil Engineering, Indian Institute of Science
Bangalore, Bengaluru, Karnataka, India
Chien Ming Wang, School of Civil Engineering, The University of Queensland,
Brisbane, QLD, Australia
Lecture Notes in Civil Engineering (LNCE) publishes the latest developments in
Civil Engineering—quickly, informally and in top quality. Though original research
reported in proceedings and post-proceedings represents the core of LNCE, edited
volumes of exceptionally high quality and interest may also be considered for publi-
cation. Volumes published in LNCE embrace all aspects and subfields of, as well as
new challenges in, Civil Engineering. Topics in the series include:
• Construction and Structural Mechanics
• Building Materials
• Concrete, Steel and Timber Structures
• Geotechnical Engineering
• Earthquake Engineering
• Coastal Engineering
• Ocean and Offshore Engineering; Ships and Floating Structures
• Hydraulics, Hydrology and Water Resources Engineering
• Environmental Engineering and Sustainability
• Structural Health and Monitoring
• Surveying and Geographical Information Systems
• Indoor Environments
• Transportation and Traffic
• Risk Analysis
• Safety and Security
To submit a proposal or request further information, please contact the appropriate

Springer Editor:
– Pierpaolo Riva at pierpaolo.riva@springer.com (Europe and Americas);
– Swati Meherishi at swati.meherishi@springer.com (Asia—except China,
Australia, and New Zealand);
– Wayne Hu at wayne.hu@springer.com (China).
All books in the series now indexed by Scopus and EI Compendex database!
Manish Shrikhande · Pankaj Agarwal ·
P. C. Ashwin Kumar
Editors
Proceedings of 17th
Symposium on Earthquake
Engineering (Vol. 4)
Editors
Manish Shrikhande Pankaj Agarwal
Department of Earthquake Engineering Department of Earthquake Engineering
Indian Institute of Technology Roorkee Indian Institute of Technology Roorkee
Roorkee, Uttarakhand, India Roorkee, Uttarakhand, India
P. C. Ashwin Kumar
Department of Earthquake Engineering
Indian Institute of Technology Roorkee
Roorkee, Uttarakhand, India
ISSN 2366-2557 ISSN 2366-2565 (electronic)

ISBN 978-981-99-1458-6 ISBN 978-981-99-1459-3 (eBook)
https://doi.org/10.1007/978-981-99-1459-3
© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature
Singapore Pte Ltd. 2023
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Contents
Probabilistic Arias Intensity Maps of Uttarakhand State (India) . . . . . . . 1

Kunal Gupta and Neelima Satyam
Statistical Analysis of Seismicity Parameters and Completeness
Period of Earthquake Catalog for Sree Padmanabhaswamy
Temple, Thiruvananthapuram District, Kerala . . . . . . . . . . . . . . . . . . . . . . . 17
M. P. Hari Padmanabhan, Siddhardha R, Sreevalsa Kolathayar,
and Ramakrishna Hegde
Effect of Open Stories on Expected Seismic Losses in Hilly Buildings . . . 27
Yati Aggarwal and Sandip Kumar Saha
Influence of Source Offset on the Resolution of Dispersion Image . . . . . . 41
Ashhad Imam, Virendra Kumar, Keshav Kumar Sharma,
and Alvin Harison
Co-seismic Deformation of Iran, 2021 Earthquake Using DInSAR
Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Hardeep, A. Bahuguna, K. Arun Saraf, and J. Das
Three-dimensional Crustal Velocity Structure of Tehri, Garhwal
Himalaya . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
R. Modi, S. Mukhopadhyay, and M. L. Sharma
Correlation Between Cone Tip Resistance and Shear Wave
Velocity for Quaternary Alluvium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
P. Mishra, A. Paul, and P. Chakrabortty
Ground Motion Predictive Equations and Its Applicability
in North-Eastern Indian Region: A Critical Appraisal . . . . . . . . . . . . . . . . 87
P. Kumar and S. S. Kumar
Influence of Site-City Interaction on the Response of Buildings
on Trapezoidal Basin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Neeraj Kumar, J. P. Narayan, Pooja Lohchab, and Sanjay Kumar
v
vi Contents
Assessment of Double Resonance from Microtremor Observations

for Jammu Region in India . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Abdullah Ansari, Falak Zahoor, K. S. Rao, A. K. Jain, Aashi Pal,
Neeraj Kumar, Sakib Majid Hajam, Pallavi Shukla, Krishna Sharma,
Faizan Fayaz, Mir Akhtar Yousuf, Shakir Riyaz, and Umer Altaf Khan
Influence of Epistemic Uncertainty on the Seismic Vulnerability
of Indian Code-Compliant RC Frame Building . . . . . . . . . . . . . . . . . . . . . . . 127
Kaushik Gondaliya, Vishisht Bhaiya, Sandip Vasanwala, Atul Desai,
and Jignesh Amin
Spatial Distribution of the Gutenberg-Richter Parameters
and Fractal Dimension and Their Correlations in Northeast India
and Its Vicinity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
R. B. S. Yadav, P. Chauhan, M. Sandhu, R. Kumar, and V. Singh
Seismic Hazard Analysis Considering the Effect of the Shape, Size,
and EQ Distribution of Seismic Sources for Different Locations
in Sikkim, NE India . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
Borah Niranjan, Mondal Joy Kumar, and Kumar Abhishek
A Comparative Study on Application of Machine Learning
Algorithms in Ground Motion Prediction Equations . . . . . . . . . . . . . . . . . . 163
A. Ahmed and M. Gade
Study of an Anomalous Behavior of Atmospheric Parameters—As
an Earthquake Precursors for Himalayan Region Earthquakes . . . . . . . . 175
M. Senthil Kumar and Natarajan Venkatanathan
Probabilistic Seismic Hazard Assessment of North East India . . . . . . . . . . 187
C. Lallawmawma, M. L. Sharma, and J. Das
Experimental Investigation of Seismic Response of Hybrid Shear
Wall with External Energy Dissipating Reinforcement . . . . . . . . . . . . . . . . 205
Ankhiparna Guha, S. R. Dash, and Goutam Mondal
1D Velocity Model for NW India in and Around Delhi . . . . . . . . . . . . . . . . 217
Deepak Kumar, G. Suresh, S. C. Gupta, M. L. Sharma,
and Hasbi Ash Shiddiqi
Anomalous Deviations in Atmospheric Parameters
as Pre-earthquake Signals-A Case Study on Sumatra
Region Earthquakes (M ≥ 6.0) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
Ramya Jeyaraman and N. Venkatanathan
Spatial Distribution of Stress Orientation by Inversion of Focal
Mechanism Solutions Using MSATSI: A Case Study Across Japan
Trench . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
Sucheta Das, Sandeep, Sonia Devi, Himanshu Mittal, Praveen Kumar,
and Monika
Contents vii
Seismic Landslide Hazard Assessment of Mandi Town, Himachal

Pradesh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
A. Kothiala, P. S. Nayek, Maheshreddy Gade, and U. V. Kala
Infill Wall Effect on Seismic Analysis of Reinforced
Concrete Buildings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
C. H. Sirajudheen and Behera Dibyadarshi
Geotechnical Seismic Base Isolation Using Rubber Sand
Mixtures—Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
S. L. Divyasree, K. M. Jithin, and Renjitha Mary Varghese
Development of Soil Amplification Factors Using 1D and 2D
Ground Response Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
A. Sharma and S. Adhikary
Probabilistic Seismic Hazard Assessment for
Assam, North-East India . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
M. Borah, M. L. Sharma, and R. N. Dubey
Effect of Randomness of Slip and Source Time Function
on Pseudo-Dynamically Simulated Ground Motion Characteristics . . . . . 325
Vishal, J. P. Narayan, and L. Joshi
Quantification of Ridge-Weathering Effects
on the Simulated Ground Motion Characteristics
Across 2D and 3D Topography Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
Vishal and J. P. Narayan
Developing a Comprehensive Historical Tsunami Database
for the Indian Ocean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
Nazeel Sabah and Daya Shanker
Seismically Induced Landslide Hazard Analyses for a Road
Corridor in the Lower Himalayas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
A. Tyagi, R. R. Nath, M. L. Sharma, and J. Das
Assessment of Proxy-Based V s30 Estimation in Roorkee,
Uttarakhand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 379
M. Srivastava and M. L. Sharma
Ground Motion Prediction Equation for NW Himalaya Region . . . . . . . . 389
Vandana, Harendra K. Dadhich, Himanshu Mittal, and O. P. Mishra
Identification of Strong Motion Generation Area of the 2019
Hualien Earthquake . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
Saurabh Sharma, A. Joshi, S. Singh, C. M. Lin, C. H. Kuo,
and K. L. Wen
Role of SH Wave in the Mapping of Shallow Subsurface . . . . . . . . . . . . . . . 411
Jyoti Singh, A. Joshi, Saurabh Sharma, and Mohit Pandey
viii Contents
A Variational Mode Decomposition Approach for Modal

Identification of Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421
S. Gupta and S. Kaloni
Geo-Factor Inference Modelling with Empirical Susceptibility
Weights Approach for GIS-Based Seismic Hazard Mapping
of Thiruvananthapuram City . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433
Madhu Mohan Velapgy and E. S. M. Suresh
Octave Tool for Probabilistic Seismic Hazard Assessment . . . . . . . . . . . . . 451
D. R. Majhi and M. Shrikhande
Site Amplification Study Using Strong Motion Data Recorded
at Various Stations in India from Far-Field Earthquakes . . . . . . . . . . . . . . 461
Sireesha Jaladi, Babita Sharma, Himanshu Mittal, and O. P. Mishra
ROSERS—A Deep Learning Framework for Earthquake Early
Warning and Its Interpretation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473
Jawad Fayaz
Early Warning System: An Efficient Earthquake Disaster
Mitigation Tool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 487
P. Kumar, Govind Rathore, Kamal, M. L. Sharma, R. S. Jakka,
Pratibha, and A. Kumar
Application of Regression Techniques for Preparing
a Homogeneous Earthquake Catalog—An Overview . . . . . . . . . . . . . . . . . 501
Ranjit Das, H. R. Wason, and Claudio Meneses
A Cluster-Based Seismic Risk Assessment: Economic Loss Using
GIS for Jaipur Sub-Urban Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 511
V. Anand, M. Mahatab, A. Sharma, D. Raj, M. K. Jat, R. Sarkar,
and S. Pal
Modelling of Empirical Accelerograms of 1999 Chamoli
Earthquake (Himalaya) Using a Modified Hybrid Approach . . . . . . . . . . . 523
A. Sharma, D. Kumar, and A. Paul
Development and Implementation of a Regional Earthquake Early
Warning System in Northern India . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 537
Govind Rathore, Pankaj Kumar, Mukat Lal Sharma, Kamal,
Ravi Sankar Jakka, and Ashok Kumar
A Critical Review of Existing Building Regulations and Bye-Laws
in Hilly Regions of India . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545
P. Das Choudhury and D. Raj
Microseismic Analysis Using Event Count and Potency
Displacement for Stability Evaluation of an Underground Cavern . . . . . . 563
Vikalp Kumar, V. R. Balasubramaniam, and K. S. Divyalakshmi
Contents ix
Characteristics of Strong Ground Motions for Delhi National

Capital Region (NCR) Using Small to Moderate Size Earthquakes . . . . . 577
Abhishek, Manisha Sandhu, and Babita Sharma
Detection of Liquefaction Phenomenon from the 2015 Nuweiba
Earthquake Using Remote Sensing Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 587
Hrik Chaudhury, Abhishek Kumar, and Rishikesh Bharti
Estimation of Site Amplification Factor and Predominant
Frequency in and Around Panchkula City, Haryana, India . . . . . . . . . . . . 599
M. Sandhu, R. B. S. Yadav, D. Kumar, and Abhishek
Use of GIS for Hypsometric Analysis for Determining Erosion
Proneness of Mandakini Watershed, Lesser Himalaya,
Uttarakhand, North India . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 613
James Xavier Paul and Daya Shanker
Understanding the Structure and Tectonic Configuration of Bengal
Basin for Earthquake Magnitude Prediction . . . . . . . . . . . . . . . . . . . . . . . . . 625
Mir Fazlul Karim and Daya Shanker
Exploring an Alternate Perspective of the Importance Factor
for Seismic Design of Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 639
Narsiram Gurjar and Dhiman Basu
Simplified Damping Modification Factor for Vertical
Response Spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 655
Ravi Kanth Sriwastav and Dhiman Basu
December 01, 2020, Haridwar, Earthquake: Fault Plane Solution
and Tectonic Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 669
Pooja Mahto and S. C. Gupta
Local Seismicity Around Tehri Dam, Garhwal Himalaya . . . . . . . . . . . . . . 683
M. L. Sharma, S. C. Gupta, J. P. Narayan, J. Das, A. Sen, S. K. Jain,
A. K. Jindal, Subhash Patel, Prajawal Tandekar, Avichal Rastogi,
Rajeev Vishnoi, Atul Jain, Virendra Singh, and S. K. Saxena
About the Editors
Dr. Manish Shrikhande is Professor and Head of the Department of Earthquake

Engineering at Indian Institute of Technology Roorkee. His current research interests
are vibration monitoring and control, seismic risk and mitigation, and computational
mechanics.
Dr. Pankaj Agarwal is Professor at the Department of Earthquake Engineering,

Indian Institute of Technology Roorkee. His research interests are earthquake-
resistant design of masonry and RC structures, post-damage assessment survey
of earthquake-affected areas, risk assessment, cyclic testing of structures, seismic
instrumentation in multi-storied buildings, health monitoring, and damage detection
in buildings. He is continuously engaged in research on structurally sound and seis-
mically efficient construction and has published several research papers in national
and international journals and conferences/seminars/symposia.
Dr. P. C. Ashwin Kumar is Assistant Professor at the Department of Earthquake

Engineering, Indian Institute of Technology Roorkee. He has been engaged in
teaching and research in earthquake-resistant design of steel structures, development
of passive devices for seismic protection, vulnerability assessment, and retrofitting
of structures.
xi
Probabilistic Arias Intensity Maps
of Uttarakhand State (India)
Kunal Gupta and Neelima Satyam
Abstract Uttarakhand, an Indian state in the Western Himalayan region, is highly

seismically active and falls under zones IV (severe) and V (very severe) as per the
seismic code of India. The region is home to some highly significant hydropower and
infrastructure projects which are either operational or in the planning phase in the
region. Therefore, it is vital to determine the area to be impacted by the high seismicity
of this region. As Arias intensity is a very efficient measure to quantify the strength of
ground motion, none of the prior studies tried to map the Arias intensity distribution
throughout Uttarakhand. Therefore, in the present study, Arias intensity is mapped
for the Uttarakhand state by the Cornell–McGuire approach. A thorough earthquake
inventory was created, considering all earthquake occurrences within a 300 km radius
of the research area, incorporating events of magnitude (M w ) > 4 from 1900 to 2020.
The inventory was declustered and homogenized into a commonly used moment
magnitude scale. Ten distinct seismogenic area source zones were established in the
study area. Seismic recurrence parameters for all sources were computed using the
earthquake inventory and tectonic framework. The hazard evaluation at the bedrock
level was conducted using a logic tree framework that included regional and global
attenuation models. The findings were presented in terms of Arias intensity hazard
maps for 475 and 2475 years return period. It was observed that Arias intensity
values vary from 0.15 to 2.13 m/s for 475-year return period and 0.21 to 6.23 m/s
for the 2475-year return period. Approximately 80% of the study area was found
to be vulnerable to co-seismic hazards. The ground motion intensity estimated in
this study will assist in the effective planning of major infrastructure projects and
earthquake-induced landslide hazard assessment.
Keywords Arias intensity · PGA · Uttarakhand · PSHA · Logic tree · Attenuation

model
K. Gupta (B) · N. Satyam

Department of Civil Engineering, IIT Indore, Indore, Madhya Pradesh, India
e-mail: phd2101104003@iiti.ac.in
N. Satyam
e-mail: neelima.satyam@iiti.ac.in
© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 1
M. Shrikhande et al. (eds.), Proceedings of 17th Symposium on Earthquake
Engineering (Vol. 4), Lecture Notes in Civil Engineering 332,
https://doi.org/10.1007/978-981-99-1459-3_1
2 K. Gupta and N. Satyam
1 Introduction
The Indian subcontinent’s northern state, Uttarakhand (Western Himalayas), has a

high seismic activity level. It is a section of the Alpine-Himalayan seismic belt,
which was created when the Indian and Eurasian Plates collided [1]. The overlap-
ping motion of these plates caused strain energy to build up. In the past, the region
has experienced earthquakes of various sizes brought on by releasing this energy.
Significantly large earthquake occurrences with magnitudes greater than 6 include
the Kangra earthquake of 1905, the Uttarkashi earthquake of 1991, and the Chamoli
earthquake of 1999. The region has also experienced several low- to high-magnitude
earthquakes over the past century, demonstrating the strong seismic vulnerability of
this region. Due to its significant seismic activity, the state of Uttarakhand is clas-
sified by the Indian seismic code as being in zones IV (severe) and V (extremely
severe) [2].
To identify the locations with a likelihood of having high ground motion param-
eter values, seismic hazard assessment is regarded as a crucial instrument. Either
the deterministic seismic hazard assessment approach (DSHA) or the probabilistic
seismic hazard assessment approach (PSHA) can be used to determine the seismic
risk in each location. The same datasets, which include earthquake origins, occur-
rence frequencies, and connections between ground motion attenuation, are used in
both methods. A semi-empirical method of seismic zonation developed by Joshi et al.
[3] is based on the deterministic modelling of finite ruptures along identified faults in
a region. Seismic hazard maps for the Indian regions of the Doon Valley [4], Assam
Valley [3], and the Uttarakhand Himalaya [5] have been created using this seismic
zonation technique based on semi-empirical modelling of the rupture plane. Khattri
et al. [6] prepared PSHA maps for the entire India and adjoining regions. Shankar
[7] and Nayak et al. [8] produced the seismic hazard maps of the Uttarakhand state
using the PSHA approach.
The PSHA for Uttarakhand and the neighbouring Himalayan region has previously
been conducted by a small number of researchers. The sole goal of these studies
was to map the seismic risk to help design earthquake-resistant structures [9]. The
distribution of the Arias intensity (Ia ) [10] over the complete state of Uttarakhand
was not attempted in any study, but the spectral acceleration (SA) and peak ground
acceleration (PGA) were. A matrix called the Ia is used to quantify the force of a
ground motion [11]. To determine how much shaking is occurring, the Ia matrix
detects the acceleration of transient seismic waves [10]. Ia provides greater details
on overall shaking energy in contrast to other scalar measurements such as SA and
PGA. The significance of Ia has been emphasized by a variety of ground vibration
simulation techniques, including spectral element methods [12], empirical function
methods [13], and source model convolution Green’s functions-based methods [14].
The Ia has a better link with deformations caused by the ground shaking [15] and is
a more trustworthy measure for assessing the risk of landslides [16–18].
In terms of Ia , this study performs the PSHA of Uttarakhand state in detail. Seis-
motectonic sources within and around the research area were examined for seismic
Probabilistic Arias Intensity Maps of Uttarakhand State (India) 3
hazard estimates. The spatial distribution of hazards due to earthquake shaking in

the study region was examined using earthquake catalogue data from 1900 to 2020.
The recently established global and regional ground motion prediction models were
considered to evaluate the ground motion characteristics. Hazard maps and curves
were used to depict the results for return periods of 475 and 2475 years. The obtained
hazard maps are essential for determining how vulnerable a place is to earthquakes.
The study will offer a fresh perspective on the assessment of earthquake-induced
hazards in the study area.
2 The Geological and Seismo-tectonic Framework

of the Study Area
The research region is in the Southern Himalayan range between 77–81° E longitude
and 28–32° N latitude (Fig. 1). With China to the north and Nepal to the east, this
region has an area of 53,483 km2 . It is classified as zones IV and V under the Indian
seismic code [2]. A total of 86% of the state’s land area is covered by the Himalayas
[19]. As the Himalayan Mountain ranges are one of the world’s youngest mountain
ranges, the region shows high geological instability and is particularly susceptible to
earthquakes.
Numerous intricate tectonic systems were formed due to the ongoing collision
of the Indian and Eurasian tectonic plates. The Indian tectonic plate is currently
advancing in the north direction at 50 mm per year rate. Himalayan seismic activity
is confined to the region along the plate boundary and is associated with the under-
thrusting and collision of the Indian plate beneath the Eurasian plate. Along with the
Fig. 1 Geographical location of the study area: a India b Uttarakhand state

Alpine-Himalayan seismic belt, this underthrusting has resulted in continual defor-

mation, faulting, and folding [20]. The Main Boundary Thrust (MBT), Main Frontal
Thrust (MFT), Main Central Thrust (MCT), and Indus Suture Zone (ISZ) are the
significant tectonic features of the area. The ISZ is located at the region’s northern
border. The MBT is presently an active thrust system, whereas the MCT is passive and
the oldest. The MFT is thought to be the newest thrust system. The Tethys Himalaya,
the Lesser Himalaya, the Outer Himalaya, and the Greater Himalaya, which pass
through the state of Uttarakhand, are the four primary thrust systems that divide the
Himalayas. The Greater Himalayas are situated between the ISZ and the MCT to the
north of the MCT. It contains an intrusive igneous rock with a coarse texture. South of
MBT, the outer Himalaya comprises Miocene molasse sediments that have been bent
and faulted. Sedimentary strata that have undergone modest metamorphism and have
undergone severe folding make up the 100 km wide Tethys Himalaya [21–23]. Due to
the high likelihood of future earthquakes, the major area of the state of Uttarakhand
falls inside the Central Seismic Gap, which crosses both Uttarakhand and Western
Nepal. Additionally, there are a number of additional Himalayan thrusts, faults, and
lineaments which affect the seismic activity of the research area and are shown in
Fig. 2.
3 Seismic Source Zone Delineation
Delineating seismic source zones is the first step in assessing the seismic risk for a
given area. Areal sources were taken into account for seismicity modelling because
there is not enough information on specific faults, and there may be undiscovered
faults in the vicinity. This led to the investigation of various area source zone arrange-
ments. Several factors, such as seismic activity, fault plane distribution, tectonic
regime, and source depth, were considered throughout the delineation process. In
total, 10 source zones were drawn around the study area (Fig. 3). The Himalayan
structural belt exhibits a number of the region’s tectonic features, as shown in Fig. 2.
The studied region is largely within earthquake zones 1, 2, 3, 4, and 5. The MCT,
Alakananda Fault (AF), North Almora Thrust (NAT), and Martoli Thrust (MT) are
important tectonic features in these zones. High tectonic activity was seen in the
region between these zones, where there was a substantial concentration of earth-
quakes. Normal faulting is present in zones 6 and 7, which are a component of
the western part of the Himalayas. The principal tectonic characteristics of this
area include the Indus Suture Zone (ISZ), the Kaurik Fault System (KFS), and the
Karakoram Fault (KF). In the Western and Central Himalayas, there is a strong corre-
lation between the active faults and regional tectonic features. Multiple moderate-
to high-magnitude earthquakes have hit this area over the past 200 years. The lesser
Himalayas are covered by zones 8, 9, and 10, and this area has historically seen the
modest earthquake activity. These zones did not experience earthquakes of a high
magnitude.
Fig. 2 Major seismo-tectonic features and earthquake occurrences within and around the state
of Uttarakhand; Mahendragarh Dehradun Fault (MDF), Moradabad Fault (MF); Great Boundary
Thrust (GBT), Ramgarh Thrust (RT), South Almora Thrust (SAT), Alakananda Fault (AF), North
Almora Thrust (NAT), Martoli Thrust (MT), Indus Suture Zone (ISZ), the Kaurik Fault System
(KFS), and the Karakoram Fault (KF)
4 Development of Earthquake Catalogue
The first stage in hazard assessment by the probabilistic method is the identifica-
tion of the seismic sources, which requires collecting historical seismicity data and
compiling an updated homogenous earthquake occurrence database. The data for
the earthquake catalogue used in this study was gathered from a number of sources,
including the India Meteorological Department (IMD), the United States Geolog-
ical Survey (USGS), the International Seismological Centre (ISC), and previously
published studies. This analysis considers the earthquakes that happened in Uttarak-
hand and the surrounding areas between 1900 and 2020. The spatial distribution
of the instrument-collected earthquake data for the given time period is depicted in
Fig. 2. Because information was gathered from a variety of agencies and sources of
literature, resulting in a diverse catalogue, the earthquake occurrences were reported
using multiple magnitude scales. Different magnitudes were transformed into the
Fig. 3 Areal seismic source zones demarcated in and around the Uttarakhand state for the present
study
moment magnitude (Mw ) scale to provide a homogeneous record. The study used
the moment magnitude scale since the surface-wave magnitude (Ms ), local magni-
tude (Ml ), and body-wave magnitude (Mb ) were only accurate across a limited range
of distance and frequency. For large earthquakes, Mw provides the most precise earth-
quake size estimation. It is necessary to establish actual relationships between them
specific to each place to convert other scales to Mw scale. Global linkages created
by [24] allowed Ms to be transformed into Mw . The region-level relationships for
the Indian Himalayan region by [25] were used for the conversion of Mb to Mw .
Ml to Mw conversions were done using [26] correlations. Equation (1) presents the
empirical relationship provided by [25] used to convert the seismic records in terms
of intensity measure to Mw .
Mw = 0.762M M I + 0.865 (1)
where MMI is Modified Mercalli Intensity.

The aforementioned relationships were used to create a homogenous earthquake
catalogue and inferred that certain earthquake data in the final database overlapped
and that the same event was likely recorded more than once. Based on the reliability
and accuracy of the source, all duplicate data entries were eliminated in the first elim-
ination stage. The primary event induced dynamic and static stress changes, resulting
in several dependent events, including foreshocks, mainshocks, and aftershocks. To
create an independent dataset, these dependent events must be removed, and the
process is known as declustering. In the present work, declustering was performed
using the [27] technique. The aftershock progression’s duration and spatial extent
were computed in terms of the mainshock. Following the declustering process, it was
discovered that the catalogue was lacking for several time periods and magnitude
ranges. Since the catalogue’s completeness is crucial for determining seismicity char-
acteristics, the catalogue’s completeness was evaluated using Stepp’s method [28].
As magnitudes smaller than 4.0 could not pose significant risks [8], therefore it was
taken as the lowest value of magnitude (Mmin ).
5 Estimation of Recurrence Parameters
PSHA requires the use of the recurrence parameters (a, b). The Gutenberg-Richter
recurrence law was used to determine these parameters. According to the law,
earthquakes occur randomly and independently in terms of both location and time
in any given region according to the Poisson distribution. The representation of
Gutenberg-Richter’s law is presented in Eq. (2).
log λm = a − bm (2)
where a and b are recurrence parameters describing the area’s seismicity and λm
represents the average rate exceedance of magnitude m. A given region’s seismicity
is related to the parameters a and b, and the parameter b describes the relative size
distribution of earthquake events. A lower b value indicates that a higher magni-
tude predominates over a low-magnitude earthquake. A larger b value, in contrast,
indicates that a smaller magnitude predominates over a higher magnitude. In Fig. 4,
Guttenberg-Richter’s relationship for the entire research region is depicted. Using the
ZMAP software, the recurrence parameter values and the magnitude of completeness
(Mc ) were calculated [29].
5.1 Maximum Magnitude (m max )
For a known seismic setting, the earthquake with the maximum magnitude (m max ) is
the one responsible for the most extreme ground shaking. According to this research,
m max refers to the upper limit of Mw for a fault under consideration that is specified
for a seismic zone. The presently used m max estimation techniques can be either
deterministic or probabilistic, and these techniques frequently use seismicity data
to make their estimates [30]. Some methods also incorporate fault features [31] and
geodetic measurements [32]. Approaches based on earthquake occurrence data are
probabilistic; in contrast, deterministic methods are based on geodetic measurements
Fig. 4 Guttenberg-Richter’s
curve for the Uttarakhand
state
and fault characteristics. The non-parametric Gaussian estimator proposed by [30]

was used in this investigation to calculate m max .
6 Ground Motion Prediction Models for Arias Intensity

Assessment
Many empirical connections have been found in recent years to link the ground
movements caused by earthquakes to the Ia . Ia is a numerical indicator of the intensity
of an earthquake’s shaking. It is described as the sum of the energies per unit weight
stored by a collection of simple undamped oscillators with resonance frequencies
ranging from zero to infinity at the conclusion of an earthquake [10]. The equation
of Ia is given in Eq. (3).
π Td
Ia = ∫ a(t)2 dt (3)
2g 0
where T d is the ground motion duration, g is the gravitational acceleration, and a(t)
is the recorded temporal acceleration history.
The first correlation between the intensity measure and the distribution of seismi-
cally generated landslides was made by [16], who also gave the first global attenuation
relationship for Ia , which is represented by Eq. (4).
log Ia = −4.1 + M − 2 log R ± 0.44 (4)
where R is the distance from the source measured from the slip surface, M is the
earthquake magnitude, and 0.44 is the standard deviation derived from the regression
analysis. Regression analysis was used to generate Eq. (4) from a dataset of 30
recordings of eight Californian events, one from Hawaii and one from Japan. Jibson
[33] amended the same dataset using information from the Tabas (Iran) earthquake
that occurred in 1987 and provided Eq. (5) for calculating Ia .
log Ia = −4.9 + 0.98M − 1.35 log R (5)
Travasarou et al. [34] discussed about the usefulness of employing Ia to evaluate

the behaviour of structures subjected to seismic loading which is influenced by the
high-frequency part of ground motion. He gave Eq. (6) for the computation of Ia .
√
M
log Ia = 2.8 − 1.981(M − 6) + 20.72 ln − 1.703 ln R 2 + 77.09
6
(6)
+(0.454 + 0.101((M − 6))SC
+(0.479 + 0.334((M − 6))S D − 0.166FN + 0.512FR
where S D and SC are dummy variables dependent on the soil types (both 0 for site
category B, 1 and 0 for site category C, and 0 and 1 for site category D) and FR and
FN are dummy variables based on the fault types (1 and 0 for normal faults, both
0 for strike-slip faults, and 0 and 1 for reverse or reverse-oblique faults). Although
[34] model is reliable and has global applicability, it has some shortcomings, most
notably in the shallow depth site response modelling. A new empirical model for Ia
computation, which has global applicability was suggested by [35] using the Pacific
Earthquake Engineering Research Center, Next Generation of Attenuation (PEER
NGA) database. The Foulser-Piggott Attenuation (FPA) model is given by Eq. (7):
/
log Ia = c1 − c2 (8.5 − M) + (c3 − c4 M) ln
2
R2 + c52 + c6 FRV (7)
where FRV is an indicator variable having a value of 1 for reverse-oblique and reverse
ruptures and 0 otherwise. Parameters for this model found through the regression
analysis are presented in Table 1.
For shallow crustal earthquakes with moment magnitudes of 5 to 8, distances of
under 100 km, and VS30 values between 200 and 1000 m/s, the FPA model can
be used. The FPA and Travasarou models were thought to be appropriate for the
Table 1 FPA model

Coefficient Value Standard deviation
coefficients for Ia (modified
from [35]) c1 4.9862 0.4716
c2 − 0.1939 0.0453
c3 − 4.0332 0.3351
c4 0.2887 0.0524
c5 6.3049 0.9341
c6 0.3507 0.1615
Uttarakhand state to evaluate the Ia because shallow crustal earthquakes are frequent
in this region.
7 Logic Tree Structure
The choice of a seismic hazard model is challenging because of the uncertainty

associated with developing and implementing an attenuation model used to evaluate
seismic hazards. These uncertainties fall into two categories: epistemic and aleatory.
Epistemic uncertainty refers to the lack of information in simulating and under-
standing complicated earthquake phenomena. Combining various ground motion
models helps defeat it. The inherent unpredictability of earthquake occurrences is
represented by aleatory uncertainty. It cannot be reduced, but more information can
help it generate accurate estimates. A reliable method to reduce epistemic uncertainty
is the logic tree, which may be used with various ground motion prediction models
[36]. A logic tree comprises a series of branches and nodes, where each branch
and node depict a particular model. Based on engineering judgement, each of these
branches might be assigned a subjective weightage. It was ensured that these equa-
tions are collectively exhaustive and mutually exclusive when deciding how much
weight to give each of these branches, with the weights from an individual node added
together to a value equalling 1 [37]. The logic tree used in this research is depicted
in Fig. 5. An area source model was
considered for the study area, and by taking into
account the observed magnitude Mmax(obs) and m max obtained using [30] estimator,
the uncertainty in the calculation of m max was reduced. By merging two attenuation
equations and assigning various weightages to each, Ia was mapped. Both the FPA
and Travasarou models, which are two widely utilized models for calculating Ia , were
taken into consideration for the present research. The Travasarou model obtained a
weightage of 0.3, whereas the FPA model was given a weightage of 0.7 due to its
usage of the most recent database.
8 Seismic Hazard Assessment
The PSHA method, which [38] proposed, was used in this work to calculate seismic
hazards. By computing the probability of exceedance of a specific value, x, of a
ground movement parameter, X , and multiplying it by the likelihood that the earth-
quake of a specified magnitude would take place at a specific location, this approach
can determine the likelihood of an earthquake at one probable location. The following
equation was used to compute the exceedance probability:
m min rmax
P(X > x) = ∫ ∫ P(X > x|m, r ) f M (m) f R (r )dr dm (8)
m min 0
Fig. 5 Logic tree framework adopted in the study
where P(X > x|m, r ) is derived from the attenuation model and f R (r ) and f M (m)
are the probability density functions (PDFs) for distance and magnitude, respectively.
The process was repeated for each combination of locations and magnitudes. Equa-
tion (8) can be generalized for potential earthquake sources (N s), each of which
has a mean rate of threshold ground motion parameter exceedance, λi = e∝i −βi m 0 ,
then the total rate of exceedance is given by Eq. (9).
Σ
Ns
m min rmax
λ(X > x) = λi ∫ ∫ P(X > x|m, r ) f Mi (m) f Ri (r )dr dm (9)
m min 0
i=1
Since the parameters in the aforementioned equation are complex, integrals cannot
be evaluated analytically to estimate accurate PSHA. As a result, for analysis,
magnitude is divided into ranges and distance is split into a number of discrete
segments.
The probabilistic approach hazard assessment of the study area based on Poisson’s
model was carried out using the R-CRISIS [39] programme and a collection of
MATLAB algorithms. Each source zone received a set of highest and lowest magni-
tude values and recurrence parameters. The triangulation approach used by the
programme to discretize the area sources was carried out repeatedly until one of
the criteria was met. These criteria are the minimum triangle size (S) and minimum
source-to-site distance ratio (R). The sensitivity study included evaluating different S
and R combinations. Each source with M vertices was preliminarily subdivided into
M − 2 triangles, and the further subdivision was performed until the specified R or S-
value was attained. To carry out this subdivision, a recursive function was employed.
A calculation was made to determine the distance between the triangle’s centroid and
the computation site. The seismicity of the local source was placed in each triangle’s
centre. R-CRISIS uses the aforementioned approach to sample seismic source models
and forecast hazards while considering all likely earthquake sites inside the source.
The outcomes of the risk analysis were displayed using hazard maps. With different
weights assigned to each for Ia computations, the logic tree technique merged the
two distinct attenuation models. The ground motion parameters were calculated at
each grid point’s centre for 2 and 10% likelihood of exceedance in 50 years.
9 Results and Discussions
The PSHA technique was used to determine the state of Uttarakhand’s seismic hazard
zones, and the results are shown as hazard maps. The regional distribution of Ia
for 475 years return period is shown in Fig. 6. It was observed that the Ia values
range from 0.15 to 2.13 m/s. Significant portions of the area are quite susceptible
to seismic shaking. Values of Ia greater than 1.1 m/s were found in the Greater
Himalayan districts such as Rudraprayag, Tehri Garhwal, Chamoli, Uttarkashi,
Pithoragarh, and Bageshwar and are extremely susceptible to earthquake-induced
landslide hazards. Dehradun, Nainital, Pauri Garhwal, Champawat, and Almora are
all within a moderate to high seismic activity zone, with intensities ranging from
0.21 to 1.21 m/s. Low values of Ia were found in the lower altitude regions, such
as Udham Singh Nagar and Haridwar, making them less susceptible to co-seismic
dangers. Ia values range from 0.21 to 6.23 m/s for 2475 years return period, as shown
in Fig. 7. While the obtained mapping of seismic hazard is comparable to that seen
for 475 years of return periods, a sizable portion of the territory has Ia values greater
than 0.32 m/s for the 2475-year return period, thus making it extremely vulnerable
to earthquake-induced hazards.
10 Conclusion
The Cornell–McGuire probabilistic approach was used to assess the hazard due to
seismic activity in the state of Uttarakhand at the bedrock level. The hazard was
evaluated in terms of Ia for 2 and 10% probability of exceedance in 50 years using
various global and regional attenuation models. For a 10% likelihood of exceeding
in 50 years, the values of Ia ranged from 0.15 to 2.13 m/s, and for a 2% likelihood of
exceeding in 50 years, Ia ranged from 0.21 to 6.23 m/s. The acquired results suggested
that the Greater Himalayan region of Uttarakhand has considerable seismic activity.
Except for Udham Singh Nagar and Haridwar, all the districts of Uttarakhand state
are at risk from earthquake-induced hazards, according to an analysis of Ia maps. Of
all the districts, the Chamoli district has the highest vulnerability to seismic hazards.
This seismic hazard zonation of the Uttarakhand state in terms of Ia at bedrock will
serve as a foundation for the evaluation of the earthquake-induced landslide hazard in
the area. This estimated ground motion intensity will help with the careful planning
of significant infrastructure projects and the safe building of structures.
Fig. 6 Arias intensity map for 475 years return period
Fig. 7 Arias intensity map for 2475 years return period

References
1. Rajendran, K., Parameswaran, R.M., Rajendran, C.P.: Seismotectonic perspectives on the

Himalayan arc and contiguous areas: inferences from past and recent earthquakes. Earth-Sci
Rev. 173, 1–30 (2017). https://doi.org/10.1016/j.earscirev.2017.08.003
2. IS 1893 (Part 1): Criteria for earthquake resistant design of structures: general provisions and
buildings. Bureau of Indian Standards, New Delhi, pp. 1–39 (2016)
3. Joshi, A., Mohan, K., Patel, R.C.: A deterministic approach for preparation of seismic hazard
maps in North East India. Nat. Hazards 43, 129–146 (2007). https://doi.org/10.1007/s11069-
007-9112-7
4. Singh, P., Patel, R.C., Lal, N.: Plio-Plistocene in-sequence thrust propagation along the
main central thrust zone (Kumaon-Garhwal Himalaya, India): new thermochronological data.
Tectonophysics 574–575, 193–203 (2012). https://doi.org/10.1016/j.tecto.2012.08.015
5. Joshi, A., Kumar, A., Castanos, H., Lomnitz, C.: Seismic hazard of the Uttarakhand Himalaya,
India, from deterministic modeling of possible rupture planes in the area. Int. J. Geophys.
2013(825276), 1–12 (2013). https://doi.org/10.1155/2013/825276
6. Khattri, K.N., Rogers, A.M., Perkins, D.M., Algermissen, S.T.: A seismic hazard map of India
and adjacent areas. Tectonophysics 108(1–2), 111–134 (1984). https://doi.org/10.1016/0040-
1951(84)90156-2
7. Shanker, D., Shubhami: Seismic hazard scenario in Western Himalaya, India. In: SnT2019
T1.2-P80 (2019)
8. Nayak, M., Sitharam, T.G., Kolathayar, S.: A revisit to seismic hazard at Uttarakhand. Int. J.
Geotech. Earthq. Eng. 6(2), 56–73 (2015). https://doi.org/10.4018/IJGEE.2015070104
9. Gopiram, A., Satyam, D.N., Ramancharla, P.K.: Dynamic analysis of transmission towers under
strong ground motion. In: Proceedings, 3rd International Earthquake Symposium, Bangladesh,
Dhaka, pp. 1–7 (2010)
10. Arias, A.: Measure of earthquake intensity. In: Hansen, R.J. (ed) Seismic Design for Nuclear
Power Plants, Mass. Massachusetts Inst. of Tech. Press, Cambridge, pp. 438–483 (1970)
11. Satyam, N.: Seismic microzonation studies for Delhi region. Dissertation, Indian Institute of
Technology Delhi (2006)
12. Dhanya, J., Gade, M., Raghukanth, S.T.G.: Ground motion estimation during 25th April 2015
Nepal earthquake. Acta Geod. Geophys. 52, 69–93 (2017). https://doi.org/10.1007/s40328-
016-0170-8
13. Kaveh, A., Bakhshpoori, T., Hamze-Ziabari, S.M.: Derivation of new equations for prediction
of principal ground-motion parameters using M5, algorithm. J. Earthq. Eng. 20(6), 910–930
(2016). https://doi.org/10.1080/13632469.2015.1104758
14. Ma, J., Dong, L., Zhao, G., Li, X.: Qualitative method and case study for ground vibration
of tunnels induced by fault-slip in underground mine. Rock Mech. Rock Eng. 52, 1887–1901
(2019). https://doi.org/10.1007/s00603-018-1631-x
15. Harp, E.L., Wilson, R.C.: Shaking intensity thresholds for rock falls and slides: evidence from
1987 Whittier Narrows and Superstition Hills earthquake strong motion record. Bull. Seismol.
Soc. Am. 85(6), 1739–1757 (1995). https://doi.org/10.1785/BSSA0850061739
16. Wilson, R.C., Keefer, D.K.: Predicting areal limits of earthquake-induced landsliding. In:
Ziony, J.I. (ed) Evaluating Earthquake Hazards in the Los Angeles Region: An Earth-science
Perspective, vol. 1360, pp. 317–345. U.S. Geological Survey Professional Paper (1985)
17. Gupta, K., Satyam, N.: Estimation of Arias intensity and peak ground acceleration (PGA) using
probabilistic seismic hazard assessment of Uttarakhand state (India). Arab. J. Geosci. 15, 437
(2022). https://doi.org/10.1007/s12517-022-09733-9
18. Satyam, N.D., Towhata, I.: Site-specific ground response analysis and liquefaction assessment
of Vijayawada city (India). Nat. Hazards 81, 705–724 (2016). https://doi.org/10.1007/s11069-
016-2166-7
19. Gupta, K., Satyam, N.: Co-seismic landslide hazard assessment of Uttarakhand state (India)
based on the modified Newmark model. J. Asian Earth Sci. X 8, 100120 (2022). https://doi.
org/10.1016/j.jaesx.2022.100120
20. Valdiya, K.S.: Tectonics and evolution of the central sector of the Himalaya. Philos. Trans. R.
Soc. London Ser. A Math. Phys. Sci. 326(1589), 151–175 (1988). https://doi.org/10.1098/rsta.
1988.0083
21. Abraham, M.T., Satyam, N., Pradhan, B., Segoni, S., Alamri, A.: Developing a prototype
landslide early warning system for Darjeeling Himalayas using SIGMA model and real-time
field monitoring. Geosci. J. 26, 289–301 (2022). https://doi.org/10.1007/s12303-021-0026-2
22. Abraham, M.T., Satyam, N., Pradhan, B., Alamri, A.M.: Forecasting of landslides using rainfall
severity and soil wetness: a probabilistic approach for Darjeeling Himalayas. Water 12(3), 804
(2020). https://doi.org/10.3390/w12030804
23. Sharma, M., Satyam, N., Reddy, K.R.: Effect of freeze-thaw cycles on engineering properties
of biocemented sand under different treatment conditions. Eng. Geol. 284, 106022 (2021).
https://doi.org/10.1016/j.enggeo.2021.106022
24. Wason, H.R., Das, R., Sharma, M.L.: Magnitude conversion problem using general orthogonal
regression. Geophys. J. Int. 190(2), 1091–1096 (2012). https://doi.org/10.1111/j.1365-246X.
2012.05520.x
25. Das, R., Wason, H.R., Sharma, M.L.: Temporal and spatial variations in the magnitude of
completeness for homogenized moment magnitude catalogue for northeast India. J. Earth Syst.
Sci. 121, 19–28 (2012). https://doi.org/10.1007/s12040-012-0144-3
26. Ristau, J.: Moment magnitude-local magnitude calibration for earthquakes in Western Canada.
Bull. Seismol. Soc. Am. 95(5), 1994–2000 (2005). https://doi.org/10.1785/0120050028
27. Gardner, J.K., Knopoff, L.: Is the sequence of earthquakes in Southern California, with
aftershocks removed, Poissonian? Bull. Seismol. Soc. Am. 64, 1363–1367 (1974)
28. Stepp, J. C.: “Analysis of completeness of the earthquake sample in the Puget Sound area,”
In contributions to seismic zoning, NOAA Tech. Report, ERL 267-ESL 30, Washington, D.C.
pp. 16–28 (1973)
29. Wiemer, S.: A software package to analyze seismicity: ZMAP. Seismol. Res. Lett. 72, 373–382
(2001). https://doi.org/10.1785/gssrl.72.3.373
30. Kijko, A.: Estimation of the maximum earthquake magnitude, mmax . Pure Appl. Geophys. 161,
1655–1681 (2004). https://doi.org/10.1007/s00024-004-2531-4
31. Wells, D.L., Coppersmith, K.J.: New empirical relationships among magnitude, rupture length,
rupture width, rupture area, and surface displacement. Bull. Seismol. Soc. Am. 84(4), 974–1002
(1994). https://doi.org/10.1785/BSSA0840040974
32. Schiffman, C., Bali, B.S., Szeliga, W., Bilham, R.: Seismic slip deficit in the Kashmir Himalaya
from GPS observations. Geophys. Res. Lett. 40(21), 5642–5645 (2013). https://doi.org/10.
1002/2013GL057700
33. Jibson, R.W.: Summary of research on the effects of topographic amplification of earthquake
shaking on slope stability. In: US Geological Survey Open-file Report, pp. 87–268 (1987).
https://doi.org/10.3133/ofr87268
34. Travasarou, T., Bray, J.D., Abrahamson, N.A.: Empirical attenuation relationship for Arias
Intensity. Earthq. Eng. Struct. Dyn. 32(7), 1133–1155 (2003). https://doi.org/10.1002/eqe.270
35. Foulser-Piggott, R., Stafford, P.J.: A predictive model for Arias intensity at multiple sites and
consideration of spatial correlations. Earthq. Eng. Struct. Dyn. 41(3), 431–451 (2012). https://
doi.org/10.1002/eqe.1137
36. Stepp, J.C., Wong, I., Whitney, J., Quittmeyer, R., Abrahamson, N., Toro, G., et al.: Probabilistic
seismic hazard analyses for ground motions and fault displacement at Yucca Mountain, Nevada.
Earthq. Spectra 17(1), 113–151 (2001). https://doi.org/10.1193/1.1586169
37. Bommer, J.J., et al.: On the use of logic trees for ground-motion prediction equations in seismic-
hazard analysis. Bull. Seismol. Soc. Am. 95(2), 377–389 (2005). https://doi.org/10.1785/012
0040073
38. Cornell, C.A.: Engineering seismic risk analysis. Bull. Seismol. Soc. Am. 58(5), 1583–1606
(1968)
39. Ordaz, M., Salgado-Gálvez, M.A.: R-CRISIS v20 validation and verification document. In:
ERN Technical Report; Mexico City, Mexico (2020)
Statistical Analysis of Seismicity
Parameters and Completeness Period
of Earthquake Catalog for Sree
Padmanabhaswamy Temple,
Thiruvananthapuram District, Kerala
M. P. Hari Padmanabhan, Siddhardha R, Sreevalsa Kolathayar,

and Ramakrishna Hegde
Abstract In the present study, seismicity data analysis of Sree Padmanabhaswamy

Temple, Thiruvananthapuram District, Kerala, India, has been performed. The area
is having a radius of 500 km with Sree Padmanabhaswamy Temple (8.48°N Latitude
and 76.94°E Longitude) as the center. Earthquake catalog from 1822 to 2021 has been
compiled and homogenized into equivalent moment magnitude (Mw) using regional
empirical relationships. Data on earthquakes were declustered using Urhammer
method to exclude foreshocks and aftershocks within a time and space window, and
then statistical analysis was done to ensure data completeness. Using the Guntenberg-
Ritchers recurrence relationship, the seismic parameters obtained are 1.234 and
0.532, respectively. The M max value calculated using Kijko’s MATLAB algorithm
found to be 6.08 ± 0.26.
Keywords Gutenberg · Richter · Statistical analysis · Foreshocks · Seismicity
M. P. H. Padmanabhan
Department of Earthquake Engineering, Srinivas University, Mangaluru, Karnataka, India
R. Siddhardha (B)
Department of Civil Engineering, NIT Warangal, Warangal, Telangana, India
e-mail: rs721001@student.nitw.ac.in
S. Kolathayar
Department of Civil Engineering, NIT Surathkal, Mangaluru, Karnataka, India
e-mail: sreevalsa@nitk.edu.in
R. Hegde
Department of Civil Engineering, Srinivas University, Mangaluru, Karnataka, India
e-mail: rkhegdecivil.cet@srinivasuniversity.edu.in
https://doi.org/10.1007/978-981-99-1459-3_2
18 M. P. H. Padmanabhan et al.
1 Introduction
Sree Padmanabhaswamy Temple is located in the center of the circle-shaped research

region, which has a radius of 500 km, as shown in Fig. 1. The temple’s locational
coordinates are 8.48°N Latitude and 76.94°E Longitude. Making a single working
catalog for a region under consideration is a crucial undertaking since earthquake
catalogs provide the first crucial input for defining seismic source zones and charac-
terizing them. In various parts of the research region, the seismological and geolog-
ical data identify several lineaments and active faults. From the Bhukosh website
of Geological Survey of India (www.bhukosh.gsi.gov.in), the location and direction
of the lineaments and faults that are the linear seismic sources were determined.
To create a seismotectonic map, these lineaments and faults were digitally recorded
using QGIS software.
The instrumental data have also been sourced from internationally renowned
earthquake databases on the internet, including the International Seismological
Center (ISC) and the USGS 2008 online bulletin. The catalogue has been homog-
enized using the regression equations developed by [1] into an equivalent moment
magnitude (Mw). Thiruvananthapuram’s current earthquake database has 66 earth-
quakes with M w > 3.0 that occurred between 1822 and 2021. There have been six
documented earthquake events with a magnitude greater than 5.0, with a maximum
of M w 6.0.
Fig. 1 Geographical extent of the study region

Statistical Analysis of Seismicity Parameters and Completeness Period … 19
2 General Analysis of Earthquake Catalog
The number of earthquakes every decade was separated into six magnitude ranges,
as indicated in Table 1: 3.0 ≤ M w ≤ 3.49, 3.5 ≤ M w ≤ 3.99, 4.0 ≤ M w ≤ 4.49, 4.5 ≤
M w ≤ 4.99, 5.0 ≤ M w ≤ 5.49, and M w ≥ 5.49. The histogram representation of the
earthquake data is shown in Fig. 2. The histogram indicates that a significant number
of earthquakes were reported between 1982 and 1991. The earthquake reporting was
subpar and inaccurate before to 1982. Unreliable findings are obtained when the
seismicity parameter is calculated using incomplete data [2]. To solve this issue, the
data’s completeness was verified using the CUVI approach.
3 Declustering Process
Foreshocks and aftershocks are removed from the catalog by declustering. The
earthquake inventory for Thiruvananthapuram, as previously indicated, has 66 earth-
quake events with M w > 3 from 1822 to 2021 A.D. The primary earthquake shocks
are discrete, Poisson-distributed events [3]. The dynamic window approach by
Uhrhammer in 1986 has been employed in this study [4]. It was found that there
were no dependent events in the catalog. All events are main shocks only.
Figure 3 depicts the entire seismotectonic map of the study area, which includes the
faults, lineaments, and declustered earthquake events. In the figure, all the magnitudes
are M w . Table 1 lists a few of the significant faults found in the research area.
4 Completeness of Catalog
Making sure a recurrence connection is complete; it is essential to confirm that all

earthquakes for each magnitude range of interest have been documented for the time
period under consideration. Utilizing the Visual Cumulative (CUVI) approach by
[5], completeness assessments have been conducted [6]. To compute the period of
catalog completion, various magnitude classes are used in the current study.
4.1 Visual Cumulative Method
Figure 4 displays plots of the total number of events over time from the start of
the catalog for four different classes of magnitudes for the whole study region as a
single source zone (a–e). The era of completeness is thought to start for a particular
magnitudes class at the earliest point at which the slope of the fitting curve may be
reasonably represented by a straight line.
Table 1 List of major faults

S.No Fault Length (km)
in research area
1 Arkavati fault 120.3
2 Bhavali fault 87
3 Cauveri fault 130
4 Kottagudi–Kokkal–Palani fault 58
5 Malayattur–Vadakkancheri fault 36
6 Ottappalam–Kuttampuzha fault 98
7 Pattikkad–Kollengol fault 40
8 Periyar fault 84
9 Sakleshpur–Bettadpur fault 83
10 Tekkadi–Kodaivannalur fault 48
11 Tenmalai fault 70
12 Tiruppur fault 85
13 Valparai–Anamudi fault 45
14 Amaradakki fault 51
15 Amirdi fault 96
16 Attur fault 160
17 Ayakkudi–Vinupaksha fault 29
18 Bhavani–Kanumudi fault 61
19 Cauveri fault 196
20 Chitradurga boundary fault 81
21 Crystalline-sedimentary contact fault 26
22 Javadi hills fault 87
23 Main fault N 45 Degrees E 66
24 Main fault N 45 Degrees E 120
25 Manamelkudi–Tonti fault 35
26 Mettur east fault 82
27 Mettur east fault 36
28 Mouth of Coleroon–point Calimere 126
5 Gutenberg–Richter Recurrence Relationship
The source model, expressed by the Gutenberg–Richter (G–R) activity parameters ‘a’
and ‘b’ for each of the seismic zones, is the fundamental input for the seismic hazard
analysis [7]. By calculating the slope of a straight line connecting the distribution of
the logarithm of the number of earthquakes to their magnitude,
log(λ M ) = a−bM
14
Number of Earthquakes in a
given magnitude range 3-3.49
12
10 given magnitude range 3.5-3.99
Number of Earthquakes
8
6
given magnitude range 4.5-4.99
4
2 Number of Earthquakes in a
0
given magnitude range 5.5-6
Years
Fig. 2 A Histogram of seismic events in the research area
Fig. 3 Map showing the research region’s seismicity

(a) Mw = 3 to3.49
2.5
Cumulative number of Earthquakes 2
1.5
0.5
0
2006 2007 2008 2009 2010 2011 2012 2013
T ime (years)
(b)
Mw = 3.5 to 3.99
30
Cumulative number of Earthquakes
25
20
15
10
0
1985 1990 1995 2000 2005 2010 2015 2020 2025
T ime (years)
(c) Mw = 4 to 4.49
20
18
16
14
12
10
8
6
4
2
0
1985 1990 1995 2000 2005 2010 2015 2020
T ime (years)
Fig. 4 (a–e) CUVI method for catalog completeness

(d) Mw = 4.5 to 4.99
10
9
Cumuative number of Earthquakes 8
7
6
5
4
3
2
1
0
1980 1985 1990 1995 2000 2005 2010 2015 2020 2025
T ime (years)
(e) Mw ≥ 5
10
9
8
7
6
5
4
3
2
1
0
1800 1850 1900 1950 2000 2050
T ime (years)
Fig. 4 (continued)
where ‘a’ and ‘b’ are the characteristic constants of the seismic zone; λM = the
mean annual rate of exceedance of magnitude M. The constants ‘a’ and ‘b’ can
be evaluated using the least square regression analysis. The ‘b’ value is sometimes
thought of as a measure of the brittle–ductile transition of the crust [8]. The regression
analysis estimates the ‘a’ and ‘b’ values from the cumulative annual rate of earthquake
occurrence and the mean of the magnitude range.
The analysis makes the assumption that the seismic activity in the area will follow
the earthquake law proposed by Gutenberg and Richter, which implies an exponen-
tial distribution of earthquake magnitudes. These have undergone comprehensive
evaluation by regression analyses using the existing catalog and is furnished in Fig. 5.
Gutenberg Richter plot

1
0.5
0
log (λM)
0 1 2 3 4 5 6
-0.5
-1 log λM = -0.5321Mw + 1.2349
-1.5
-2
Moment magnitude (Mw)
Fig. 5 G–R relationship for single source zone using CUVI method
6 Maximum Magnitude
The maximum magnitude (M max ) is a crucial parameter for seismologists, insurance

companies, and disaster management organizations. The term ‘M max ’ refers to the
maximum earthquake magnitude that can occur in the area under consideration. The
selection of maximum magnitude (M max ) in Peninsular India is seriously questionable
due to the short time period covered by the earthquake catalog in comparison to
the frequency of high-magnitude earthquakes [9]. In order to estimate the value of
M max , the Kijko–Sellevoll–Bayes (K-S-B) method proposed by Kijko [10] takes into
account both the complete and the partial portions of the earthquake database. The
M max value was calculated using Kijko’s MATLAB algorithm (M max ). The obtained
M max value is 6.08 ± 0.26.
7 Conclusions
The current study examines the occurrence rate for a number of magnitude criteria in
order to verify the catalog’s completeness. Completeness threshold has been deter-
mined using CUVI method. The regional recurrence relations are obtained based on
nearly 199 years (1822–2021 A.D.) of past data. A crucial seismicity parameter is the
magnitude–frequency relationship’s b value. The estimated values of a and b, which
are crucial input parameters in the Gutenberg–Richter recurrence relationship, are
1.234 and 0.532, respectively, for the research region.
References
1. Kolathayar, S., Sitharam, T.G., Vipin, K.S.: Spatial variation of seismicity parameters across
India and adjoining areas. Nat. Hazards (Springer Publications) 60(3), 1365–1379 (2012)
2. Schorlemmer, D., Zechar, J.D., Werner, M.J., Field, E.H., Jackson, D.D., Jordan, T.H., &
RELM Working Group: First results of the regional earthquake likelihood models experiment.
In Seismogenesis and Earthquake Forecasting: The Frank Evison Volume II (pp. 5–22) (2010).
Springer, Basel
3. Gardner, J.K., & Knopoff, L.: Is the sequence of earthquakes in Southern California, with
aftershocks removed, Poissonian?. Bull. Seismol. Soc. Am. 64(5), 1363–1367 (1974)
4. Uhrhammer, R.: Characteristics of Northern and Central California seismicity. Earthq. Notes
57(1), 21 (1986)
5. Tinti, S., Mulargia, F.: Effects of magnitude uncertainties on estimating the parameters in
the Gutenberg-Richter frequency-magnitude law. Bull. Seismol. Soc. Am. 75(6), 1681–1697
(1985)
6. Nath, S.K., Mandal, S., Adhikari, M.D., Maiti, S.K.: A unified earthquake catalogue for South
Asia covering the period 1900–2014. Nat. Hazards 85(3), 1787–1810 (2017)
7. Gutenberg, B., Richter, C.F.: Frequency of earthquakes in California. Bull. Seismol. Soc. Am.
34(4), 185–188 (1944)
8. Amitrano, D.: Brittle-ductile transition and associated seismicity: experimental and numerical
studies and relationship with the b-value. J. Geophys. Res. Solid Earth 108(B1), 1–8 (2003)
9. Khan, M.M., Munaga, T., Kiran, D.N., & Kumar, G.K. Seismic hazard curves for Warangal
city in Peninsular India. Asian J. Civ. Eng. 21, 543–554 (2020)
10. Kijko, A.: Estimation of the maximum earthquake magnitude, Mmax . Pure Appl. Geophys.
161(8), 1655–1681 (2004)
Effect of Open Stories on Expected
Seismic Losses in Hilly Buildings
Yati Aggarwal and Sandip Kumar Saha
Abstract Due to the lack of adequate flat land in the hilly region, it is often opted to
construct buildings on the slope. Further, to accommodate the need for large covered
open spaces (e.g., parking and shops), open stories are provided in the hilly buildings
at levels based on the location of the approach road. In this study, an attempt is made
to determine the variation in the expected seismic losses due to the presence of
open story in reinforced concrete (RC) hilly buildings. For this, two basic types
of hilly buildings with different story ratios are considered. Bi-directional nonlinear
dynamic analyses are performed under 22 ground motions. The expected repair cost is
estimated considering the repair cost of four drift-sensitive building components. The
component-wise and floor-wise deaggregation of the expected repair costs due to the
considered components are studied. It is observed that the unreinforced masonry wall
contributes to the majority of the expected seismic repair cost in all the considered
hilly buildings. Moreover, the presence of open stories just above the uppermost
foundation level is observed to be the most critical in terms of estimated seismic
losses to structural components.
Keywords Hilly buildings · Open story · Seismic loss estimation
1 Introduction
An increase in population causes indiscriminate growth in the activities of building

infrastructure in the Indian Himalayan region. Due to scarcity of flat land in hilly
regions, buildings are usually constructed over slopes. It is evident from the literature
[1, 2] as well as from post-earthquake reconnaissance studies [3, 4] that reinforced
Y. Aggarwal (B) · S. K. Saha

School of Civil and Environmental Engineering, Indian Institute of Technology Mandi, Kamand,
Himachal Pradesh, India
e-mail: d17050@students.iitmandi.ac.in
S. K. Saha
e-mail: sandip_saha@iitmandi.ac.in
https://doi.org/10.1007/978-981-99-1459-3_3
28 Y. Aggarwal and S. K. Saha
concrete (RC) hilly building configurations perform poorly in comparison to the

buildings with regular configuration. The poor performance of hilly buildings is due
to the presence of short columns at intermediate floors. Moreover, to cater to the
need of open spaces for vehicle parking or other utility purposes, open stories are
provided in hilly buildings. Contribution of unreinforced masonry walls is often
ignored during the analysis and design phases of the building. When open stories are
provided in hilly buildings to serve any specific purpose, consequential difference
in the performance of hilly buildings is observed as reported by Aggarwal and Saha
[5].
Until now, we are aware of the seismic performance of hilly buildings without and
with open stories [5]. Nevertheless, their seismic loss assessment is also important
as it helps to gauge the performance of these buildings in terms of monetary losses
and count of vulnerable occupants, therefore enabling the decision-makers to plan
for mitigation policies prior to an impending disaster.
Herein, an attempt has been made to quantify the seismic loss in terms of financial
losses due to repair or reconstruction of hilly building components without and with
open stories. Thus, the objectives of present study are (i) to estimate the component-
wise repair cost, taking into account the contribution of unreinforced masonry wall,
RC beam, RC column, RC beam-column joint, in hilly buildings without and with
open story, and (ii) to study the repercussions of having an open story on seismic
loss estimation in hilly buildings.
2 Numerical Modeling and Ground Motion Selection
In this section, nonlinear numerical modeling of RC hilly building configurations

is explained and then followed by the selection of scenario-based ground motion
records. These ground motion records are scaled to make them compatible with the
code-defined response spectra. Considered hilly buildings are then subjected to these
scaled ground motions and their seismic responses are noted.
2.1 Numerical Modeling of Hilly Buildings
For studying the effect of open story, two RC hilly building configurations, i.e.,
stepback (SB) and split-foundation (SF), are selected. The elevation and plan views
of the considered buildings are depicted in Fig. 1. Three story ratios, (a) 0.5, (b) 1,
and (c) 2, are selected, keeping the number of floors below the uppermost foundation
level (UFL) uniform as four. Column height below the plinth level in SB varies from
1.1 to 2.75 m to match the ground slope. In split-foundation building, column height
below the plinth level is 1.1 m. Table 1 enlists the dead and live loads assigned to the
buildings as prescribed by the Indian standards [6, 7]. The behavior of slab is modeled
as a rigid diaphragm. The grade of concrete and steel is considered as M30 and Fe500,
Effect of Open Stories on Expected Seismic Losses in Hilly Buildings 29
respectively. The second moment of area of RC beam and RC column is reduced

to 35 and 70% of the gross second moment of area as per the recommendations of
IS 1893 [8] to account for cracked section properties. The cross section details and
dynamic characteristics of the considered buildings are mentioned in Table 2.
The lumped plasticity model is used to simulate the material nonlinearity in RC
beams and RC columns. The force–deformation (backbone curve) behavior of RC
beam and column is defined based on the ASCE 41 [9] recommendations. To account
for the cyclic degradation, energy-based degrading model is used. The parameters
used in this model are used from the study performed by Surana et al. [2]. Uni-
directional bending (M 3 ) plastic hinges are allocated to RC beams, whereas axial-
bi-directional bending (P-M 2 -M 3 ) plastic hinges are allocated to RC columns. The
length of flexural plastic hinge is calculated using the expression proposed by Paulay
and Priestley [10] and is assigned at a distance of half of the flexural plastic hinge
length from both ends of the RC beams as well as RC columns. Shear plastic hinges
2-story SB 4-story SB 8-story SB 2-story SF 4-story SF 8-story SF
Across slope
direction
8 stories
@ 3.3m
8 stories
@ 3.3m
Along slope
direction
UFL UFL
4 stories
@ 3.3m
4 stories
@ 3.3m
LFL LFL
1.1m 1.1m
(a) (b)
6m 3m 6m
Downhill side
Uphill side
8 bays @3.2m = 25.6m

(c)
Fig. 1 Hilly building configurations a elevation view of stepback buildings, b elevation view of
split-foundation buildings, and c building plan
Table 1 Details of dead and

Component Magnitude
live loads considered in this
study Slab finishing load 1 kN/m2
Masonry load 20 kN/m3
Reinforced concrete load 25 kN/m3
Roof treatment 1.5 kN/m2
Roof live 1.5 kN/m2
Live load 3 kN/m2
Height of parapet wall 1.5 m
Table 2 Cross section details and fundamental periods of hilly buildings

Building Column Beam Short Slab Fundamental Fundamental
type (mm) (mm) column thickness period in period in
(mm) (mm) across-slope along-slope
direction (s) direction (s)
2-story SB 350 × 350 300 × 450 × 150 0.29 0.2
300 450,
750 ×
750
4-story SB 350 × 350 400 × 450 × 0.50 0.36
300 450,
800 ×
800
8-story SB 300 × 300, 400 × 500 × 1.05 0.72
350 × 350 300 500,
800 ×
800
2-story SF 350 × 350 300 × 450 × 0.33 0.23
300 450
4-story SF 350 × 350 400 × 450 × 0.52 0.38
300 450
8-story SF 350 × 350, 400 × 500 × 1.00 0.71
450 × 450 300 500
(force-controlled) are assigned to short columns to simulate the shear failure in short
columns. The shear hinge length is considered as 1.5 times the depth (d) of the short
column and is assigned at a distance of d from the end of the short column. Geometric
nonlinearity in the form of P-delta effect is also considered in this study.
To simulate the effect of unreinforced masonry wall, equivalent diagonal struts are
modeled as per the recommendations of IS 1893 [8]. The strength of unreinforced
masonry wall is considered as 4.1 MPa (fair category as per ASCE 41[9]). The
thickness (t) of equivalent strut is same as the thickness of unreinforced masonry
wall. The width (wds ) of equivalent strut is estimated on the basis of relative stiffness
of wall and frame:
wds = 0.175αh−0.4 L ds (1)
where L ds is the length of equivalent diagonal strut, and αh is estimated as:

/
E m t sin 2θ
αh = h 4
(2)
4E f Ic h
where h is the height of unreinforced masonry wall; E m and E f are the elastic moduli
of masonry and RC frame, respectively; I c is second moment of area of columns; θ is
the angle between equivalent strut and beam. The compression-only axial behavior
SB-IA SB-IRB SB-IRT SB-IRBoth
SF-IA SF-IRB SF-IRT SF-IRBoth
Fig. 2 Hilly buildings with prospective locations of open story
of equivalent strut is considered. The thickness of exterior and interior unreinforced

masonry walls is considered as 230 mm and 150 mm, respectively. The presence
of masonry infill is considered during both analysis and design phases of the hilly
buildings.
Further, to assess the consequences of having an open story in hilly buildings,
masonry walls (infills) are removed from the floors which are close to the approach
road. Four prospective cases are selected based on the field survey and assessed in
this study. These are (a) infills considered throughout the building represented as
IA, (b) open story at lowermost foundation level (LFL) represented as IRB, (c) open
story at UFL represented as IRT, and (d) open story at both LFL and UFL represented
as IRBoth, as shown in Fig. 2.
2.2 Ground Motion Selection and Scaling
A set of 22 ground motion records is selected to conduct this study with the limi-
tation on moment magnitude (M w ) and site to source distance (Rjb ). The magnitude
of selected strong far-field ground motion records is in between 6.0 and 8.0, and
site to source distance is between 15 and 30 km to avoid near-field effects as per
the guidelines of FEMA P695 [11]. The ground motion records are taken from
PEER strong ground motion database [12], and their details are mentioned in the
authors’ previous work [5]. Both the horizontal components of the records are used
for performing bi-directional nonlinear time history analysis. Across-slope direc-
tion of building is subjected to ground motion component with higher magnitude of
peak ground acceleration (PGA), and along-slope direction is subjected to another
component of ground motion.
The selected ground motion records are scaled with respect to spectral acceleration
at fundamental period of the building corresponding to 5% damping (S a (T 1 , 5%)).
The spectral acceleration is scaled to match the target spectrum [13] for rocky soil
in seismic zone V specified by Indian seismic design code [8]. The fundamental
periods of the considered buildings are different in their two translational directions;
therefore, scaling of two ground motion components is done separately to match the
target spectrum.
2.3 Numerical Study
Nonlinear time history analyses are performed. These analyses employ Newmark-β
integration method, to assess the seismic performance of 24 hilly buildings. These
24 building models include two different building configurations, three different
number of stories, and four different prospective locations of open story. Ground
motions are assigned in two translational directions of the buildings. The 5% propor-
tional damping is assigned to the first period of vibration and the period at which
at least 90% of the cumulative modal mass is participating. For performing numer-
ical modeling of buildings and dynamic analyses, FEM-based structural analysis
software, SAP2000v21 [14], is used. Seismic response in terms of peak interstory
drift ratio (IDR) is noted in all the considered buildings. Median of 22 responses is
calculated to estimate the expected seismic losses due to one non-structural and three
structural building components.
3 Seismic Performance of Hilly Buildings
Seismic response of buildings in terms of peak IDR is an important engineering

demand parameter because seismic losses are estimated using IDR, especially in
structural components. IDR in across-slope direction of the buildings is reported in
this study because the stiffness of considered buildings is less in lateral direction
(across-slope direction) as compared to their longitudinal direction. Thus, higher
interstory drift is obtained in lateral direction of the building. In each considered
building, the median of 22 peak IDR is calculated at every floor level. Figure 3
shows the variation in peak IDR corresponding to relative story height in considered
the hilly buildings. The shaded portion in Fig. 3 represents the building portion below
the uppermost foundation level. It is observed from Fig. 3 that in two-story buildings
when all masonry infills are present, peak IDR remains within the permissible limit
of 0.4% as recommended by [8]. However, when two-story hilly buildings have open
story, peak IDR increases drastically and exceeds the permissible limit. In four-story
and eight-story hilly buildings without and with open story, peak IDR exceeds the
permissible limit of 0.4%. Presence of an open story decreases stiffness of story,
thereby increase in peak IDR is observed.
1.00
IA
IRB
0.75
IRT
IRBoth
0.50
Relative Story Height
0.25
2-story SB 4-story SB 8-story SB
0.00
1.00
0.75
0.50
0.25
2-story SF 4-story SF 8-story SF
0.00
0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5 2.00.0 1.0 2.0 3.0 4.0
IDR (%) - Across Slope Direction
Fig. 3 Variation in peak interstory ratio in lateral direction of hilly building for different prospective
locations of open story
Moreover, it is observed that the increase in peak IDR is higher when open story
is at UFL in comparison to the LFL of the building. About 3–4 times increase in
peak interstory ratio is observed when open story is at UFL. This is because the
quantity of masonry infills is lesser at LFL; thus, removal of these masonry infills
imparts lesser flexibility to the building in comparison to removal of masonry infills
from UFL. Interestingly, when open story is present at both UFL and LFL, two
spikes are observed in split-foundation buildings, whereas only one spike of increased
peak IDR is observed in stepback buildings. Again, the reason is same that the
quantity of masonry infills at LFL is less in SB building in comparison to SF building.
Thus, dynamic analysis of hilly buildings has revealed that the maximum peak IDR
(maximum among all floor levels) increases significantly due to the presence of open
story.
4 Seismic Loss Estimation
The expected seismic loss is estimated considering four building components, i.e.,
unreinforced masonry wall, RC beam, RC column, and RC beam–column joint. For
this, response (peak IDR) of hilly buildings, without and with open story, is utilized
to estimate the probability of exceeding a damage state (P(DSi |I D R)) using the
expressions mentioned in FEMA P58[15]. The fragility functions for the considered
building components are referred from relevant literatures as shown in Fig. 4. For
unreinforced masonry wall, fragility functions are taken from the experimental study
performed by Cardone and Perrone [16]. For RC beam and RC column, fragility
functions are referred from the library of FEMA P58[15]. Fragility functions for
Unreinforced masonry wall [14] Reinforced concrete beam [13]

1.0
MedianDS-1 = 0.0015
DispersionDS-1 = 0.50
MedianDS-2 = 0.0040 MedianDS-1 = 0.0137
DS-1
Probability of exceedance
DS-2 DispersionDS-2 = 0.50 DispersionDS-1 = 0.21

0.5
DS-3 MedianDS-3 = 0.0100 MedianDS-2 = 0.0264
DS-4 DispersionDS-3 = 0.40 DispersionDS-2 = 0.33
DispersionDS-4 = 0.35 DispersionDS-3 = 0.74
0.0
Reinforced concrete column [13] Reinforced concrete beam-column joint [15]
1.0
MedianDS-1 = 0.0068
DispersionDS-1 = 0.89
0.5 MedianDS-2 = 0.0275 MedianDS-3 = 0.0392
0.0
0.00 0.05 0.10 0.00 0.05 0.10
Peak interstory drift ratio
Fig. 4 Fragility functions for the considered building components
reinforced beam–column joint are adopted from the work of Brown and Lowes [17].
The probability of exceeding a damage state is calculated considering the median of
responses in which no simulated collapse instances are observed. For the instances,
in which simulated collapse is observed, probability of collapse (Pc ) is estimated
as the ratio of number of cases for which simulated collapse is observed and total
number of cases (i.e., 22 ground motion records). Total probability of exceeding a
damage state (Pt (DSi |I D R)) is calculated as [18]:
Pt (DSi |I D R ) = Pc + (1 − Pc ) × P(DSi |I D R ) (3)
After calculating the probability of exceeding a damage state for considered

building components, their repair cost under different damage states is also required
to be estimated for seismic loss estimation. Repair cost functions for considered
building components under different damage states are adopted from the work of
Aggarwal and Saha [19]. They proposed region-specific repair cost functions for
estimating the repair cost (RE ) (in Indian rupees) of building components under
different damage states. The general form of repair cost functions is as follows:
(4)
where m, n, p, k are the upper limits of the controlling parameters i, j, q, l and their
values depend on the type of building component; x, y, z, and w are the independent
variables that account for the cross-sectional dimensions of the building components.

The values of coefficients ai, j,q,l are taken, from appendices of Aggarwal and Saha
[19], corresponding to Himachal Pradesh.
Using estimated total probability, quantity of considered building components
present in hilly buildings and repair cost of each building component for each damage
state expected repair cost (L e ) of the buildings are estimated as per the following
expression [15]:
Σ
n1 Σ
n2 Σ
n3
Le = Nc × R E × Pt (DSi |I D R ) (5)
Fk=1 C j=1 DSi=1
where N c is the quantity of each component at a floor; n1 , n2 , n3 are the number of

floors, type of components per floor and considered damage states in a building and
also the upper limits of F k , C j , and DS i , respectively.
Component-wise (separately for each considered building component) repair cost
for considered hilly buildings is estimated using the expression discussed above. The
obtained component-wise repair cost is then normalized with the reconstruction cost
of its respective hilly building and denoted as component-wise repair cost ratio. The
reconstruction cost for all the considered hilly buildings is mentioned in Table 3.
Figure 5 shows the variation in component-wise repair cost ratio in all the considered
hilly buildings without and with open story. It is observed that the component-wise
repair cost ratio is lesser in SB than SF building. This is probably due to the fact that
in SF building, there is comparatively larger livable space available below the UFL
in comparison to SB building that adds to the greater number of components in SF
building. Also, the peak IDR of SB building below the UFL is lower in comparison to
SF building due to the presence of short columns at intermediate story in SB building
that increase its story stiffness; therefore, increase in component-wise repair cost ratio
is observed in SF building.
It is also observed that component-wise repair cost ratio increases as the number
of stories increases. It is important to note here that repair cost ratio of unreinforced
masonry wall is lesser for open story as compared to without any open story. However,
the repair cost ratio of other structural components, especially RC beam and RC
column, increases significantly. This concludes that due to the presence of an open
story IDR increases and thus higher damages are expected which further increases
the repair cost ratio of the structural components.
Table 3 Reconstruction cost of the considered buildings (in 103 |)

Building SB SF
Configuration IA IRB IRT IRBoth IA IRB IRT IRBoth
2-story 17,936 17,370 16,140 15,575 19,776 18,801 17,980 17,005
4-story 27,163 26,616 25,427 24,880 29,034 28,091 27,298 26,354
8-story 44,330 43,783 42,594 42,047 46,817 45,873 45,081 44,137
Unreinforced masonry wall RC beam RC column RC beam-column joint
0.06 0.09 0.15

2-story SB 4-story SB 8-story SB
0.12
0.04 0.06
0.09
0.06
Component-wise repair cost ratio
0.02 0.03
0.03
0.00 0.00 0.00

IA IRB IRT IRBoth IA IRB IRT IRBoth IA IRB IRT IRBoth
0.06 0.09
2-story SF 4-story SF 0.12 8-story SF
0.04 0.06
0.08
0.02 0.03 0.04
0.00 0.00 0.00

Fig. 5 Deaggregation of component-wise repair cost ratio
Further, floor-wise repair cost by considering the repair cost of all building compo-
nents in a floor under all damage states is estimated. Then, floor-wise repair cost is
normalized with the reconstruction cost of that floor and represented as floor-wise
repair cost ratio as shown in Fig. 6. Figure 6 shows the variation in floor-wise repair
cost ratio in all the considered buildings. It is noticed that the location of experiencing
maximum floor-wise repair cost ratio is typically at uppermost foundation level irre-
spective of the presence or absence of open story. However, marginal increase in
floor-wise repair cost ratio at uppermost foundation level is observed in most cases
when open story is considered.
Figure 7 shows the floor-wise repair cost ratio in hilly buildings after ignoring the
contribution of unreinforced masonry walls. Significant increase in the floor-wise
repair cost ratio is observed in hilly building with an open story. It is known that
due to open story, peak IDR increases (Fig. 3). This increase in response causes
higher damages specifically to the location of open story and higher damages cause
higher repair costs. Figure 5 shows that maximum seismic loss is contributed by
unreinforced masonry walls. Therefore, at the location of open story, where no unre-
inforced masonry walls are present, marginal increase in floor-wise repair cost ratio
is observed when the contribution of unreinforced masonry walls is considered in
Fig. 6. Nevertheless, it is interesting to observe from Fig. 7 how repair cost of other
structural members increases due to the presence of open story.
The repair cost of all floor levels (including all four considered building compo-
nents) is aggregated to represent the overall repair cost of the building. This repair
cost is normalized with the reconstruction cost of the corresponding building and
1.0 2-story SB 4-story SB 8-story SB
0.5
Relative story height
0.0
1.0 2-story SF 4-story SF 8-story SF
0.5
0.0
0.00 0.02 0.04 0.06 0.08 0.10 0.00 0.05 0.10 0.15 0.0 0.1 0.2 0.3 0.4 0.5 0.6
Floor-wise repair cost ratio
Fig. 6 Floor-wise variation in repair cost ratio
1.0 4-story SB
2-story SB 8-story SB
0.5
Relative story height
0.0
1.0 4-story SF
2-story SF 8-story SF
0.5
0.0
0.00 0.02 0.04 0.06 0.08 0.10 0.00 0.05 0.10 0.15 0.20 0.0 0.1 0.2 0.3 0.4 0.5 0.6
Floor-wise repair cost ratio
Fig. 7 Floor-wise variation in repair cost ratio without considering the repair cost of unreinforced
masonry infill
2-story SB 0.09 4-story SB 0.4 8-story SB

0.06
0.3
0.04 0.06
0.2 0.367 0.367
0.080 0.080
0.050 0.046
0.02 0.041 0.040 0.03 0.053 0.054
0.1
Repair cost ratio
0.131 0.128
0.00 0.00 0.0
2-story SF 0.09 4-story SF 0.4 8-story SF

0.06
0.3
0.04 0.06
0.060 0.091 0.2

0.082
0.051 0.045 0.03
0.02 0.060 0.054
0.029 0.1
0.122 0.110 0.126 0.124
0.00 0.00 0.0
Fig. 8 Variation in repair cost ratio for hilly buildings
denoted as repair cost ratio. Figure 8 shows the repair cost ratio of considered build-
ings. It is observed that stepback buildings without open story have lower repair cost
ratio in comparison to split-foundation buildings without open story. Also, when
open story is present, a decrease in repair cost ratio is observed (in most cases).
Thus, it is concluded that hilly buildings with open story have lower overall repair
cost ratio; however, repair cost of their structural components is significantly higher.
5 Conclusions
In this study, open stories are considered at different locations based on the connec-
tivity with approach road, i.e., no open story, open story at LFL, open story at UFL,
and open stories at both LFL and UFL. Seismic loss estimation, in terms of monetary
losses, is performed for 24 hilly buildings without and with open story. The following
conclusions are drawn from this study:
• Unreinforced masonry wall contributes maximum to the seismic losses. Therefore,
contribution of non-structural elements should not be ignored while performing
seismic loss assessment.
• The floor at uppermost foundation level experiences maximum monetary loss
irrespective of the presence of open story.
• Presence of an open story in hilly buildings increases the damage to structural
components thereby increasing the repair cost of structural components.
• In two-story and four-story hilly buildings, the repair cost ratio of building without
open story building is highest. Whereas in eight-story buildings, the repair cost
ratio is highest with open story at UFL.
References
1. Singh, Y., Gade, P., Lang, D.H., Erduran, E.: Seismic behavior of buildings located on slopes –
an analytical study and some observations from Sikkim earthquake of September 18, 2011. In:
15th World Conference on Earthquake Engineering, Lisboa, Portugal (2012)
2. Surana, M., Singh, Y., Lang, D.H.: Seismic characterization and vulnerability of building stock
in hilly regions. Nat. Hazard. Rev. 19(1), 04017024 (2018)
3. Varum, H., Damaru, R., Furtado, A., Barbosa, A.R., Gautam, D., Rodrigues, H.: Seismic perfor-
mance of buildings in Nepal after the Gorkha earthquake. Impacts Insights Gorkha Earthq.,
Elsevier, pp 47–63 (2018)
4. Sharma, M.L., Maheshwari, B.K., Singh, Y., Sinvhal, A.: Damage pattern during Sikkim, India
earthquake of September 18, 2011. In: 15th World Conference on Earthquake Engineering,
Lisboa, Portugal (2012)
5. Aggarwal, Y., Saha, S.K.: Seismic performance assessment of reinforced concrete hilly
buildings with open story. Structures 34, 224–238 (2021)
6. IS 875 Part-1.: Indian standard code of practice for design loads (other than earthquake) for
buildings and structures. Dead loads- unit weights of building materials and stored materials.
Bureau of Indian Standards,New Delhi, India (1987)
7. IS 875 Part-2.: Indian standard code of practice for design loads (other than earthquake) for
buildings and structures. Imposed loads. Bureau of Indian Standards,New Delhi, India (1987)
8. IS 1893.: Criteria for earthquake resistant design of structures. Bureau of Indian Standards,New
Delhi, India (2016)
9. ASCE 41.: Seismic evaluation and retrofit of existing buildings. American Society of Civil
Engineers, Reston, Virginia, USA (2013)
10. Paulay, T., Priestley, M.J.N.: Seismic design of reinforced concrete and masonry build-
ings.Wiley, New York, p. 744 (1992)
11. FEMA P695. Quantification of buildings seismic performance factors, Washington (2009)
12. PEER.: Pacific earthquake engineering research center next generation of ground motion
attenuation models west 2 database (2012). www.peerberkeley.edu/ngawest2
13. PEER.: User manual for the PEER ground motion database web application, Beta Version
(2010)
14. CSI.: Integrated software for structural analysis & design SAP2000, version 20.0.1. Analysis
reference manual, Computers and Structures Inc., Berkeley, USA (2021)
15. FEMA P58.: Seismic performance assessment of buildings, volume 1- methodology, Federal
Emergency Management Agency, Washington, DC, USA (2018)
16. Cardone, D., Perrone, G.: Developing fragility curves and loss functions for masonry infill
walls. Earthq. Struct. 9(1), 257–279 (2015)
17. Brown, P.C., Lowes, L.N.: Fragility functions for modern reinforced-concrete beam-column
joints. Earthq. Spectra 23(2), 263–289 (2007)
18. Benjamin, J.R., Cornell, C.A.: Probability, statistics, and decision for civil engineers. Courier
Corporation (2014)
19. Aggarwal, Y., Saha, S.K.: Component repair cost functions in indian context for seismic loss
estimation of reinforced concrete buildings. Structures 44, 1974–1994 (2022)
Influence of Source Offset
on the Resolution of Dispersion Image
Ashhad Imam , Virendra Kumar, Keshav Kumar Sharma,

and Alvin Harison
Abstract A non-invasive method called multichannel analysis of surface waves

(MASW), which has been used for geotechnical site characterization since the late
1990s, has become very prevalent. In field surveys, things like the quality of the input
source and the geophone parameters (resolution, spacing, orientation layout) can have
a big impact on the dispersion curve. This paper will look at how the distance between
the source and the first geophone receiver (offset) affects the resolution of disper-
sion images in the MASW method. An experimentation site was chosen within the
NIT Jamshedpur campus. According to borehole data, the location (22°46, 37.2,, N,
86°08, 38.5,, E) has a stiff silty clay soil (up to a depth of 5 m) followed by a dense
to very dense weathered mica schist. A 10 kg sledgehammer was used to strike the
striker plate, which generated surface waves that were recorded by 24 geophones
(of 4.5 Hz frequency) arranged in a linear array pattern. Wave fields were recorded
using a sampling frequency of 1000 Hz and a varying offset distance (1, 2, 4, 6, 8,
10, and 12 m) with 1 m geophone spacing and five stacking. Based on the results
obtained, an optimum offset was determined as 8 m for 1000 Hz sampling frequency
and 1 m geophone spacing, producing a high signal-to-noise ratio of 92%, resulting
in an appropriate resolution of the dispersion image.
Keywords Offset distance · Dispersion images · Active MASW · Geophones
A. Imam (B)
Sam Higginbottom University of Agriculture Technology and Sciences, Prayagraj, Uttar
Pradesh 211007, India
e-mail: ashhad.imam@shiats.edu.in
V. Kumar · K. K. Sharma · A. Harison
National Institute of Technology Jamshedpur, Jamshedpur, Jharkhand 831014, India
https://doi.org/10.1007/978-981-99-1459-3_4
42 A. Imam et al.
1 Introduction
Geophysical investigations concerning the creation and propagation of seismic waves

are frequently employed in earthquake geotechnical engineering. Source mecha-
nisms, seismic wave transmission in the field, and site characteristics all have an
impact on earthquake ground motions [1]. The use of geophysical investigation,
utilizing seismic waves, has made its mark in the recognizance and characterization
of the subsurface. Since the development of techniques of seismic wave for identi-
fying the subsurface characteristics of soil, geophysical exploration has significantly
advanced. It is one of the most popular non-destructive seismic techniques. MASW
uses a multireceiver module to identify subsurface segmentation of layers, i.e., shear-
wave velocity profiling down the depth from the ground surface [2–4]. To determine
the shear-wave velocity (V s ) changes below the identified area, which are primarily
liable for the surface waves’ investigated propagation velocity pattern, MASW
initially monitors seismic surface waves originated by various seismic causes, such
as sledgehammers. The dynamic soil parameter known as Vs has been linked to
the bulk modulus, seismic amplification, fundamental vibration frequency, shear
modulus, and Poisson’s ratio [5]. V s has been successfully used in several impor-
tant geotechnical and earthquake engineering domains, including the evaluation of
seismic site effects, the estimation of soil liquefaction potential, the characterization
of seismic sites, and seismic microzonation studies [6–8]. The strongest energy trend
must be extracted, which makes the accuracy of the V s profile directly dependent on
the dispersion images resolution [9].
Multichannel analysis of surface waves (MASW) is the most popular surface wave
technique worldwide. The approach has peculiar data collecting, analysis, and inter-
pretation procedures. There is a potential for significant uncertainty to be introduced
into the outcomes at each stage of the MASW. Even a single mistake can lead to
erroneous findings that are hard to spot by a third party. Researchers have previously
reported a large number of such occurrences. Since the MASW data are eventually
helpful for the generation of the design response spectrum, such techniques may
result in significant losses [10, 11]. Moreover, the assumption of homogeneous or
horizontally layered soil models is made in order to perform the inversion of the
surface wave dispersion curve; nevertheless, this hypothesis could represent a limi-
tation on the practical use of the MASW technique [12]. Additionally, an effective
validation of results using an invasive technique, such as a borehole survey or cross-
hole/downhole method, must be performed for MASW data in order to support the
accuracy of the information it provides regarding subsurface anomalies. Because
of all of these factors, there is an urgent requirement for the spread of informa-
tion regarding the recommendations for the reliable practise of MASW testing. In
a similar line, a report by Jakka et al. [10] includes a full overview of the MASW
approach, its three steps of data acquisition, processing, and inversion, as well as
methods for increasing confidence in the results.
Several data acquisition parameters, particularly offset distance, have an impact on
the image resolution [2, 13–15]. MASW frequently experiences near-field effects,
Influence of Source Offset on the Resolution of Dispersion Image 43
which can lead to either an underestimation or an overestimation of the Rayleigh

wave phase velocity due to the presence of body wave’s contamination close to the
source [16].
Despite the fact that there has been a lot of study on active MASW surveys, the
majority of them employ different parametric configurations to meet the requirements
of the investigation, according to a critical analysis of the literature. There is hardly
any literature that makes an effort to offer recommendations for the use of different
factors in the ongoing MASW survey to improve the precision of the estimations.
However, it is also found that there is a lack of consensus among the studies on the
most effective variables to make precise subsurface profiles. The trustworthiness of
the result is significantly influenced by the resolution of the image of the dispersion.
The best acceptable factors (offset distance) can be chosen, but it is important to opt
an appropriate resolution of the dispersion images from the acquired field data. This
article’s goal is to emphasize the data acquisition parameter (offset distance) that
governs each testing and evaluation stage.
Owing to the above discussions, a study was planned and the experimental field
data was gathered through an active MASW survey, and a commercial data interpre-
tation software i.e. Parkseis was used that processes and interprets the data. It is found
that varying field topography and combinations result in different dispersion images
with different resolutions. The best dispersion images with the highest resolution are
to be identified with the help of this study. The suitability is assessed using visual
inspections and the signal to noise ratio obtained for different cases of the study.
The present paper demonstrates the effect of varying offset distance along with
the soft site condition on the quality of the dispersion images (a detailed report is
presented by Imam et al. [17] on data acquisition parameters and site characterization
for the proposed site). As part of the field experimental program, an Active MASW
survey was conducted on the NIT Jamshedpur campus in Jamshedpur, India (Fig. 1).
The area (22°46, 37.2,, N, 86°08, 38.5,, E) is characterized by a stiff silty clay soil
(depth up to 5 m) followed by a dense and very dense weathered mica schist. Due
to the existence of rock-type strata, the borehole investigation was limited to a depth
of 15 m. Non-invasive procedures, such as MASW, come into play when invasive
techniques (boreholes) cannot be employed to accomplish a deeper depth exploration.
As per the SPT-N standards measured at the location, the average shear-wave velocity
determined was to be around 350 m/s.
2 Methodology
The three primary phases of an active MASW survey are collection of data, dispersion
analysis, and inversion analysis [2]. A network of on-the-ground geophone receivers
is used to capture the vibration caused by waves that are propagating from an active
impulsive source [18–20]. The dispersion image is produced by converting the time
signatures to the frequency domain and then identifying the dispersion curve using
combinations of phase velocities and frequencies that have the local highest energies
44 A. Imam et al.
0m
Dense silty clay with

(G.L)
Surface Materials
5m
weathered mica
Medium dense
a) Jamshedpur Map
schist
Depth
10 m
weathered mica
Very dense
schist 15 m
c) Lithology from borehole data
b) Study Area (NIT Jamshedpur)
Fig. 1 Location details of the site chosen for field investigation using active MASW within NIT
Jamshedpur campus and the corresponding lithology obtained from borehole data
[18, 21]. The flowchart in Fig. 2 shows steps of the whole process of a MASW survey
(active or passive).
2.1 Details of Test Setup
An active MASW investigation that was carried out in the field is schematically
depicted in Fig. 3. An impulsive sledgehammer strike that passes through the soil
generates seismic waves, which can be detected by a linear arrangement of geophone
receivers. The receivers are connected to a portable seismograph and a computer
system which will act as an interface to present the recorded field data. A 10 kg sledge-
hammer was used to make an impact on the striker plate which further generated
STEP 1: DISPERSION
DATA ACQUISITION DISPERSION ANALYSIS
IMAGING
INVERSION 'GRADIENT
STEP 2: DISPERSION BASED ITERATIVE SHEAR WAVE VELOCITY
CURVE EXTRACTION APPROACH' ANALYSIS - PROFILE 1D & 2D
INVERSION
Fig. 2 Flowchart of MASW method for data acquisition and processing
surface waves propagating on the ground recorded by 24 numbers of geophones (of

4.5 Hz frequency) which were arranged in a linear array pattern. With 1 m geophone
spacing, wave fields were controlled using a sampling frequency of 1000 Hz and
different offset distances (1, 2, 4, 6, 8, and 12 m). Following the collection of the raw
wave fields during field experiments, they are subjected to preprocessing, dispersion,
and inversion analyses, among other stages of analysis. Parkseis 3.0 as processing
software was utilized in the current study to analyze and interpret the field recorded
data and thereby resulting in the generation of an appropriate shear-wave velocity
profile along the depth.
Fig. 3 Schematic diagram for data acquisition setup of MASW

46 A. Imam et al.
To obtain accurate shear-wave velocity profile from dispersion curve, maximum

signal-to-noise ratio (S/N) is required. Also it affects the extraction process from
obtained dispersion image if the resolution is inadequate. After being transmitted
from the inversion process to the next phase, dispersion analysis helps to evaluate
dispersion curves.
To produce an accurate shear-wave velocity profile, data acquisition parameters’
impact should be taken into account for quality of the dispersion images. With the
high signal-to-noise (S/N) ratio, presumably surface wave is one of the important
parameter which is mostly observed in a strong quality data set; however, the noise
is common factor to weak quality data set. Occurrence of all body wave and higher
modes surface wave is counted as noise. This part documents the offset distance
influence for generated dispersion images on visually identifiable characteristics.
3.1 Offset Distance Influence
From the source to first geophone receiver the offset distance, producing high-
resolution dispersion images and reliable wavefield records is critical [2]. To capture
the planer wave field, receiver array is designed, generally surface waves, which
travelled a defined distance after generation of wave field from its source. Waves
which are longer in wavelength travel more prior to becoming planer, and vice versa.
The far-offset effect is generally observed in geophone array, the distance beyond
which impact generated wave field is not properly recorded in geophone array. There
are near-offset effects and far-offset effects which are discussed in the literature [2, 13,
16]. The wave field’s wavelength composition determines the properties of both the
near-offset and far-offset effects. The ideal offset distance in a surface wave survey
must be adjusted to the range of distance from the source where surface waves
are most effectively generated in order to determine subsurface velocity (V s ) for a
planned depth range (e.g., 0–30 m). Source offset (X1 ) and receiver spread length (L)
are two field geometry variables that are closely related. The ideal configuration of
these two factors effectively suggests that there is a requirement of ideal positioning
of offset that it should not to be too close or too far away from seismic source.
In order to assess the impact of the offset distance on the dispersion image, a linear
type of spread was recorded at the site location at a length of 23 m with geophones
spaced at 1 m. In this manner, with a varied offset distances (1, 2, 4, 6, 8, 10, and
12 m), seven MASW data were collected. For spread in 23 m length, the produced
dispersion images are shown in Fig. 5a–g and correspond to the collected wavefield
recordings shown in Fig. 4a–g. Figures 5f and g show the findings for offset distance
of 10 and 12 m, respectively. Frequently, it is also observed that all the dominant
waves are entirely not recorded by the geophone array, which is responsible for
missing data in the collected record [17].
a) Offset Distance = 1 m b) Offset Distance = 2 m c) Offset Distance = 4 m
d) Offset Distance = 6 m e) Offset Distance = 8 m f) Offset Distance = 10 m
g) Offset Distance = 12 m
Fig. 4 Effect of offset distance on seismic wave records obtained from the study region for 512
samples at 1, 2, 4, 6, 8, 10, and 12 m (sampling frequency: 1000 Hz, geophone spacing: 1 m @ five
stacking)
A near-offset effect was identified when tests conducted with 1 and 2 m offset
distances. A weak M0 dispersion band decoded from the dispersion images. M0
dispersion band also termed as weak fundamental mode, also having energy content
mainly gathered at the lower-frequency zone (Refer Fig. 5a–b). When offset distance
was set to 4 m, fundamental mode is found to be less dominant than the higher
mode. For an 8 m offset distance, Fig. 5e) generated fundamental dispersion band is
showing more distinct outcome, but the higher modes are not evident enough when
offset distances are greater than 8 m. Due to far offset effect at 10 and 12 m offset
48 A. Imam et al.
a) Offset Distance = 1 m b) Offset Distance = 2 m
c) Offset Distance = 4 m d) Offset Distance = 6 m
e) Offset Distance = 8 m f) Offset Distance = 10 m
g) Offset Distance = 12 m
Fig. 5 Impact of offset distance on dispersion images at 1, 2, 4, 6, 8, 10, and 12 m (sampling

frequency: 1000 Hz, geophone spacing: 1 m @ five stacking)
distances (Fig. 5f–g), there is dominating adulteration throughout for analysis of

significant frequency range of the generated dispersion images. Due to the low S/N
ratio, adulteration of the obtained dispersion curve occurs, which has an adverse
effect on the shear-wave velocity profile for a considerably reduced depth of study.
The results for the offset distance are consistent with and agree well with the
suggestions made by earlier studies. The recommended values, according to some
sources, can have a tolerance of 20% and need to be updated regularly depending on
testing done in different places with different types of soil [13, 17, 21].
The main purpose of this research is to discover a fair offset distance to obtain
high-resolution dispersion image to match with fundamental curve or M0 dispersion
curve. M0 dispersion curve opens the doors to present the substrata characteristics
efficiently via inverted profiles with a wide range of frequency. For the presented
site, an 8-m offset with 1-m geophone spacing was found to be the best solution
with highest signal-to-noise ratio as 92%. Based on the research with some realistic
uncertainties, to obtain a high-resolution dispersion image for firmer strata, 6–8 m
offset distance can be considered optimum. Based on this optimum offset distance
decides M0 dispersion curve is extracted, and through this, inversion analysis could
be done and shear-wave velocity profile could be developed.
3.2 Selection and Extraction of Dispersion Curve
The outcome of the inversion analysis is substantially impacted by the magnitude of

the dispersion curve that is found in the frequency domain. According to the available
research, the depth of investigation decreases as the amplitude of the frequencies in
a band increases. It is a well-known fact that higher frequencies are related to lower
wavelengths, which can penetrate lower depths in the subsurface in order to disclose
its information. This is one of the reasons why higher frequencies are associated with
lower wavelengths. If a lower-frequency band is chosen, there is a possibility that an
inaccurate reading in shallow depths is obtained.
The dispersion image in Fig. 6a represents the highest concentration of propa-
gating energy over various combinations of frequency and phase velocities, with the
darkest shade being the maximum concentration of propagating energy. The color
represents the amplitude of the signal, while the white dots correspond to the funda-
mental mode (M0 dispersion curve), based on which the signal-to-noise ratio was
maintained as 92% for all the data points.
Figure 6b depicts the 10-layer 1-D shear-wave velocity (V s ) profile obtained
from the inversion analysis of the extracted dispersion curve automatically selected
across a wide range of frequency band. The approach of automatic selection may
lead to a consistent shear-wave velocity profiles leading to a high reliability of the
post-processing analysis.
The soil profile (lithology) from the borehole data acquired from the study location
was compared to the 1-D V s profile obtained (Fig. 1c). The V s profile generated from
the MASW survey with a 8 m offset offered information on deeper strata (up to 30 m)
that could not be studied by borehole survey due to existence of hard strata in the site
of investigation. The information obtained from MASW approach indicated sound
agreement with the lithological information gathered from borehole survey.
50 A. Imam et al.
(a)
(b)
Fig. 6 a Extraction of dispersion curve from the dispersion image (offset 8 m); b 1D shear wave
velocity profile using inversion analysis
4 Conclusions
This study provides a description of the active MASW survey that was carried out in
Jamshedpur, the potential location for NIT Jamshedpur. The suggested site’s under-
lying geology has been researched as part of a larger study for Jamshedpur in prepa-
ration for future foundation excavation and design development [17]. Suggested
observations are submitted below as per the experimental findings:
• The investigations highlighted that suitable sampling frequency is site dependent
[22–24]. This can also be supported by the report published by Penumadu and
Park [24], which states that the acquisition time for rigid/stiff-type medium should
be 500 ms and the sampling frequency should be 1000 Hz. Based on which an
appropriate sampling frequency for the proposed study was adopted as 1000 Hz, it
allows for the suitable completion of phase propagation and wavefield recordings.
• To obtain the best possible record of propagating wave fields, offset ranges were
suitably adopted for the site, such that the requirement of completing phase prop-
agation through the receiver array without causing noise adulteration in the wave-
field record is met, and the planar wave propagation is satisfied, resulting in the
best resolution dispersion image indicating a high range of signal-to-noise ratio.
Based on the results obtained, an optimum offset was determined as 8 m, producing
a high signal-to-noise ratio of 92%, resulting in an appropriate resolution of the
dispersion image.
• Depending on the power of the seismic source used, near and far source effects
become more noticeable as the spread length grows, making the choice of disper-
sion curve for detecting the fundamental mode difficult. Hence, use of large
receiver spacing is avoided since it causes significant attenuation of vibrant energy
as it passes through the array, resulting in incomplete wave propagation. Hence,
it is recommended to adopt a receiver (geophone) spacing of 1 m if a 10 kg
sledgehammer is used for making an impact on the ground [17, 22, 23].
References
1. Joshi, A., Bhardwaj, P.: Site characterisation using multi-channel analysis of surface waves at
various locations in Kumaon Himalayas, India. J. Ind. Geophys. Union 22(3), 265–278 (2018)
2. Park, C.B., Miller, R.D., Xia, J.: Multichannel analysis of surface waves. Geophysics 64(3),
800–808 (1999)
3. Park, C.B., Miller, R.D., Xia, J., Ivanov, J.: Multichannel analysis of surface waves (MASW)-
active and passive methods. Lead. Edge 26(1), 60–64 (2007)
4. Desai, A., Jakka, R.S.: Role of A-priori information in minimizing uncertainties in MASW
testing. Indian Geotech. J. 52, 1182–1196 (2022)
5. Dikmen, U., Arisoy, M.O., Akkaya, I.: Offset and linear spread geometry in the MASW method.
J. Geophys. Eng. 7(2), 211–222 (2010)
6. Eker, A.M., Akgün, H., Kockar, M.K.: Local site characterization and seismic zonation study
by utilizing active and passive surface wave methods: a case study for the northern side of
Ankara, Turkey. Eng. Geol. 151, 64–81 (2012)
7. Picozzi, M., Strollo, A., Parolai, S., Durukal, E., Ozel, O., Karabulut, S., Zschau, J., Erdik,
M.: Site characterization by seismic noise in Istanbul, Turkey. Soil Dyn. Earthq. Eng. 29(3),
469–482 (2009)
8. Trupti, S., Srinivas, K.N.S.S.S., Kishore, P.P., Seshunarayana, T.: Site characterization studies
along coastal Andhra Pradesh-India using multichannel analysis of surface waves. J. Appl.
Geophys. 79, 82–89 (2012)
9. Taipodia, J., Baglari, D., Dey, A.: Recommendations for generating dispersion images of
optimal resolution from active MASW survey. Innov. Infrastruct. Solut. 3, 14 (2018)
10. Jakka, R.S., Desai, A., Foti, S.: Guidelines for minimization of uncertainties and estimation of a
reliable shear wave velocity profile using MASW testing: a state-of-the-art review. In: Sitharam,
T.G., Jakka, R.S., Kolathayar, S. (eds.) Advances in Earthquake Geotechnics. Springer Tracts
in Civil Engineering, pp. 211–253. Springer, Singapore (2023)
11. Jakka, R.S., Roy, N.: Uncertainties in site characterization using surface wave techniques
and their effects on seismic ground response. In: Geotechnics for Natural and Engineered
Sustainable Technologies. Developments in Geotechnical Engineering, pp 371–383. Springer,
Singapore (2018)
52 A. Imam et al.
12. Evangelista, L., Magistris, F.S.: Some limits in the use of the MASW technique in soils with
inclined layers. Geotech. Geol. Eng. 33, 701–711 (2015)
13. Park, C.B.: Imaging dispersion of MASW data-full vs. selective offset scheme. J. Environ.
Eng. Geophys. 16(1), 13–23 (2011)
14. Socco, L.V., Boiero, D., Foti, S., Wisen, R.: Laterally constrained inversion of ground roll from
seismic reflection records. Geophysics 74(6), 35–45 (2009)
15. Kanli, A.I., Tildy, P., Pronay, Z., Pinar, A., Hermann, L.: Vs 30 mapping and soil classification
for seismic site effect evaluation in Dinar region, SW Turkey. Geophys. J. Int. 165(1), 223–235
(2006)
16. Roy, N., Jakka, R.S.: Near-field effects on site characterization using MASW technique. Soil
Dyn. Earthq. Eng. 97, 289–303 (2017)
17. Imam, A., Sharma, K.K., Kumar, V., Singh, N.: Subsurface profiling of a region in Jamshedpur
city using active MASW: a case study. Acta Geophys. 70(4), 1601–1617 (2022)
18. Park, C.B., Miller, R.D., Xia, J.: Imaging dispersion curves of surface waves on multi-
channel record. In: Proceedings of the 68th Annual International Meeting of Society of Explo-
ration Geophysics. SEG Technical Program Expanded Abstracts, pp. 1377–1380. Society of
Exploration Geophysicists (1998)
19. Xia, J., Miller, R.D., Park, C.B., Ivanov, J., et al.: Utilization of high-frequency Rayleigh waves
in near-surface geophysics. Lead. Edge 23(8), 753–759 (2004)
20. Xia, J., Miller, R.D., Xu, Y., Luo, Y., Chen, C., Liu, J., Ivanov, J., Zeng, C.: High-frequency
Rayleigh-wave method. J. Earth Sci. 20, 563–579 (2009)
21. Park, C.B., Miller, R.D., Xia, J.: Offset and resolution of dispersion curve in multichannel anal-
ysis of surface waves (MASW). In: 14th EEGS Symposium on the Application of Geophysics
to Engineering and Environmental Problems, pp. 1–6. European Association of Geoscientists &
Engineers (2001)
22. Park, C.B., Miller, R.D., Miura, H.: Optimum field parameters of an MASW survey. In: Japanese
Society of Exploration Geophysics (SEG-J), Extended Abstracts, pp. 1–6 (2002)
23. Foti, S., Hollender, F., Garofalo, F., et al.: Guidelines for the good practice of surface wave
analysis: a product of the InterPACIFIC project. Bull. Earthq. Eng. 16, 2367–2420 (2018)
24. Penumadu, D., Park, C.B.: Multichannel analysis of surface wave (MASW) method for
geotechnical site characterization. Geo-Frontiers Congress, Texas, United States (2005)
Co-seismic Deformation of Iran, 2021
Earthquake Using DInSAR Technique
Hardeep, A. Bahuguna, K. Arun Saraf, and J. Das
Abstract In the present study, the co-seismic deformation of an earthquake of M w

6.4 magnitude (maximum intensity between VII-IX) occurred NNW side of the
Bandar Abbas in Iran on 14th November 2021 at 12:08:38 (UTC) was estimated.
The epicentre of the event was 27.73° N, 56.068° E with a focal depth of 10 km.
The epicentral region was near the southern margin of the collision zone of the
Eurasian plate and the Arabian plate. Co-seismic deformation of the earthquake was
estimated using the differential InSAR (DInSAR) technique, which is carried out
using microwave SAR data pairs from both ascending and descending pass of the
Sentinel-1A satellite. LOS ground deformation for the Iran earthquake that occurred
on 14th November 2021 is estimated using GMTSAR open software. SRTM DEM
of 30 m is used in the processing of the SAR data. The preliminary results indicate
that the earthquake deformation field has thrust fault characteristics. In ascending
pass, subsidence of about − 200 mm SSE side of the fault, i.e. away from the line
of sight (LOS) of satellite and uplift of about 300 mm NNW side of the fault, i.e.
towards the line of sight of the satellite. In descending pass, subsidence of about −
135 mm SSE side of the fault and uplift of about 348 mm NNW side of the fault.
Keywords Sentinel-1A · DInSAR · Co-seismic deformation · Thrust fault ·

Microwave SAR data
Hardeep (B) · K. A. Saraf

Department of Earth Sciences, IIT Roorkee, Roorkee, Uttarakhand, India
e-mail: hardeep@es.iitr.ac.in
K. A. Saraf
e-mail: arun.saraf@es.iitr.ac.in
A. Bahuguna · J. Das
Department of Earthquake Engineering, IIT Roorkee, Roorkee, Uttarakhand, India
e-mail: abahuguna1@eq.iitr.ac.in
J. Das
e-mail: josodhir.das@eq.iitr.ac.in
https://doi.org/10.1007/978-981-99-1459-3_5
54 Hardeep et al.
1 Introduction
Early use of remote sensing in earthquake studies are used to decipher or differ-
entiate the active fault lines and lineaments. Later, remote sensing outspreads its
applications towards detecting pre-seismic, co-seismic and post-seismic changes
and deformations. Synthetic Aperture Radar Interferometry (InSAR) which comes in
the microwave electromagnetic band, is rapidly growing remote sensing technology
which is begin used extensively in the study of the seismology. Iran regions come
under the active seismic zones and observed some destructive earthquake such as
Bam (2003), Saravan (2013) and Iran-Iraq border (2017) earthquakes. Several works
on these earthquakes have been done using DInSAR techniques such as [1–4].
Microwave SAR data images used since late 1980s to study the ground deforma-
tion due to earthquake using DInSAR technique [5–7]. SAR raw data images are
generated after high data processing. With time, accuracy in spatial and temporal
resolution is increased with improved instruments and increased number of satellites
to provide data [8, 9]. This off-field DInSAR technique have some major advantages
than on-field techniques (GPS, Triangulation surveys), like less time consuming, low
cost of money, estimation of ground deformation of inaccessible areas, high accu-
racy, etc. [10–12]. Post seismic deformation is estimated by [13] of the 2001 Bhuj,
India earthquake. Fault plane solutions are also derived from the estimated ground
deformation using DInSAR technique [14–18].
In this work, Sentinel-1A, 1B SAR data images are used to estimate the ground
deformation of Iran earthquake occurred NNW side of the Bandar Abbas. Sentinel-
1A, 1B data are available at European Space Agency (ESA) Copernicus Open Access
Hub. Generic Mapping Tool SAR (GMTSAR) is used for the ground deformation
estimation. Shuttle Radar Topography Mission (SRTM) DEM of 30 m spatial resolu-
tion is used in the processing. Ground deformation for both ascending and descending
pass of sentinel-1A are estimated.
2 Study Area
An earthquake of M w 6.4 magnitude NNW side of the Bandar Abbas in Iran on

14th November 2021 at 12:08:38 (UTC) occurred. The epicentre of the event is
27.73° N, 56.068° E with a focal depth of 10 km. The preliminary fault plane solution
indicates that the earthquake occurs due to shallow dipping north-eastward thrust
fault or steeply dipping south-westward thrust fault. Iran plateau is seismically active
region where some major and devastating earthquake occurred in history. Based on
the geological features, tectonic and seismic activity, Iran plateau are divided into
many seismotectonic zones [19–22] but mainly in 6 zones as Alborz in north side,
Azerbaijan in NW side, Zagros Mountain chains in SW side, Cenrtal Iran, Kopet
Dagh in NE side and Makran Thrust in SE side [23].
Co-seismic Deformation of Iran, 2021 Earthquake Using DInSAR … 55
Fig. 1 Seismo-tectonic map of the study area
Iran (2021) earthquake occurs in young Zagros Mountains zone shown in Fig. 1
which is active continental collision zone and experience high seismic activity in the
regions. The Arabian plate is moving towards the north with respect to the Eurasia
at a rate of about 23 mm/yr. [24]. Zagros Thrust and Fold Zone (ZTFZ) are the
resulted from Arabian and Eurasian plate collisions. The Zagros Mountains have
been upthrust due to the result of northwest converging of these two plates.
In this zone, accumulated stress releases with small rupture deformation so that
high numbers of small magnitude events in range of M w 3–6 occurs more than other
zones and events with M w greater than 6 occurs with less return period as compare
to other parts of the Iran active zones [23]. Fault plane solutions of most earthquake
occurs in Zagros Mountains zone have thrust fault characteristics [25]. Iran-Iraq
Border (2017) earthquake also occurred in the Zagros Mountain chain and had thrust
fault characteristics.
3 Methodology and Data Set
Differential SAR Interferometry (DInSAR) technique uses for the estimation of

ground deformation due to an earthquake. In this technique, the Microwave SAR data
images are used which contains reflection amplitude as well as the phase information
of the received signal. Sentinel-1A, 1B raw data images are available at European
56 Hardeep et al.
Space \Agency (ESA) Copernicus Open Access Hub (https://scihub.copernicus.eu/

dhus/#/home). Sentinel-1A launched on 3rd April 2014 and provided SAR data under
the Copernicus Programme by ESA and have 12 days of revisit time period. After
the launch of Sentinel-1B, the revisit time period reduces to 6 days. This satellite
recorded the SAR data in C-band of microwave portion of EM spectrum and contains
three sub-swaths for the complete one image coverage of 250 km areas. Each sub-
swath further includes nine continue bursts and spatial resolution of SAR image is 5
and 20 m along range and azimuth directions, respectively.
Phase information of SAR data is used for the estimation of ground deformation.
Two SAR images, one before the occurrence of the earthquake (Master image) and
another after the occurrence of the earthquake (Slave image) is used in DInSAR
technique. Co-registration is the process of registering the pixels of master and slave
image so that they represent the same surface area and this process significantly
affects the accuracy of the deformation results. Then interferogram is generated
which is process of complex multiplication of master and slave image in which their
amplitude gets multiplied and phase information gets subtracted. Since phase is lies
in the range of 0 to 360 degree or from − π to π, the phase information lies in
wrapped form (its value repeat after 2π). With interferogram, coherence map also
generated which provides reliability or accuracy of estimated ground deformation.
The wrapped phase have many unwanted source along with ground deformation
source and errors which are removed by some correction and filtering on the wrapped
phase (Φ). Wrapped phase of different sources are as:
Φint = Φ f + Φtopo + Φde f + Φatm + Φerr (1)
where Φf is flat earth phase; Φtopo is topographic phase; Φdef is deformation

phase; Φatm is atmospheric phase; Φerr is error component. After removing all the
unwanted wrapped phase from the total wrapped phase map, only deformation phase
values obtained. Then, this wrapped phase are unwrapped with using statistical-cost
network-flow algorithm for phase unwrapping (SNAPHU). Then, unwrapped phase
is converted into LOS ground deformation of the satellite. Since the wavelength of
Cband is in the range of cm, LOS ground deformation values also lies in the same
range. Flow chart of processing steps in GMTSAR are shown in Fig. 2. Sentinel-
1A SAR data images used for LOS ground deformation estimation of Iran (2021)
earthquake are given in Table 1.
4 Results and Discussion
An earthquake of M w 6.4 occurred 63 km NNW of Bandar abbas, Iran on 14th

November, 2021 at 12:08:38 (UTC). At least death of 1 person is reported after
the event. LOS ground deformation for both passes of Sentinel-1A, i.e. ascending
and descending is estimated. Pair of SAR data for ascending pass are from 13th
November, 2021 as master image and from 25th November, 2021 as slave image
Fig. 2 Flow chart of processing in GMTSAR
Table 1 Sentinel-1A (IW) SAR data set parameters of Iran earthquake

Sensor Incident angle Data type Pass Path Frames Orbit Acquisition
(θ i ) date
Sentinel-1A 40.5°–41.9° Master Descending 166 499 40488 09/11/2021
Slave Descending 166 499 40663 21/11/2021
Sentinel-1A 40.1°–41.5° Master Ascending 57 83 40554 13/11/2021
Slave Ascending 57 83 40729 25/11/2021
is used. Similarly, for descending pass, pair of SAR data are from 9th November,
2021 as master and 21st November, 2021 as slave image is used. Wrapped phase or
interferogram of Iran (2021) for both pass is shown in Fig. 3a and b.
Two well developed fringe lobes are observed in both ascending and descending
wrapped phase. In ascending pass (Fig. 3a), one large wrapped phase fringe lobe in
the North and NW side and slightly smaller fringe lobe in South side can be observed.
Near the fault line, fringes becomes relatively narrow indicating steeper change in
the terrain area. In North side, more deformation occurs which forms narrow fringes
in the lobe and in South side, relatively less deformation occurs which indicated by
the wider fringes in the lobe. Similarly in the descending pass (Fig. 3b), two fully
developed fringe lobes are observed. One large wrapped phase lobe in the North and
NW side with relatively narrow fringe indicates the more deformation in this part.
Another slightly smaller fringe lobe in South side with wider fringe indicates less
deformation in this part.
58 Hardeep et al.
Fig. 3 Wrapped phase map of Iran (2021) earthquake a for ascending pass b for descending pass.
Phase is wrapped in π to − π range. In ascending pass, one large, wrapped fringe lobe in North,
NW side and one small fringe lobe in South side are observed. In descending pass, one slightly
larger fringe lobe in North, NW and other smaller lobe in South are observed
Due to the different incident angle of sensor (angle between LOS of sensor onboard
satellite and nadir line of satellite), slight changes in the fringe lobe area in ascending
and descending mode can be observed. Incident angle in ascending pass is from 40.1°
to 41.5° whereas in descending pass, it is from 40.5° to 41.9° as given in Table 1. So,
LOS ground deformation estimated by the DInSAR technique have imbedded effects
of incident angle over the area. After removing the wrapped phase from topographic
source, flat earth source, atmospheric source and applying filtering for removal of
any errors, wrapped phase due to deformation source is obtained, which is further
unwrapped using the SNAPHU. The unwrapped phase obtained using SNAPHU for
both passes are shown in Fig. 4a and b.
One unwrapped phase is equal to one complete wavelength of microwave radiation
used. In ascending pass (Fig. 4a), large unwrapped phase lobe in North, NW side and
slightly smaller lobe in South side can be observed. Range of unwrapped phase is
from − 66.2 to 45.2. Similarly in descending pass (Fig. 4b), large unwrapped phase
lobe in North, NW side and smaller lobe in South side can be observed with range
of 77.2 to 29.39.
Estimated LOS ground deformation due to Iran (2021) earthquake is shown in
Fig. 5a and b. In ascending pass (Fig. 5a), upliftment of about 209 mm in North, NW
side is estimated, i.e. towards the satellite. Similarly, subsidence of about − 199 mm
in South side is estimated, i.e. away from the satellite. In descending pass (Fig. 5b),
upliftment of about 340 mm in North, NW side is estimated and subsidence of about
129 mm in South side is estimated.
Further, in ascending pass, incident angle vary from 40.1° to 41.5° (left to right).
In ascending deformation map (Fig. 5a), it can be observed that in South side where
subsidence occurs, i.e. movement away from LOS of satellite; however, have low
incident angle value so it have more accurate result of subsidence. Similarly in
North, NW side where upliftment occurs, i.e. movement towards the LOS of satellite,
however it have more incident angle value so it have slightly less accurate results
of upliftment. For the descending pass, range of incident angle are from 40.5° to
41.9° (right to left). In descending deformation map (Fig. 5b), in South side where
subsidence occurs, i.e. movement away from LOS of satellite and also have large
incident angle value so it has low values of subsidence and less accurate as compared
to ascending deformation map (Fig. 5a). Further, in North, NW side where upliftment
occurs, i.e. movement along LOS of satellite and have low incident angle value, it
have more value of upliftment and more accurate results as compared to ascending
deformation map (Fig. 5a).
5 Conclusion
Sentinel-1A SAR data images are used to analyse the Iran (2021) earthquake of
M w 6.4 magnitude (maximum intensity between VII-IX) with focal depth of 10 km.
Ground deformation is estimated using DInSAR technique with GMTSAR. Esti-
mated ground deformation have thrust fault characteristics like other earthquakes
60 Hardeep et al.
Fig. 4 Unwrapped phase map of Iran (2021) earthquake a for ascending pass b for descending
pass. One unwrapped phase is equal to one complete wavelength of microwave radiation used. In
ascending pass, one large, unwrapped lobe in North, NW side and small unwrapped phase lobe in
South side are observed. In descending pass, one slightly larger fringe lobe in North, NW and other
smaller lobe in South are observed
Fig. 5 LOS ground deformation of Iran (2021) earthquake a for ascending pass b for descending
pass. In ascending pass, large lobe in North, NW side indicates upliftment and relatively smaller lobe
in South side indicates subsidence. In descending pass, large lobe along North, NW side indicate
upliftment and along South side indicates subsidence
62 Hardeep et al.
have in Zargos mountains region of Iran. DInSAR-based ground deformation have

accuracy in cm range which provides good results in less time period and also in
the inaccessible areas compares to other field-based ground deformation estimation
techniques. Using the high resolution DEM also improves the final results. Based
on DInSAR technique, short term (earthquake) and long term (time series analysis)
ground deformation can be estimated in seismically active regions.
References
1. Huang, Z., Zhang, G., Shan, X., Gong, W., Zhang, Y., Li, Y.: Co-seismic deformation and fault
slip model of the 2017 Mw 7.3 Darbandikhan, Iran–Iraq earthquake inferred from D-InSAR
measurements. Remote Sens. 11(21), 2521 (2019)
2. Liu, F., Elliott, J.R., Craig, T.J., Hooper, A., Wright, T.J.: Improving the resolving power of
InSAR for earthquakes using time series: a case study in Iran. Geophys. Res. Lett. 48(14),
e2021GL093043 (2021)
3. Peyret, M., Rolandone, F., Dominguez, S., Djamour, Y., Meyer, B. Source model for the Mw
6.1, 31 March 2006, Chalan-Chulan earthquake (Iran) from InSAR. Terra Nova 20(2), 126–133
(2008)
4. Vajedian, S., Motagh, M., Mousavi, Z., Motaghi, K., Fielding, E.J., Akbari, B., Wetzel, H-
U., Darabi, A.: Coseismic deformation field of the Mw 7.3 12 November 2017 Sarpol-e Zahab
(Iran) earthquake: a decoupling horizon in the northern Zagros Mountains inferred from InSAR
observations. Remote Sens. 10(10), 1589 (2018)
5. Gabriel, A.K., Goldstein, R.M., Zebker, H.A.: Mapping small elevation changes over large
areas: differential radar interferometry. J. Geophys. Res. Solid Earth 94(B7), 9183–9191 (1989)
6. Massonnet, D., Rossi, M., Carmona, C., Adragna, F., Peltzer, G., Feigl, K., Rabaute, T.:
The displacement field of the Landers earthquake mapped by radar interferometry. nature
364(6433), 138–142 (1993)
7. Scantland, S., Biegert, E.: Radar locates offshore oil slicks. Oceanogr. Lit. Rev. 3(44), 279–280
(1997)
8. Fielding, E.J., Lundgren, P.R., Burgmann, R., Funning, G.J.: Shallow fault-zone dilatancy
recovery after the 2003 Bam earthquake in Iran. Nature 458, 64–68 (2009)
9. Stramondo, S., Moro, M., Doumaz, F., Cinti, F.R.: The 26 December 2003, Bam, Iran earth-
quake: surface displacement from Envisat ASAR interferometry. Int. J. Remote Sens. 26,
1027–1034 (2005)
10. Rosen, P.A., Hensley, S., Joughin, I.R., Li, F.K., Madsen, S.N., Rodriguez, E., Goldstein, R.M.:
Synthetic aperture radar interferometry. Proc. IEEE 88(3), 333–382 (2000)
11. Sharma, K., Saraf, A.K., Das, J., Baral, S.S., Borgohain, S., Singh, G.: Mapping and change
detection study of Nepal-2015 earthquake induced landslides. J. Indian Soc. Remote Sens.
46(4), 605–615 (2018)
12. Zia, M., Sharma, K., Saraf, A.K., Das, J., Baral, S., Das, M.: Ground deformational studies
using ALOS-PALSAR data between 2007 and 2010 of the central Kutch area, Gujarat, India.
Nat. Hazards 71(3), 1379–1388 (2014)
13. Saraf, A.K., Das, J., Biswas, A., Rawat, V., Sharma, K., Suzat, Y.: SAR interferometry in
post-seismic ground deformation detection related to the 2001 Bhuj earthquake, India. Int. J.
Remote Sens. 33(4), 1296–1308 (2012)
14. Ferretti, A., Savio, G., Barzaghi, R., Borghi, A., Musazzi, S., Novali, F., Prati, C., Rocca, F.:
Submillimeter accuracy of InSAR time series: experimental validation. IEEE Trans. Geosci.
Remote Sens. 45(5), 1142–1153 (2007)
15. Ganas, A., Elias, P., Briole, P., Cannavo, F., Valkaniotis, S., Tsironi, V., Partheniou, E.I.: Ground
deformation and seismic fault model of the M6.4 Durres (Albania) Nov. 26, 2019 earthquake,
based on GNSS/INSAR observations. Geosciences 10(6), 210 (2020)
16. Li, Z., Cao, Y., Wei, J., Duan, M., Wu, L., Hou, J., Zhu, J.: Time-series InSAR ground deforma-
tion monitoring: atmospheric delay modeling and estimating. Earth-Science Rev. 192, 258–284
(2019)
17. Osmanoğlu, B., Sunar, F., Wdowinski, S., Cabral-Cano, E.: Time series analysis of InSAR
data: methods and trends. ISPRS J. Photogramm. Remote. Sens. 115, 90–102 (2016)
18. Sakkas, V. Ground deformation modelling of the 2020 Mw 6. 9 Samos earthquake (Greece)
based on INSAR and GNSS data. Remote Sens. 13(9), 1665 (2021)
19. Niazi, M., Basford, J.R.: Seismicity of Iranian plateau and Hindu Kush region. Bull Seismol
Soc Am 58(1), 417–426 (1968)
20. Nowroozi, A.A.: Seismo-tectonics of the Persian plateau, eastern Turkey, Caucasus, and Hindu-
Kush regions. Bull Seismol Soc Am 61(2), 317–341 (1971)
21. Shoja-Taheri, J., Niazi, M.: Seismicity of the Iranian plateau and bordering regions. Bull.
Seismol. Soc. Am. 71(2), 477–489 (1981)
22. Wilson, A.T.: Earthquakes in Persia. Bull. Sch. Orient. Afr. Stud. 6(01), 103–131 (1930)
23. Salamat, M., Zöller, G., Zare, M., Amini, M.: The maximum expected earthquake magnitudes
in different future time intervals of six seismotectonic zones of Iran and its surroundings. J.
Seismolog. 22(6), 1485–1498 (2018)
24. Barka, A., Reilinger, R. (1997). Active tectonics of the Eastern Mediterranean region: deduced
from GPS, neotectonic and seismicity data. Ann. Geophys. 40(3), 587–610 (1997)
25. Jackson, J., McKenzie, D.: The relationship between plate motions and seismic moment tensor,
and the rates of active deformation in Mediterranean and Middle East. Geophys. J. Int. 93(1),
45–73 (1988)
Three-dimensional Crustal Velocity
Structure of Tehri, Garhwal Himalaya
R. Modi, S. Mukhopadhyay, and M. L. Sharma
Abstract We estimated a three-dimensional (3D) structure of crustal velocity of

Tehri, Garhwal Himalaya using seismic data of local earthquake events collected
from January 2008 to December 2015 by a seismological network of 12 remote
stations. The network is spread over an area of nearly 100 × 80 km around Tehri
dam. The study area between 29°50´ N to 31°50´ N latitude and 77°50´ E to 80°E
longitude, falls in the Garhwal Himalayan region of the central seismic gap between
Main Central Thrust (MCT) and Main boundary Thrust (MBT). Inversions were
performed simultaneously for P- and S-wave velocity anomalies and source coordi-
nates. Dataset for tomographic inversions consists of 3543 local earthquake events
with 29451 P- and 28692 S-wave picks. Results of inversion include P- and S-wave
velocity anomalies and revised locations of earthquake sources. Tomograms of P-
and S-wave velocity anomalies point to strong lateral heterogeneities in the region
of investigation. Several zones of low velocity are depicted in the Lesser Himalayan
region of the study area which indicates the presence of fluid filled structures in the
region. Our velocity model shows that the Uttarkashi earthquake of 1991 is located
at the transition zone of a high and a low velocity anomaly. Tomographic inversions
could well resolve the investigated area up to a depth of 20 km.
Keywords Tomography · Local earthquake data · Three-dimensional velocity

model · Crustal velocity structure · Travel time inversion · Garhwal Himalaya
1 Introduction
A realistic three-dimensional (3D) velocity model besides improving the location

accuracy of earthquakes also sheds light on significant physical processes responsible
R. Modi (B) · S. Mukhopadhyay

Seismology Division, Bhabha Atomic Research Centre, Trombay, Mumbai 400085, India
e-mail: rajnim@barc.gov.in
M. L. Sharma
https://doi.org/10.1007/978-981-99-1459-3_6
66 R. Modi et al.
for the occurrence of earthquakes and the material composition of the earth. Improved
earthquake location accuracy facilitates improving our tectonic understanding of
the earth, for example, by means of seismicity studies. Significant efforts have been
made in this area and various techniques have been developed over the years to model
the 3D velocity structures of the earth such as local earthquake tomography, surface
wave tomography and ambient noise tomography. An appropriate technique can be
chosen depending on the type of data available.
Numerous seismic networks are operative the world over on various scales and
the abundance of data available from these networks is being efficiently utilized to
develop 1D and 3D realistic velocity models. In the present work, the focus is on the
estimation of a realistic 3D structure of P- and S-wave velocities for the Tehri region
of Garhwal Himalaya which is a highly seismically active region, also important from
socio-economic point of view. The presence of a 260.5 m high earth and rock-fill
dam, with a storage volume of 3.54 km3 in Tehri also calls to a study of the region
to estimate the effect of reservoir-induced seismicity in the region, if any.
As the realistic velocity model to be estimated depends on the unknown earthquake
locations which in turn depend on the unknown velocity models, we have employed
the method of simultaneous inversion where both velocity models and earthquake
locations are inverted simultaneously.
The estimated realistic velocity model and the relocated earthquake locations will
help to improve the location of future earthquakes originating from this area thereby
improving the tectonic understanding of the region.
2 Study Area
Formed due to the continued convergence of the Indian and the Eurasian tectonic
plates at approximately 4 cm/year, the Himalayas are one of the most seismically
active regions of the world. Along this Himalayan arc of approximately 2400 km
length on the northern border of the Indian subcontinent, several thrust zones have
been postulated, the most prominent being the Main central thrust (MCT), Main
boundary thrust (MBT) and Main frontal thrust (MFT) [1]. Khattri [2] proposed a
seismic gap between the locations of the Kangra earthquake, 1905 (Mw 7.8) and the
Bihar-Nepal earthquake, 1934 (Mw 8.0) called central gap, a 700 km long section
on the plate boundary. Several moderate and large earthquakes have occurred in
the Central seismic gap over the years but no great earthquake has occurred in the
region in the past 200 years. The space geodetic measurements in the region suggest
a substantial slip deficit [3]. The region is, thus, of great interest to seismologist.
The study area bounded between 29°50´ N–31°50´ N latitude and 77°50´ E–80° E
longitude falls in the Garhwal Himalayan region of the central seismic gap between
MCT and MBT. Figure 1 shows major thrust faults of the study area, such as MCT,
MBT, MFT and Srinagar thrust.
Three-dimensional Crustal Velocity Structure of Tehri, Garhwal Himalaya 67
Fig. 1 Distribution of seismic events (black outlined magenta colored circles) in the study area.
Gray lines show the ray paths from sources to receivers, green triangles show recording stations,
major fault lines are shown in black, prominent rivers in the region are shown in blue and Tehri
reservoir is shown in cyan color. Yellow stars indicate the location of the Uttarkashi earthquake,
1991 and the Chamoli earthquake, 1999
3 Seismic Observation Network and Dataset
The seismic observation system is a local seismic network comprising 12 recording

stations operative in the environs of Tehri dam located in the region of Garhwal
Himalaya. The network covers an approximate area of 100 × 80 km. Each recording
station is equipped with a three-component short-period seismometer (Model: CMG
40T-1, M/s Guralp system, UK) to record the ground motion. The data acquisition
system (Model: 130–01/03, M/s REFTEK, USA) digitizes the recorded signal using a
24-bit digitizer at a rate of 100 samples/sec/component. The geographical parameters
of the network stations are listed in Table 1.
The dataset for the study comprises of 3543 local seismic events recorded by
the above-mentioned recording network from January 2008 to December 2015.
Initial location analysis of seismic events was carried out by using SEISAN soft-
ware package [4] with a three-layer 1D velocity model given by Kumar et al. [5].
The criteria for selecting the events for 3D tomographic inversions include greater
than or equal to 10 (P + S) phase recordings and a maximum distance to the nearest
station to be less than 100 km. The dataset, thus, is not restricted to events that orig-
inated from within the network only. Including out-of-the-network events provides
a higher ray density and better ray coverage in the study area and thereby improves
the results of 3D tomographic inversion. The use of out-of-the-network events in
68 R. Modi et al.
Table 1 Parameters of the recording stations of the local seismic network around Tehri region
Station no. Station code Geographical coordinates Elevation (m)
Latitude (N) Longitude (E)
1 AYR 30° 18.18’ 78° 25.86’ 2106
2 CHN 30° 18.31’ 78° 37.14’ 2244
3 CNT 30° 24.59’ 78° 44.69’ 1953
4 KHU 30° 34.81’ 78° 29.68’ 1730
5 NTT 30° 22.58’ 78° 25.78’ 1914
6 PRT 30° 27.48’ 78° 28.54’ 2128
7 SRL 30° 07.92’ 78° 38.03’ 1424
8 SRT 30° 36.76’ 78° 17.93’ 1617
9 SUR 30° 24.66’ 78° 17.39’ 2754
10 GYN 30° 45.34’ 78° 25.26’ 2113
11 RAJ 30° 50.64’ 78° 14.29’ 1908
12 VIN 30° 33.99’ 78° 39.32’ 1640
tomographic inversion studies is very well reasoned by Koulakov [6]. As shown in

Fig. 1, the source-receiver pairs used in the study provide a good ray density and
good ray coverage in the investigated area, especially in the region where network
stations are located. As seismic activity occurs at shallow depth in the study region,
in vertical direction, the ray path coverage is up to a depth of 20 km.
4 Methodology
For 3D local earthquake tomography, an algorithm called Local Earthquake Tomog-

raphy Software (LOTOS 12) was employed which works on the principle of simulta-
neous inversion of earthquake locations and velocity models [7]. The process starts
with a rough estimation of source locations based on a grid search algorithm where
reference travel-time tables constructed using simplified methods of ray tracing are
used. The sources are then relocated in a 3D velocity model using a more sophis-
ticated 3D ray bending tracing algorithm based on Fermat’s principle of travel-
time minimization. For parameterization of velocity distribution, a grid of nodes is
defined in the study volume depending on the density of rays between the sources and
receivers. No nodes are defined in the region where there are no rays. To reduce any
effect of parameterization of nodes, inversions are performed with four different grid
orientations. Velocity anomalies from all the grids are combined for final results.
The matrix inversions are performed simultaneously for P- and S-wave velocity
anomalies, source parameters and station corrections using LSQR method [8, 9].
Input data to the code are P- and S-wave travel-time data of local earthquakes
and station coordinates. One-dimensional (1D) P and S-wave velocity models are
given as initial reference velocity models where velocity values vary vertically with
depth. Velocities are linearly interpolated between different depth levels. A set of free
parameters for source locations, grid construction and inversion, etc. are required to
be defined. Further, damping parameters are needed to be defined to regulate the
amplitude of anomalies and the smoothness of solution. Optimal values of damping
parameters are estimated from the inversions performed with synthetic models which
are given in the form of some anomalies. Damping parameters are set so as to obtain
the best recovery of synthetic models after inversion. The same values of damping
parameters are then used for real datasets.
5 Results
1D crustal velocity model to be used as initial velocity model for tomography

studies is derived using VELEST algorithm [10] which solves the coupled hypocenter
velocity model problem by simultaneous inversion of travel-time data of earthquake
sources and velocity model. A dataset of 225 well-located earthquake events is used
to obtain the 1D velocity model for P- and S-waves which is shown in Table 2.
Because of the shallow nature of seismicity, inversion for the 1D velocity model
resolved the velocity structure of the upper crust up to a depth of 20 km. Moho
depth of ~ 44 km and the corresponding velocity information is obtained from the
travel-time curves using cross-over distance technique. This model is used as initial
reference velocity model for 3D tomography studies.
Based on the results of synthetic model tests, parameters like 3D grid spacing,
smoothness and amplitude damping are optimized. A number of test runs are
performed with LOTOS to fix the most optimal values of these inversion param-
eters. In lateral directions, the grid nodes are placed a distance of 2 km each while in
vertical direction node spacing is set at 1 km. Due to the topography of the study area
and the altitude of recording stations, the first grid is defined at an altitude of 3 km
above sea level. Inversions are carried out for the 3D velocity models and source
Table 2 1D velocity model

Depth to the top of P-wave velocity S-wave velocity
for Tehri, Garhwal Himalaya
layer (km) (km/s) (km/s)
used as initial reference
velocity model 0 5.56 3.28
4 5.75 3.36
8 5.75 3.43
10 5.93 3.43
14 6.13 3.71
18 6.13 3.88
20 6.47 3.88
44 7.69 4.55
70 R. Modi et al.
Table 3 Average residuals of P- and S-waves before and after inversion and total variance reduction
Initial average residual (s) Average residual after fourth Total variance reduction
iteration (s)
P- wave S- wave P- wave S- wave P- wave S- wave
0.23 0.32 0.16 0.22 29.58 28.44
locations with the optimized inversion parameters. Table 3 shows average travel-
time residuals of P- and S-waves and total variance reduction after four iterations of
tomographic inversions.
The reduction in the average travel-time residual after 3D tomographic inversions
is a clear indication of an improvement in the earthquake locations accuracy with
3D velocity model. The P- and S- wave velocity anomalies are computed at various
horizontal depth levels such as 2, 4, 6, 8, 10, 12, 14, 16, 18 and 20 km. The tomograms
corresponding to P- and S- wave velocity anomalies at depth sections 4, 10, 14 and
20 km are presented in Fig. 2.
5.1 Three Dimensional Velocity Structure
The 3D velocity structure of the study area is given as a percentage deviation of P-

and S-wave velocities from the reference model velocities at respective depth levels.
To address the issue of uncertainty involved in S phase picking as compared to P
phase picking, we used a 1.3 times larger weighting factor for P phases as compared
to S phases for tomographic inversions. Further, based on the results of inversions
performed with the checkerboard synthetic model, for the S-velocity model we used
a smoothness parameter 2.3 times larger and a damping parameter 3 times larger
than for the P-velocity model.
Figure 2 shows P- and S-wave velocity anomalies in the horizontal sections at
depths of 4, 10, 14 and 20 km with a backdrop of the most prominent tectonic features
in the area of investigation. Strong lateral heterogeneities are clearly represented by
the tomograms in the study area. A number of discontinuous high and low velocity
zones can be seen scattered in the investigated area at various depth levels. High
density of rays from sources to receivers in the central part of study area (Fig. 1)
represents better resolution for this part as compared to the peripheral part of the study
region. Up to a depth of 10 km, the anomalies are small-sized in the region where
network stations are located. Low velocity anomalies for both P- and S-wave velocity
models are stronger and more widespread below the depth of 10 km. Tomograms
depict a slightly lower amplitude of S-wave anomalies than P-wave anomalies. A
strong and shallow high velocity anomaly traverses the region between Srinagar thrust
and MCT which fades and shifts to further northeast of MCT as we go deeper. A
large low velocity anomaly is identified in the Garhwal lesser and higher Himalayan
P anomalies S anomalies
Fig. 2 Results of tomographic inversions. Images in the first row show P-wave velocity anomalies
at depth levels of 4, 10, 14 and 20 km. Images in the second row show S-wave velocity anomalies at
depth levels of 4, 10, 14 and 20 km. Depth is mentioned at the top right corner of the tomogram with
the corresponding reference velocity mentioned at the bottom left corner. With respect to these refer-
ence velocities are the velocity anomalies computed. Colors on color scale denote the percentage
deviation of velocities from reference velocities. Locations of the recording stations are shown
as black triangles. Yellow stars indicate the location of Uttarkashi earthquake, 1991 and Chamoli
earthquake, 1999. Major fault lines are shown in black
72 R. Modi et al.
region along the strike direction of MCT from 14 to 20 km depth. Both P- and S-
velocity tomograms depict several zones of low velocity in the lesser Himalayan
region between MBT and MCT. These low velocity zones occur in the region where
recording stations are located and run across the Srinagar thrust in lesser Himalaya
which may be attributed to fluid-filled structures in the study region. A strong low
velocity zone at a depth below 10 km is depicted around Uttarkashi earthquake of
1991. Above 10 km depth, the location of Uttarkashi earthquake is at the transition
boundary of a positive and negative anomaly.
5.2 Checkerboard Test
To test the resolution potential of source-receiver pairs as described in Fig. 1, we have

performed checkerboard synthetic model tests. LOTOS algorithm permits various
kinds of synthetic models such as vertical and horizontal anomalies in the form of
polygons or periodic anomalies (checkerboard test). In the present work, we have
defined the synthetic model as a 2D checkerboard pattern of size 20 × 20 km in
the x and y directions, respectively. The pattern of anomalies does not change along
the vertical direction. The amplitude of anomalies varies as ± 5% with respect to
the reference 1D velocity model. The travel-times for synthetic tests are computed
by 3D ray tracing between the real source-receiver pairs (as shown in Fig. 1). After
computing the synthetic travel-time data, the program forgets about the source coor-
dinates and velocity model. From the station coordinates, travel-time data and a
starting 1D velocity model which is different from the true velocity model, the tomo-
graphic inversions are performed in the same way as for the real data, starting from
the preliminary source location step. Figure 3 shows the results of the checkerboard
tests for P- and S-wave velocity models after four iterations. The recovered velocity
anomalies (both P-wave and S-wave) depict that the central part of the study area
is well resolved (reasoned previously as well). Significant horizontal spreading is
noticed in the peripheral regions in the study area. The amplitude of recovered P-
wave anomalies matches the original checkerboard anomalies. In case of S-wave
anomalies, however, the amplitude is slightly reduced. As evident from Fig. 3, the
real source-receiver configuration is capable of resolving the velocity anomalies up
to a depth of 20 km for both P- and S-wave velocity models.
6 Conclusion
A 3D P- and S-wave velocity structure of Tehri, Garhwal Himalayan region is

modeled by simultaneous inversion of the local earthquake data and velocity model. A
significant reduction in the travel-time residual is obtained indicating improved earth-
quake locations after tomographic inversions. Resolution capability of the source-
receiver pairs is demonstrated using Checkerboard synthetic model testing. The
P anomalies S anomalies
Fig. 3 Results of checkerboard synthetic model tests. Images in the first row show the recovered P
wave velocity anomalies at depth levels of 4, 10, 14 and 20 km. Images in the second row show the
recovered S-wave velocity anomalies at depth levels of 4, 10, 14 and 20 km. The grid plotted over
the recovered anomalies shows the synthetic checkerboard anomaly grid where alternate boxes are
set as positive and negative velocity anomalies of amplitude ± 5%. Depth is mentioned at the top
right corner of the tomogram with the corresponding reference velocity mentioned at the bottom
left corner. With respect to these reference velocities are the velocity anomalies computed. Colors
on color scale denote the percentage deviation of velocities from reference velocities. Locations of
the recording stations are shown as black triangles. Yellow stars indicate the location of Uttarkashi
earthquake, 1991 and Chamoli earthquake, 1999. Major fault lines are shown in black
74 R. Modi et al.
velocity structure of P- and S-waves in the study area is given as a percentage devi-
ation from the reference velocity models. The tomograms of P- and S-wave velocity
anomalies indicate strong lateral heterogeneities in the study area from the surface
up to a depth of 20 km. Several zones of low velocity are delineated in the lesser
Himalayan region bounded between MBT and MCT which suggests the presence of
fluid-filled structures that may be contributing to the high seismicity of the region.
Our model depicted that the location of Uttarkashi earthquake, 1991 is at the transi-
tion boundary of a positive and negative anomaly. The positive anomaly disappears
below the depth of 10 km and a large negative anomaly surrounds this region.
References
1. Seeber, L., Armbruster, J.G.: Great detachment earthquakes along the Himalayan arc and long-
term forecasting. In: Earthquake Prediction—An International Review, Maurice Ewing Series,
AGU, vol. 4, pp. 259–277 (1981). Wiley Online Library, Washington, DC
2. Khattri, K.N.: Great earthquakes, seismicity gaps and potential for earthquake disaster along
the Himalayan plate boundary. Tectonophysics 138, 79–92 (1987)
3. Bilham, R., Gaur, V.K., Molnar, P.: Earthquakes. Himalayan seismic hazard. Science 293(5534),
1442–1444 (2001)
4. Ottemoller, L., Voss, P., Havskov, J.: SEISAN earthquake analysis software for Windows,
Solaris, Linux and MACOSX (Version 9.1), p. 368. Bergen University, Bergen (2011)
5. Kumar, A., Pandey, A.D., Sharma, M.L., Gupta, S.C., Verma, A.K., Gupta, B.K.:Processing
and preliminary interpretation of digital data obtained from digital seismic array in Garhwal
Himalaya. In: Proceedings of the 10th Symposium on Earthquake Engineering, University of
Roorkee, Roorkee, India, vol. 1, pp. 141–152 (1994)
6. Koulakov, I.: Out-of-network events can be of great importance for improving results of local
earthquake tomography. Bull. seism. Soc. Am. 99(4), 2556–2563 (2009a)
7. Koulakov, I.: LOTOS code for local earthquake tomographic inversion: benchmarks for testing
tomographic algorithms. Bull. seism. Soc. Am. 99(1), 194–214 (2009b)
8. Paige, C.C., Saunders, M.A.: LSQR: an algorithm for sparse linear equations and sparse least
squares. ACM Trans. Math. Softw. 8(1), 43–71 (1982)
9. Nolet, G.: Seismic tomography: with applications in global seismology and exploration
geophysics. Springer Science & Business Media (1987)
10. Kissling, E., Ellsworth, W.L., Eberhart-Phillips, D., Kradolfer, U.: Initial reference models in
local earthquake tomography. J. Geophys. Res. 99(B10), 19635–19646 (1994)
Correlation Between Cone Tip
Resistance and Shear Wave Velocity
for Quaternary Alluvium
P. Mishra, A. Paul, and P. Chakrabortty
Abstract Several correlations were envisaged in the literature to find out shear
wave velocity (V s ) using cone penetration test (CPT) data, but still, a considerable
discrepancy is found when CPT correlations are compared for the studied site. A
comparison study is conducted with the previously given correlations which also
shows the requirement of developing a new correlation for the site. Empirical corre-
lations have been derived in this study from measured field data at 5 sites in the
IIT Patna Campus for the estimation of shear wave velocity using CPT data. The
shear stress-related parameter, i.e., shear wave velocity is defined in terms of various
parameters obtained from the CPT. Various parameters
( ,) such as cone tip resistance
(qc ), soil behavior index (I C ), effective stress σ 0 , , and depth (z) are correlated
with V s because some factors are interrelated, like the soil layer’s aging, effective
,
confining stress, etc. Regression analysis is carried out to correlate V s with qc , I C , σ 0 ,
and z. Empirical correlations are developed for a site and validated with the rest of
the other locations in the campus area. This research will help to predict the subsur-
face material properties by reducing the overall time and cost of an investigation.
By using this correlation, important soil parameters used in design can be obtained
more accurately with only CPT data.
Keywords Shear wave velocity · CPT · Correlation · Penetration tests ·

Quaternary alluvium
P. Mishra (B) · A. Paul · P. Chakrabortty

Department of Civil and Environmental Engineering, IIT Patna, Bihta, Patna, Bihar, India
e-mail: priyam_2121ce10@iitp.ac.in
A. Paul
e-mail: abhik_1921ce04@iitp.ac.in
P. Chakrabortty
e-mail: pradipt@iitp.ac.in
https://doi.org/10.1007/978-981-99-1459-3_7
76 P. Mishra et al.
1 Introduction
Shear wave velocity (V S ) is one of the fundamental properties of the soil used
in various geotechnical fields. The V S is directly related to the small strain shear
modulus. The small strain shear modulus (G) is required in analytical procedures
for estimating the dynamic soil response and soil-structure interaction analysis [1].
In soil dynamics, V s is used to evaluate the maximum shear modulus (Gmax ), as a
soil stiffness parameter. In ground vibration problems, V s is used to calculate the
actual amount of strain during the event. The V s is used to calculate the fundamental
frequency of vibration and soil stiffness for the case of vibration under footings. For
the measurement of V s in the laboratory, resonant column or bender element tests
are used. Geophysical techniques are one of the best methods for determining shear
wave velocity in the field. The primary field tests used to measure V S are the cross-
borehole test (CHT), downhole testing (DHT), and multichannel analysis of surface
waves (MASW). All the geophysical tests are advantageous because they identify
soil properties in their natural configuration with less disturbance compared to labo-
ratory testing. It has been also observed that the degree of saturation significantly
affects the G [2] and eventually V s .
In practical problems, CHT and DHT require a borehole and it is not economically
feasible as part of routine geotechnical testing. Moreover, MASW testing does not
require any borehole and it can be performed quickly and economically [3]. But
obtaining data from this test is quite complex because of the dispersive nature of
the waves radiated. Another method of seismic cone penetration test (SCPT) gives
excellent V S profiling, but it needed specialized equipment and sound technical
expertise to collect data during the test. A velocity seismometer is equipped with
cone penetration test (CPT) equipment to obtain the measurement of dynamic shear
modulus. CPT is widely used to obtain vertical profiling and soil classification. The
advantage of using CPT is that it does not require any sample collection as needed
in SPT, moreover gives continuous data logging and is repeatable. The addition of
a seismic velocity sensor that gives direct V s with CPT is advantageous. Separate
testing for V s is not only time-consuming and expensive, but it also depends upon
the importance of the project. The development of a multi-linear regression equation
[4] between V S and cone penetration testing parameters is beneficial as it gives an
estimate of V S without actual measurements.
Several CPT-V S correlations have been developed for different soil types. Mayne
and Rix [5] uses 31 different natural clays for correlating the cone tip resistance (qc )
and V S also improved the correlation by using the void ratio. Mayne and Rix [5]
proposed correlations for specific soil types like sand, and clays using qc and sleeve
friction (f s ). Hegazy and Mayne [6] reviewed the previous correlation of V s with qc
and f s and found that V S majorly depends on qc and the soil type. Robertson and
Wride [7] presented the soil behavior Index (I C ), which ( , )tells about the soil type and
is the function of qc , f s , and vertical effective stress σ0 .
Information regarding soil types and their behavior is best given by I C , so V S is
correlated with qc and I C . Andrus et al. [8] have considered cone tip resistances,
Correlation Between Cone Tip Resistance and Shear Wave Velocity … 77
sleeve friction, depth, overburden pressure, soil behavior type index, and geologic
age at the time of generating the correlation between CPT-V S . Here different geolog-
ical aging parameter and increasing or decreasing geological age plays a significant
role in increasing or decreasing qc . Robertson [9] performed the SCPT and gener-
ated the correlation between the net cone tip resistance, V s , and shear modulus by
considering the normalized cone tip resistance (Qtn ) and I C value. McGann et al. [10]
reviewed different available CPT-V S relations and calculated residual value from the
difference between predicted and measured V S . Combination of different parameters
from available models which are giving a minimum value of residual is used to find
the functional form of correlation equation. Mola-Abasi et al. [11] have proposed
a polynomial model with the conclusion that the cone tip is more crucial than the
sleeve friction in their predicted model. More recently Mousa and Hussein [12]
developed seven different correlation equations with different pairs of CPT param-
eters. Numerous researchers have attempted to correlate the CPT parameters with
V S . Although researchers like Andrus et al. [8] have shown and incorporated some
prime factors which can influence the change of V s profile. But still, some factors
like geological factors, soil textural factors, etc. are left, which may be the cause of
variation. Application of correlation to a site, which has a similar type of geological
conditions may give reliable output. If those correlations do not give appropriate
results, then the cause associated with it in terms of a local factor needs to be intro-
duced in the existing correlation. Roy et al. [13] have considered different available
V s –N correlations and tried to address the effect of uncertainty while selecting any
V s –N correlation, especially in terms of site response analysis. Although these uncer-
tainties need to be considered at the selection stage of any correlation, in this article
factors considered which are solely responsible for causing the change in V s profile
using cone penetration are studied. Emphasis is given to CPT-related parameters that
should be present in correlation to predict V s . Mayne and Rix [5] showed that if any in-
situ void ratio (e0 ) can be incorporated, then correlation gets significantly improved.
Traditionally, V S can be measured by the MASW test which required a huge open
space area. MASW is a great geophysical testing tool for the estimation of V s . The
V s has been estimated in three steps, i.e., (1) Data acquisition (2) Data processing
by dispersion curve generation (3) Inversion and V s profile creation. In this article,
shear wave velocity has been correlated with different parameters estimated using
CPT. Jakka et al. [14] have discussed various uncertainties associated with MASW
testing, such as (i) Model-based uncertainty (ii) Data measurement uncertainty (iii)
Inversion uncertainty and proposed some guidelines to minimize that uncertainty.
MASW testing has been performed, and results have been analyzed using proper
guidelines like the one suggested by Jakka et al. [14] so that the minimum uncer-
tainty can be induced. On the other side, CPT is a widely used test for assessing soil
properties, subsurface stratigraphy associated with soil materials, and liquefaction
potential. Different researchers have given correlations with CPT also, but it has been
realized (shown in the next section) that the available correlations are not applicable
to the studied soil. So, these correlation needs to be revised, and more significant
factors need to be considered in correlations. Therefore, for all the above reasons, an
attempt had been made to propose new correlations to compute V s from CPT data.
78 P. Mishra et al.
Fig. 1 Location of CPT and MASW tests at IIT Patna campus used in this study
In this article, all the significant CPT-V S correlations available in the literature are
investigated and plotted using the CPT data from the IIT Patna campus. The location
of CPT testing results used from the IIT Patna campus in this study is shown in
Fig. 1. Plot using a different available model of V S does not show good agreement
with experimentally measured shear wave velocity using MASW tests. Therefore,
a site-specific equation has been developed in this study for predicting the Vs value
from CPT test data.
2 Some Existing CPT-V S Correlations
In the first part of this study, different existing CPT-V S correlations (listed in Table 1)
with their number of sample datasets along with the coefficient of determination (R2 )
are reviewed. Using these empirical equations, V S has been calculated and plotted
along with the depth up to 30 m in Fig. 2. These plots of V S are varying with each
other, with a huge difference in between them. Measured shear wave velocity (V SM )
is also plotted in a continuous line. The difference between actual and predicted V S
is noticeable. Some of the models give huge differences while some show little gaps
at varying depths. This lag between plots of predicted V S and measured V S tells us
that none of the existing models can predict the V S correctly for this location. It also
shows that existing correlations are site-specific and vary from region to region.
Table 1 List of some well-known existing CPT-V s correlations

Study Geologic No of R2 Vs
age samples
( / )0.3
Hegazy and Quaternary 323 0.695 (10.1 log(qc ) − 11.4)1.67 . f s qc ∗ 100
Mayne [15]
Piratheepan Holocene 60 0.73 32.3qc0.089 f s0.121 D 0.215
(2002)
( ,
)0.25
σ0
Hegazy and – 558 0.854 0.0831 ∗ qc1n ∗ Pa ∗ e1.786IC
Mayne [6]
Mayne [16] Quaternary 161 0.82 118.8 log( f s ) + 18.5
Andrus et al. Holocene 185 H–0.73 2.62qt0.395 Ic0.912 D 0.124 S F
[8] (H) & P–0.43
Pleistocene
(P) age
Robertson Quaternary 1035 – 10(0.55I c+1.68) (qt − σv )0.25
[9]
McGann – 513 – 18.4qc0.144 f s0.0832 z 0.278
et al. [10]
100(1.40 + 1.59 f s + 0.09qc
Mola-Abasi – 37 – )
et al. [11] −1.33 f s2 − 0.002qc2 + 0.05 f s qc
Zhang and – – 0.798 1915qt0.317 Ic0.210 D 0.057 S F a

Tong [17]
Vs (m/s)
0 200 400 600 800
0
10
Depth (m)
15
20
25
30
Andrus et al (2007) Hegazy and Mayne (2006)

Hegazy and Mayne (1995) Molabassi et al
Mayne (2006) Piratheepan (2002)
McGann et al (2015) Robertson (2009)
Zhang et al Vsm
Fig. 2 Variation of shear wave velocity along depth estimated using different available CPT-V s
models and experimentally measured for the studied site
80 P. Mishra et al.
3 Methodology
CPT and MASW tests have been conducted at different locations on the campus.
Readings of qc and f s are taken at an interval of 20 cm each. So, CPT values are
averaged up to 1 m. While V s values obtained from the MASW survey are obtained
for 1 m intervals. CPT testing has been done up to a depth of 30 m while V s values
obtained from the dispersion curve in MASW testing are also up to a depth of greater
than 30 m. So, correlation developed using these values can be used up to 30 m depth.
CPT testing gives qc and f s as output as shown in Fig. 3. Soil behavior Index (I C )
can be calculated by the following expressions given in the literature (Robertson and
Wride [7]) using qc and f s :
{ }0.5
IC = (3.47 − log Q tn )2 + (log Fr + 1.22)2 (1)
( )
(qt − σv ) pa n
Q tn = , (2)
pa σ0
fs
Fr = × 100(%) (3)
(qt − σ0 )
( ,/ )
n = 0.381Ic + 0.05 σv pa − 0.15 ≤ 1.0 (4)
where Qtn = Normalized cone tip resistance; F r = Friction ratio; and Pa =

atmospheric pressure (100 kPa)
Cone tip resistance qc (kPa) Sleeve friction fs (kPa)

0 5 10
0 200 400
0
0 CPT-1
CPT-1
CPT-3
CPT-3 5
5 CPT-8
CPT-8
CPT-12
CPT-12 CPT-15
10 CPT-15 10
Depth (m)
Depth (m)
15 15
20 20
25 25
30 30
Fig. 3 Experimentally determined CPT results a Cone tip resistance (qc ) and b Sleeve friction (f s )
values at different locations
Table 2 Soil classification

Soil behavior type index, Zone Soil behavior type
according to the soil behavior
IC
type index (after Robertson
[18]) IC < 1.31 7 Gravelly sand to dense
sand
1.31 < I C < 2.05 6 Sands: clean sand to silty
sand
2.05 < I C < 2.60 5 Sand mixtures: silty sand
to sandy silt
2.60 < I C < 2.95 4 Silt mixtures: clayey silt to
silty clay
2.95 < I C < 3.60 3 Clays: silty clay to clay
I C > 3.60 2 Organic soils: peats
In the above equations, qt is the normalized value of the cone tip for pore water
pressure acting just behind the tip of the cone with a value of area ratio and are related
as qt = qc + u2 (1 − a), where qc = cone tip resistance; u2 = dynamic pore pressure,
and a is the area ratio. Generally, the value of qc and qt are nearly equal except for
some soils with greater fineness content.
Using all the above parameters, the first normalized cone tip resistance (Qtn ) is
calculated with an initial value of n as 1 for the first iteration. Then final values I C and
n are calculated by iterations. Based on Qtn and friction ratio (F r ), Robertson [18]
proposed a soil behavior chart in which 9 different zone of soil types (as mentioned
in Table 2) has been described. Using the chart and values of CPT parameters (i.e.,
qc and f s ) soil type has been identified and shown in Fig. 4 and vertical soil profiling
is done using I C . Here, I C is used because the soil sample is not taken in CPT testing,
so information about soil type and its behavior can be obtained from the I C index
(shown in Fig. 5).
In Fig. 4, it can be seen that all the soils in the studied region lie from Zone–3 to
Zone–6, which are classified as Clay to Sand. The majority of the soil found on the
campus using the above chart is Sandy soil in nature Nilay et al. [19]. Soil variability
with depth is also shown in Fig. 5. From the I C plot, it can be seen that at most CPT
locations in IIT Patna soil is a sandy to silty mixture up to a depth of 30 m. While a
layer of clayey soil is also observed between a depths of 18–24 m.
4.1 Regression Analysis
The equation for predicting V S is obtained using non-linear regression with multiple
variations. The concept behind using a non-linear model is based on the evaluation
of previously available models like Mayne and Rix [5], and Andrus et al. [8]. From
82 P. Mishra et al.
1000.00
CPT 1
CPT 3
CPT 8
100.00 CPT 12
CPT 15
Qtn
10.00
1.00
0.100 Fr (%) 1.000 10.000
Fig. 4 Classification of soil using normalized cone tip resistance (Qtn ) and friction ratio (F r ) based
on Robertson [18]
Ic Value
1.50 2.50 3.50
0
CPT - 1
2 CPT - 3
CPT - 8
4
CPT - 12
Clayey silt to silty clay
6 CPT - 15
silty sand to sandy silt
8
10
Sand mixutre
Silt Mixture
12
Depth (m)
Sand
Clay
14
16
18
20
22
24
26
28
30
Fig. 5 Variation of soil behavior index (I C ) along the depth

previous studies, it is found that qc is the most important parameter related to the
undisturbed shear strength of the soil. Hegazy and Mayne [6] showed that cone tip
resistance data represent better variability in V s than in f s . As discussed, and shown
in the previous section, I C is an indicator of soil type, so our shear wave velocity (V s )
( , ) type (I C ). Other parameters in this correlation
will depend upon this soil behavior
equation are effective stress σ 0 and depth (z). These parameters are those over
which shear wave velocity (V s ) and depth are highly correlated. If we plot shear
wave velocity against depth, it will be seen that with ( , ) an increase in depth, V s also
increases, the same applied with effective stress σ 0 ; it also increases with depth.
So, in CPT-V S correlation consideration of these two parameters will be useful.
For this correlation equation, data collected from campus are shown in Fig. 6a.
This pair of cone penetration and MASW testing data is used for regression analysis.
It is listed as C15-M13, where C15 and M13 are the location of cone penetration and
MASW tests, respectively. Using non-linear multiple variable statistical regression
analysis, the following equation has been developed with an R2 value of 0.90.
Vs = 828.693qc−0.013 Ic−0.203 σ0,−0.534 z 0.712 (5)

,
where V s is in m/s; qc is in kPa; σ 0 is in kPa, and z is depth in meters.
a) C15-M13 b) C1-M3 c) C3-M1

Shear wave (m/s) Shear wave (m/s) Shear wave (m/s)
0 200 400 0 500 0 200 400
0 0 0
Measured Vs Measured Vs
Predicted Vs Predicted Vs
5 5 5
10 10 10
Depth (m)
Depth (m)
Depth (m)
15 15 15
20 20 20
25 25 25
Measured Vs
Predicted Vs
30 30 30
Fig. 6 Plots of shear wave velocity (V S ) for different CPT sites using the regression equation to
validate the model. a Site near C15 b Site near C1 c Site near C3
84 P. Mishra et al.
4.2 Validation of Results
To validate the above non-linear CPT-V S equation, CPT values from other locations
in the site are plugged into the above equation. All other selected CPT locations and
their predicted value of V S show a good agreement. The plots between predicted and
measured values for other CPT locations are shown in Fig. 6. Using the regression
Eq. (5) given in this study, predicted shear wave velocity (V SP ) is shown in the
same figure. To check the efficiency of the proposed equation, the term coefficient
of determination R2 is used. The value of R2 of two locations, namely C1-M3 and
C3-M1 [shown in Fig. 6b and c], have been found as 0.90 and 0.91, respectively.
Both CPT locations with their predicted and measured values of V s are shown in
Fig. 7. In this figure, both the results are plotted over a 45° line. The closer the
results with this line, the better the accuracy of predicted V S . Deviation of V s value
from predicted and measured, has been also presented by a term called velocity ratio
denoted by ‘K’, i.e., the ratio of measured to the predicted value. The value of K is
plotted against depth and shown in Fig. 8. Scatter plot which is closest to K = 1,
represents that predicted and measured V S are the same. CPT site location C3 shows
a good relationship as the K value is nearly close to 1 up to 20 m depth, while C1
initially shows predicted value is less than measured up to a depth of 5 m, then it
crosses and started over predicting the V S value. Predicted values are in the range of
± 30% approximately.
Fig. 7 Variation between

measured and predicted V S 290
for different sites
270
250
Predicted VSP (m/s)
230
210
190
C1
170
C3
150
150 170 190 210 230 250 270 290
Measured VSM (m/s)
Fig. 8 Velocity ratio (K) Velocity ratio (K)

compared along with the 0.5 1 1.5
depth 0
C1
5
10
Depth (m)
15
20
25
30
5 Conclusions
In the first part of this study, soil classification is performed using I c values along the
depth. The available CPT-V s correlations are used to predict the value of shear wave
velocity. Existing empirical correlations may be region-specific or applicable to a
particular soil type, but none of the relations gives reliable values for the site studied
here. All models give different results, and there is a considerable difference (up to
280%) between the measured and predicted values for the selected site. So, for this
reason, a non-linear multiple variable regression equation is developed (Eq. (5)). This
equation is further validated with data obtained from other sites inside the campus.
Validation results match the measured value of V S at a different location on the
campus with sufficient accuracy. The proposed regression equation is not validated
for the site other than the studied region. The proposed CPT-V S equations can be used
for other similar sites with caution. Before using these correlation, site conditions,
geologic similarity, etc. needs to be considered. If the direct measurement of V S is
possible, then preference should be given to all those methods. Therefore, further
study is needed to check the applicability of the proposed model globally.
86 P. Mishra et al.
References
1. Youd, T.L., Idriss, I.M.: Liquefaction resistance of soils: summary report from the 1996 NCEER
and 1998 NCEER/NSF workshops on evaluation of liquefaction resistance of soils. J. Geotech.
Geoenvironmental Eng. 127(4), 297–313 (2001)
2. Chakrabortty, P., Roshan, A.R., Das, A.: Evaluation of dynamic properties of partially saturated
sands using cyclic triaxial tests. Indian Geotech. J. 50(6), 948–962 (2020)
3. Chakrabortty, P., Kumar, U., Puri, V.: Seismic site classification and liquefaction hazard
assessment of Jaipur city, India. Indian Geotech. J. 48(4), 768–779 (2018)
4. Das, A., Chakrabortty, P.: Simple models for predicting cyclic behaviour of sand in quaternary
alluvium. Arab. J. Geosci. 15, 385 (2022)
5. Mayne, P.W., Rix, G.J.: Correlations between shear wave velocity and cone tip resistance in
natural clays. Soils Found. 35(2), 107–110 (1995)
6. Hegazy, Y.A., Mayne, P.W.: A global statistical correlation between shear wave velocity and
cone penetration data. In: Site and Geomaterial Characterization (American Society of Civil
Engineers (ASCE), Geotechnical Special Publication (GSP) 149), Proc GeoShanghai, pp. 243–
248 (2006)
7. Robertson, P.K., Wride, C.E.: Evaluating cyclic liquefaction potential using the cone penetra-
tion test. Can. Geotech. J. 35(3), 442–459 (1998)
8. Andrus, R.D., Mohanan, N.P., Piratheepan, P., Ellis, B.S., Holzer, T.L.: Predicting shear-wave
velocity from cone penetration resistance. In: Proceedings of the 4th International Conference
on Earthquake Geotechnical Engineering, Thessaloniki, Greece, paper no. 1454, pp. 1–12
(2007)
9. Robertson, P.K.: Interpretation of cone penetration tests—a unified approach. Can. Geotech. J.
46(11), 1337–1355 (2009)
10. McGann, C.R., Bradley, B.A., Taylor, M.L., Wotherspoon, L.M., Cubrinovski, M.: Develop-
ment of an empirical correlation for predicting shear wave velocity of Christchurch soils from
cone penetration test data. Soil Dyn. Earthq. Eng. 75(5), 66–75 (2015)
11. Mola-Abasi, H., Dikmen, U., Shooshpasha, I.: Prediction of shear-wave velocity from CPT
data at Eskisehir (Turkey), using a polynomial model. Near Surf. Geophys. 13(2), 155–168
(2015)
12. Mousa, A., Hussein, M.: Prediction of shear wave velocity in fine-grained soils from cone
penetration test data: toward a global approach. Transp. Res. Rec. J. Transp. Res. Board 2676(6),
565–582 (2022)
13. Roy, N., Shiuly, A., Sahu, R.B., Jakka, R.S.: Effect of uncertainty in V S -N correlations on
seismic site response analysis. J. Earth Syst. Sci. 127(7), 1–21 (2018)
14. Jakka, R.S., Desai, A., Foti, S.: Guidelines for minimization of uncertainties and estimation
of a reliable shear wave velocity profile using MASW testing: a state-of-the-art review. In:
Advances in Earthquake Geotechnics, pp. 211–253 (2022)
15. Hegazy, Y.A., Mayne, P.W.: Statistical correlations between V S and CPT data for different soil
types. In: Proceedings of the 1st international symposium on cone penetration testing (CPT’95),
vol. 2, pp. 173–178 (1995)
16. Mayne, P.W.: In-situ test calibrations for evaluating soil parameters. Characterisation Eng.
Prop. Nat. Soils 3(4), 1601–1652 (2007)
17. Zhang, M., Tong, L.: New statistical and graphical assessment of CPT-based empirical
correlations for the shear wave velocity of soils. Eng. Geol. 226, 184–191 (2017)
18. Robertson, P.K.: Soil classification using the cone penetration test. Can. Geotech. J. 27(1),
151–158 (1990)
19. Nilay, N. Chakrabortty, P., Popescu, R.: Liquefaction hazard mapping using various types of
field test data. Indian Geotech. J. 52, 280–300 (2022)
Ground Motion Predictive Equations
and Its Applicability in North-Eastern
Indian Region: A Critical Appraisal
P. Kumar and S. S. Kumar
Abstract The level of ground shaking, to which earthquake-resistant structures are

subjected, is estimated using Ground Motion Prediction Equation (GMPE) in terms
of ground motion parameters. More than 1000 GMPEs worldwide and more than
50 GMPEs for India are available to estimate ground motion parameters like PGA,
PGV, PGD, spectral acceleration, etc. However, the selection of a GMPE for seismic
hazard analysis, out of available ground motion models becomes a critical task for
seismic hazard analysis. In the present study, the GMPEs were selected for the
Northeast region of India according to qualitative analysis and then ranked to those
GMPEs for their suitability using the quantitative information-theoretic approach,
based on the estimation of the log-likelihood value. Further, the ranking of models
for discrete ranges of distance was estimated considering respective weights for the
Northeast region. Based on the variable performance of models, for the different range
of data, it has been observed that (NDMA in Govt. of India New Delhi, 2010) and
(Bajaj and Anbazhagan in Soil Dyn. Earthq. Eng. 126, 2019) are the best-suited model
for the study area. However, the weighted ground motion models have shown the
efficiency and robustness in the process of selecting and ranking GMPEs for diverse
applications, particularly for the Northeast Indian region, which has witnessed some
devastating earthquakes in past and is vulnerable as per seismic hazard analysis.
Keywords Ground motion · GMPE · PGA · LLH
1 Introduction
Northeast India is considered as one of the most seismically active regions in the world
due to frequent occurrence of the earthquakes of moderate to large magnitude. The
continuous subsidence of Indian plate under the Eurasian plate and the Indo-Burma
subduction zone is the major region to trigger such events. The devastating effect of
P. Kumar (B) · S. S. Kumar

Department of Civil Engineering, NIT Patna, Bihar 800005, India
e-mail: prabhakark.phd20.ce@nitp.ac.in
https://doi.org/10.1007/978-981-99-1459-3_8
88 P. Kumar and S. S. Kumar
these earthquakes can be minimized by identifying the seismic risk and the strong
ground motion, induced by the earthquakes, prior to the design and construction
of the structures. The strong ground motion parameters can be evaluated using a
ground motion prediction equation (GMPE) for a respective area and seismic risk
can be assessed by conducting the region-specific seismic hazard analysis. Also, the
assessment of the conditional chance that, in the event of the earthquake, a specific
acceleration limit can be surpassed at the site of interest, requires a GMPE.
GMPEs are widely used relationships for calculating the seismic ground motion’s
amplitude parameters such as the response spectrum, peak ground acceleration, and
peak ground velocity, take into consideration the impacts of the source, path, and loca-
tion. For the estimation of ground motion parameters, incorporating an appropriate
attenuation model, at a specific site for each earthquake scenario is the fundamental
part of Probabilistic Seismic Hazard Analysis (PSHA). Seismic hazard study for any
region must start with an appropriate GMPE to calculate the ground motion in terms
of peak ground acceleration (PGA) or spectral acceleration (SA). The availability of
number of GMPE pertaining to a single region requires the selection of an appropriate
one, mere random selection may carry forward practical problems associated with
the model [1]. There are various methods available for the selection of GMPE which
can be brought under the category of qualitative and quantitative approaches. Cotton
et al. [2] offered a list of qualitative suggestions for disliking ground motion equa-
tions whereas Scherbaum et al. [1] have used exceeding probability for the selection
and ranking of the GMPEs. Further, Scherbaum et al. [3] have also suggested an
information-theoretic strategy for the ranking of GMPEs based on the consideration
of log-likelihood (LLH) for the refinement of earlier work.
Early-stage GMPEs have demonstrated a reduction in epistemic uncertainty with
an increase in model complexity, such as source and site properties. However, the
recent research revealed that the uncertainty associated with the earthquake, source,
path, and location, necessitates the use of numerous GMPEs for seismic hazard
assessments using logic tree architecture following their selection and ranking. The
purpose of this study is to find appropriate GMPEs for various regions of Northeast
India and to rank and weight the same for seismic hazard analyses by taking into the
account of data of region-specific earthquakes and GMPEs that are already accessible.
Eight of the regionally applicable GMPEs are considered in the proposed work
(Table 1). GMPEs considered in the study vary from each other in the context of
applicable distance range. Hence, the analysis is bifurcated into three distance bins
(I) up to 100 km, (II) 100–300 km, and (III) 300–500 km. All the applicable GMPEs
are used to estimate the PGA value for each distance bin. Further, these PGA values
were utilized to calculate the rank and weights for GMPEs based on LLH values and
Data Support Index (DSI), taking account of all the earthquakes considered in this
study. In an effort to avoid overestimating and underestimating the PGA at shorter
and larger distances, a segmented-based ranking of GMPEs is explored.
Ground Motion Predictive Equations and Its Applicability … 89
Table 1 Abbreviations for

S.N Attenuation relationship Abbreviation for equation
GMPEs
1 Nath et al. (2009) NA09 [4]
2 NDMA (2010) ND10 [5]
3 Nath et al. (2012) NA12 [6]
4 Anbazhagan et al. (2013) AN13 [7]
5 Raghukanth and Kavitha RK14 [8]
(2014)
6 Singh et al. (2016) SI16 [9]
7 Kumar et al. (2017) KU17 [10]
8 Bajaj and Anbazhagan BA19 [11]
(2019)
2 Seismotectonics of the Study Area
India’s Northeastern region and its environs can be classified into six tectonic blocks
[12]. The Arakan-Yoma belt region is part of the Indo-Burma subduction zone and
other zones namely Eastern Himalaya, Surma Valley, Shillong Plateau, Assam Valley,
and Naga Hill region, which are not in the subduction zone, correspond to the active
region. Due to the fact that earthquakes are occurring in the crust, intraslab, and
interface subduction zones, the NEI is regarded as a seismically unique region in the
world.
The thrust planes, Main Central Thrust (MCT), Main Boundary Thrust (MBT),
Main Frontal Thrust (MFT), and their subsidiary thrusts, make up the Himalayan
structures in the Northeastern region. A notable tectonic complexity is formed in
the Mishmi region by the Mishmi thrust, Lohit thrust, Po Chu fault, and Tidding
suture. The Tsangpo suture, Tuting, Kopili, Oldham, Dubri, Dauki, and Bame faults
are further significant deformities [14]. A no. of thrusts such as Disang thrust, Lohit
thrust, Naga thrust, and eastern boundary thrust is also present in the zone, making
Northeast one of the most seismically active regions in the world that demands
a regional seismic hazard analysis. A reliable hazard analysis requires selecting a
GMPE which predicts the amplitude parameter closely to actual values. Various
tectonic regions along with the faults and recorded events used in the study are
depicted in Fig. 1.
3 GMPEs Used in the Study
Eight suitable GMPE models were selected to analyze their relative closeness to
actual PGA values. NA09 used the Finite fault stochastic model (EXSIM) of Motaze-
dian and Atkinson [15] which models the spectral amplitude based on the Convolution
theorem [16]. NA09 utilized relations of Motazedian and Atkinson [15] for source
Fig. 1 Seismotectonic map of Northeastern India along with Locations of earthquakes used in the
study (Modified after Rahman [13])
modeling and adopted the GMPE form proposed by Campbell and Bozorgnia [17]
and reported a similar trend of Amplitude-site distance plot for distances < 30 km.
Nath’s relation applies to northeast India over a wide range of Magnitudes. GMPE
developed by ND10 is applicable for the whole of India and, it was developed consid-
ering India has various seismotectonic zones as per the works of Sabeer et al. [18]
and determining potential maximum magnitude for each of 32 sources assigned to
tectonic zones using prehistoric data by Kijko and Graham [19] method. NDMA’s
GMPE relationship was derived from 80,000 samples originating from 38,860 events
which makes it applicable for a large distance and Magnitude range (Table 2).
The concept of Apparent Station was first utilized in the region by AN13 to get an
even plot of ground motion and corresponding magnitude for the development of a
GMPE relation based on FINSIM model of Beresnev and Atkinson [20]. Anbazhagan
et al. [7] model was derived from 420 ground motions from 14 EQ recorded at 30
stations placed at an interval of 10 km, and includes the past seismicity data obtained
from a correlation between MMI and surface PGA values. NA12 model has been
developed for the Shillong region considering the earthquake to be originated from
two separate crust zones based on the depth criteria. This model is based on the works
Table 2 Functional form and applicability range of GMPEs used in the study
S. GMPE Functional form of GMPE Magnitude Distance
No. range range
loge y = 9.143 + 0.247MW

1 NA09 −0.014(10 − MW )3 4.8–8.1 Upto
100 km
−2.697 loge Rr up + 32.9458 exp(0.0663 ∗ MW )
y = PGA (g)
loge y = c1 + c2 M + c3 MW2
+ C4 Rhypo

2 ND10 +c5 ln Rhypo + c6 exp(c7 MW ) 4.0–8.5 Upto
500 km
+c8 log Rhypo f o + ln ∈
y = PGA (g);
c1 to c8 are the regression coefficients
loge y = c1 + c2 MW + c3 (10 − MW )3
3 NA12 4.8–8.1 Upto
+c4 loge Rr up + c5 exp(c6 MW ) 100 km
y = PGA (g);
log10 y = c1 + c2 MW
4 AN13 √ 3.0–8.7 Upto
−1.792 log10 R 2 + h 2 + exp(c3 MW ) 300 km
y = PGA (g); c1 to c3 are the regression coefficients;
R is the closest distance to rupture;
h is the focal depth
loge y = c1 + c2 M + c3 MW2
+ C4 Rhypo

5 RK14 +c5 ln Rhypo + c6 exp(c7 MW ) 4.0–8.5 Upto
1000 km
+c8 log Rhypo f o + ln ∈
/
f o = max loge (R 100 , 0)
loge y = c1 + c2 (MW − 6) + c3 (MW − 6)2
6 SI16 4.0–8.5 Upto
− loge Rhypo − c4 Rhypo 300 km
y = PGA (g);
log10 y = −1.497 + 0.3882MW
7 KU17 4.0–6.8 Upto
−1.19 log10 Rhypo + exp(0.2876MW ) 200 km
y = PGA (g)
ln y = c1 + c2 (M − 6) + c3 (9 − M)2 + c4 ln R
8 BA19 4.0–9.0 Upto
+am ln R(M − 6) + c7 R + σ 750 km
y = PGA (g); cm = c5 if M W < 6.0, otherwise c6 ;
of Boore [21] which has been developed for the lower crust zone (depth < 25 km)
and upper crust zones.
SI16 utilized the stochastic point source model for the development of regional
GMPE after utilizing the 30,000 records obtained from 1000 simulated earth-
quakes. The equation presented is shaped in the functional form used by Iyenger
and Raghukant [22]. KU17 utilized 216 records originating from 24 earthquakes to
obtain an attenuation relationship for Northeast India. The records obtained from the
Indian strong motion network were then regressed using a 2-step stratified regression
approach [23]. BA19 developed an attenuation relationship for the whole Himalayan
region after dividing it into 4 different regions, i.e., Kashmir Himalaya, Kumaon
Himalaya, Bihar-Nepal Himalaya, and Northeastern Himalaya. A separate duration
model, geometric attenuation and kappa factor were also developed by BA19 for the
respective region which was later used for simulation. This GMPE uses a functional
form which is developed particularly for the Himalayan region.
All the above GMPEs differ in terms of the parameters and methodology used for
the development of the equation which creates confusion among the seismic hazard
analysts during the selection of a GMPE. The functional form and the applicability
range of each GMPE in terms of magnitude and distance are reported in Table 3.
Thus, an effort is made in the present study which will allow an easy selection of a
model for hazard analysis.
4 Methodology
The development of GMPEs, which are used to forecast the amplitude characteristics
like peak ground acceleration (PGA), velocity, or displacement, for an earthquake has
been widespread in recent years [24]. This is because of GMPEs have a substantial
impact on the modeling of seismic risk and hazard: deriving more precise predic-
tions from Ground Motion Models (GMMs) leads to more precise estimations of
seismic risk and hazard. To strengthen the selection of GMMs, various statistical and
probabilistic methods have been devised. Several tests, like the Pearson correlation
coefficient and Chi-Square Misfit, employ direct observations for model selection,
while others, like the chi-square and Kolmogorov–Smirnov tests, analyze the form
and distribution of model mismatch residuals. The likelihood (LH) value test [1]
employs the assumption of log-normal distribution for each GMPE and calculates
the probabilities of residuals; the log-likelihood (LLH) [3] method uses information
theory to assess how feasible the model is for the provided data.
LLH method is based on the likelihood value of the observed ground motion that
will occur if a model is believed to be valid. The match between the data and model
is indicated by one value, the negative average log-likelihood (LLH) [25]. Figure 2
shows the working approach of LLH method in which model R represents the actual
GMPE and A, B, and C are some other models which try to predict the ground motion
as close as possible. However, it should be noted that actual model R is also unknown
and only some information is provided which is based on the discrete observations
Table 3 Chronicle of earthquakes used in the study
Earthquake ID Date Time Magnitude Magnitude scale Latitude Longitude Depth No. of stations
(DD-MM-YYYY) (UTC) (M W ) (°N) (°E) (km) (Rock)
NE India EQ01 10-09-1986 07:50:26 4.5 MS 25.385 92.077 43 7
India-Bangladesh EQ02 06-02-1988 14:50:45 5.8 MS 24.688 91.57 15 8
border
India-Burma border EQ03 06-08-1988 00:36:25 7.3 MW 25.149 95.127 90 12
India-Burma border EQ04 08-05-1997 02:53:15 6.0 MB 24.894 92.25 34 6
Ground Motion Predictive Equations and Its Applicability …
Bhutan EQ05 21-09-2009 08:53:04 6.2 MW 27.3 91.5 8 3

93
Fig. 2 Working model of LLH-based ranking approach
corresponding to x 1 , x 2 and x 3 . The respective probability density function for each

of the predictive models is shown in Fig. 2. Based on this data LLH leads us to a
conclusion in regards to the best suitable GMPE model (A, B, C). LLH value is given
by Eq. (1).
1 Σ
N
LLH = − log2 (P(xi )) (1)
N i=1
where P(x) is the probability density function (PDF) of the GMPE models, x i
represents the observations from the antecedent earthquakes upto a total no. of N
datasets.
The total distance has been divided into three distance bins of up to 100, 100–300,
and 300–500 km to demonstrate the performance of the GMPEs at various distance
ranges. For each of the three distance segments, the average LLH values for each
of the 8 GMPEs have been determined wherein, the high LLH value correlates to
a model which is less likely to have produced the data and the small LLH suggests
that the respective model is near to the model that has generated the data. Further,
LLH-based weights for each of the GMPE model is calculated (Eq. (2)) utilizing
the respective LLH value according to the method proposed by Delavaud et al. [26]
which represents the extent to which data enhances or reduces a model’s weight in
correlation to its level of non-informativeness.
2−L L H (Pi (x))

Wi = Σ N (2)
−L L H (Pi (x))
i=1 2
Wi − Wu
DS Ii = × 100 (3)
Wu
where W i is the initial weight, W u (= 1/M) is uniform weight, and M is the number
of GMPE model used for calculating LLH. The Data Support Index (DSI) (Eq. (3))
value which represents the percentage by which the weight of a model is reduced or
increased, is used for the calculation of the final weight. The final weight is calculated
only for GMPEs having a positive DSI value.
Ranking of GMPEs not only helps to select the suitable GMPEs but also leads to
a reduction in epistemic uncertainty associated during selection [27]. In the present
study, 5 earthquakes from the northeast Indian region were selected to act as the
discrete observations of the actual model. Further for each earthquake PGA values
were calculated using the respective equation. The records of each event are collected
for stations at NEHRP site class A from COSMOS virtual data center [28]. After
that ranking is determined using LLH-based approach for each distance bin.
Ranking based on LLH value is presented in Fig. 3a, b, c. The respective weight
of the selected models is given in Table 4. When a GMPE is not valid for a distance
bin it was not considered for the ranking and GMPEs having negative DSI value
was also not considered for the weight which can be used for multi-model hazard
estimation. For distance bin I, the ND10 model was ranked 1 for EQ01 and EQ04,
and BA19 is ranked 2 for two of the earthquakes. For bin II and III ND10 model and
BA19 model are ranked 1 and 2, respectively, for all the earthquakes. KU17 acquired
the third rank for bin II for EQ02 and EQ04 as it was not considered for EQ03 and
EQ05 because the recording station does not fall in the applicability range of the
model. Since a low number of data points were employed to determine the ranking,
an increase in uncertainty is unavoidable which further increased the variance and
also leads to divergence of ranking order. The effect of this can be seen as few of
the models developed for Sikkim Himalaya were ranked higher, however, it can be
reduced by increasing the data points [25]. Thus, the results obtained in the study can
modify given the increment of no. of records used in the analysis. The weights of the
model given in Table 4 are significantly different from one another which represent
the importance of the use of the weighted model for future seismic hazard analysis.
The overall suitability of the GMPE models based on the results obtained (Table 4)
is given in Table 5 which classifies the GMPEs as best fit, moderately fit, and poorly
fit models. ND10 and BA19 models are the best fit model throughout the distance
range. At a higher distance range, the ND10 model was the best-fitted model and its
weight is also high when compared with the next best fit model AB19 which is also
highlighted in Fig. 3.
(a)
8
7
6
5
4
Rank
3
2
EQ04
1 EQ02
EQ01
0
NA09 ND10 NA12 AN13 RK14 SI16 KU17 BA19
(b)
6
4
Rank
2
EQ05
1 EQ04
EQ03
0 EQ02
AN13 RK14 SI16 KU17 BA19 ND10
(c)
3
Rank
1
EQ05
EQ03
0
ND10 RK14 BA19
Fig. 3 Ranking of each GMPE for distance bin I (up to 100 km) (a), distance bin II (100–300 km)
(b) and distance bin III (300–500 km) (c) considering all the earthquakes mentioned in Table 1
6 Conclusion
The GMPE suitable for the northeast region to assess PGA values at the bedrock
level has been examined in the present study. Eight GMPE models were selected for
the study and the log-likelihood value of each is calculated using the information-
theoretic approach of Scherbaum et al. [3]. The recorded data of the past five earth-
quakes in the region is used for the analysis after segmenting the data into three
different distance segments. Further, the weight of each GMPE is calculated using
Table 4 Weight of the GMPEs against each of the earthquakes used in the study for various distance
segments
EQ01 Bin I
GMPEs LLH DSI W Rank
NA09 1.817 3.04 0.17 4
ND10 0.768 113.08 0.35 1
NA12 2.738 − 45.59 NF NF
AN13 3.241 − 61.60 NF NF
RK14 21.40 − 100.00 NF NF
SI16 0.986 83.22 0.30 2
KU17 1.741 8.56 0.18 3
BA19 1.870 − 0.70 NF NF
EQ02 Bin I Bin II
GMPEs LLH DSI W Rank LLH DSI W Rank
NA09 2.001 92.63 0.26 3 NA
ND10 5.470 − 82.60 NF NF 0.938 125.32 0.48 1
NA12 4.982 − 75.61 NF NF NA
AN13 1.930 102.22 0.27 1 2.411 − 18.81 NF NF
RK14 6.591 − 92.01 NF NF 9.324 − 99.33 NF NF
SI16 8.302 − 97.56 NF NF 3.176 − 52.25 NF NF
KU17 2.320 54.35 0.21 4 1.898 15.79 0.25 3
BA19 1.957 98.57 0.27 2 1.739 29.29 0.27 2
EQ03 Bin II Bin III
NA09 NA NA
ND10 2.093 207.31 0.78 1 6.326 125.83 0.82 1
NA12 NA NA
AN13 4.539 − 37.37 NF NF NA
RK14 23.583 − 100.00 NF NF 41.382 − 100.00 NF NF
SI16 47.650 − 100.00 NF NF NA
KU17 NA NA
BA19 3.922 36.94 0.22 2 7.932 25.83 0.18 2
EQ04 Bin I Bin II
NA09 2.492 8.39 0.17 4 NA
ND10 1.614 99.17 0.31 1 1.055 182.12 0.52 1
NA12 4.850 − 78.86 NF NF NA
AN13 2.659 − 3.50 NF NF 3.629 − 52.63 NF NF
(continued)
Table 4 (continued)
RK14 17.232 − 100.00 NF NF 20.801 − 100.00 NF NF
SI16 3.660 − 51.77 NF NF 6.201 − 92.03 NF NF
KU17 2.112 41.00 0.22 3 2.413 10.09 0.20 3
BA19 1.716 85.57 0.29 2 1.943 52.45 0.28 2
EQ05 Bin II Bin III
NA09 NA NA
ND10 0.925 119.29 0.45 1 0.163 124.850514 0.9 1
NA12 NA NA
AN13 1.786 20.78 0.25 3 NA
RK14 5.610 − 91.47 NF NF 4.249 − 86.76018172 NF NF
SI16 8.697 − 99.00 NF NF NA
KU17 NA NA
BA19 1.469 50.40 0.31 2 2.024 21.09033232 0.1 2
Note NA—not applicable for respective distance bin, NF = not fit for weighting
Table 5 Sorting the GMPEs

Model fitting Distance bin Distance bin 2 Distance bin 3
based upon their fit to the data
1
Best fit ND10, BA19, ND10, BA19 ND10
KU17
Moderately fit NA09, AN13, KU17, AN13 BA19
SI16
Poorly fit NA12, RK14 RK14, SI16 RK14
the DSI value. Overall, NDMA [5] and Bajaj and Anbazhagan [11] model comes out
as the best suitable model for assessing PGA at the bedrock level for the Northeast
region. The final weight is calculated only for GMPEs having a positive DSI value.
The use of weights as suggested, in the calculation of PGA for hazard analysis can be
used to reduce the epistemic uncertainty associated with selection up to a significant
extent. The result indicates the use of a segmented ranking system since the use of
a single distance range can increase or decrease the rank of the model not fit for the
whole range. The use of lesser dataset for the analysis can inherit some uncertainty.
However, it can be reduced with the use of larger datasets.
References
1. Scherbaum, F., Cotton, F., Smit, P.: On the use of response spectral-reference data for the selec-
tion and ranking of ground-motion models for seismic-hazard analysis in regions of moderate
seismicity: the case of rock motion. Bull. Seismol. Soc. Am. 94(6), 2164–2185 (2004). https://
doi.org/10.1785/0120030147
2. Cotton, F., Scherbaum, F., Bommer, J.J., Bungum, H.: Criteria for selecting and adjusting
ground-motion models for specific target regions: application to central Europe and rock sites.
J. Seismol. 10(2), 137–156 (2006). https://doi.org/10.1007/s10950-005-9006-7
3. Scherbaum, F., Delavaud, E., Riggelsen, C.: Model selection in seismic hazard analysis: an
information-theoretic perspective. Bull. Seismol. Soc. Am. 99(6), 3234–3247 (2009). https://
doi.org/10.1785/0120080347
4. Nath, S.K., Raj, A., Thingbaijam, K.K.S., Kumar, A.: Ground motion synthesis and seismic
scenario in Guwahati city-a stochastic approach. Seismol. Res. Lett. 80(2), 233–242 (2009).
https://doi.org/10.1785/gssrl.80.2.233
5. NDMA.: Development of probabilistic seismic hazard map of India. In: Technical Report of the
Working Committee of Experts (WCE), National Disaster Management Authority (NDMA),
Govt. of India New Delhi (2010)
6. Nath, S.K., Thingbaijam, K.K.S., Maiti, S.K., Nayak, A.: Ground-motion predictions in
Shillong region, northeast India. J. Seismol. 16(3), 475–488 (2012)
7. Anbazhagan, P., Kumar, A., Sitharam, T.G.: Ground motion prediction equation considering
combined dataset of recorded and simulated ground motions. Soil Dyn. Earthq. Eng. 53, 92–108
(2013). https://doi.org/10.1016/j.soildyn.2013.06.003
8. Raghukanth, S.T.G., Kavitha, B.: Ground motion relations for active regions in India. Pure
Appl. Geophys. 171(9), 2241–2275 (2014). https://doi.org/10.1007/s00024-014-0807-x
9. Singh, N.M., Rahman, T., Wong, I.G.: A new ground-motion prediction model for Northeastern
India (NEI) crustal earthquakes. Bull. Seismol. Soc. Am. 106(3), 1282–1297 (2016). https://
doi.org/10.1785/0120150180
10. Kumar, A., Mittal, H., Kumar, R., Ahluwalia, R.S.: Empirical attenuation relationship for peak
ground horizontal acceleration for North-East Himalaya. Vietnam J. Earth Sci. 39(1), 47–57
(2017)
11. Bajaj, K., Anbazhagan, P.: Regional stochastic GMPE with available recorded data for active
region–application to the Himalayan region. Soil Dyn. Earthq. Eng. 126, 105825 (2019)
12. Goswami, H.C., Sarmah, S.K.: Probabilistic earthquake expectancy in the northeast Indian
region. Bull. Seismol. Soc. Am. 72(3), 999–1009 (1982)
13. Rahman, T.: Seismological model parameters for northeastern and its surrounding region of
India. Earthq. Sci. 25(4), 323–338 (2012)
14. Thingbaijam, K.K.S., Nath, S.K., Yadav, A., Raj, A., Walling, M.Y., Mohanty, W.K.: Recent
seismicity in Northeast India and its adjoining region. J. Seismol. 12(1), 107–123 (2008)
15. Motazedian, D., Atkinson, G.M.: Stochastic finite-fault modeling based on a dynamic corner
frequency. Bull. Seismol. Soc. Am. 95(3), 995–1010 (2005). https://doi.org/10.1785/012003
0207
16. Boore, D.M.: Stochastic simulation of high-frequency ground motions based on seismological
models of the radiated spectra. Bull. Seismol. Soc. Am. 73(6A), 1865–1894 (1983)
17. Campbell, K.W., Bozorgnia, Y.: Updated near-source ground-motion (attenuation) relations for
the horizontal and vertical components of peak ground acceleration and acceleration response
spectra. Bull. Seismol. Soc. Am. 93(1), 314–331 (2003). https://doi.org/10.1785/0120020029
18. Seeber, L., Armbruster, J.G., Jacob, K.H.: Probabilistic assessment of seismic hazard for
Maharashtra. Unpublished Report, Government of Maharashtra, Earthquake Rehabilitation
Cell Mumbai (1999)
19. Kijko, A., Graham, G.: “Parametric-historic” procedure for probabilistic seismic hazard anal-
ysis part II: assessment of seismic hazard at specified site. Pure Appl. Geophys. 154(1), 1–22
(1999). https://doi.org/10.1007/s000240050218
20. Beresnev, I.A., Atkinson, G.M.: FINSIM–a FORTRAN program for simulating stochastic
acceleration time histories from finite faults. Seismol. Res. Lett. 69(1), 27–32 (1998)
21. Boore, D.M.: Phase derivatives and simulation of strong ground motions. Bull. Seismol. Soc.
Am. 93(3), 1132–1143 (2003). https://doi.org/10.1785/0120020196
22. Iyengar, R.N., Kanth, S.T.G.R.: Attenuation of strong ground motion in peninsular India.
Seismol. Res. Lett. 75(4), 530–540 (2004). https://doi.org/10.1785/gssrl.75.4.530
23. Joyner, W.B., Boore, D.M.: Measurement, characterization, and prediction of strong ground
motion. In: Earthquake Engineering and Soil Dynamics II, Proceedings of American Society of
Civil Engineers Geotechnical Engineering Division Specialty Conference, pp. 43–102 (1988)
24. Douglas, J.: Ground motion prediction equations 1964–2021 (2021)
25. Delavaud, E., Scherbaum, F., Kuehn, N., Riggelsen, C.: Information-theoretic selection of
ground-motion prediction equations for seismic hazard analysis: an applicability study using
Californian data. Bull. Seismol. Soc. Am. 99(6), 3248–3263 (2009). https://doi.org/10.1785/
0120090055
26. Delavaud, E., Scherbaum, F., Kuehn, N., Allen, T.: Testing the global applicability of ground-
motion prediction equations for active shallow crustal regions. Bull. Seismol. Soc. Am. 102(2),
707–721 (2012)
27. Anbazhagan, P., Bajaj, K., Patel, S.: Seismic hazard maps and spectrum for Patna considering
region-specific seismotectonic parameters. Nat. Hazards 78(2), 1163–1195 (2015). https://doi.
org/10.1007/s11069-015-1764-0
28. Archuleta, R., Steidl, J., Squibb, M.: The cosmos virtual data center. In: Directions in Strong
Motion Instrumentation. Nato Science Series: IV: Earth and Environmental Sciences, vol. 58,
pp. 209–222 (2005). Springer, Dordrecht. https://doi.org/10.1007/1-4020-3812-7_13
Influence of Site-City Interaction
on the Response of Buildings
on Trapezoidal Basin
Neeraj Kumar, J. P. Narayan, Pooja Lohchab, and Sanjay Kumar
Abstract The paper presents the response of cluster of buildings located on soft
sediment in double resonance situation. In order to comprehend the influence of Site-
city interaction on the response of the buildings, two symmetrical models–Cluster 1
(9 buildings) and Cluster 2 (25 buildings) situated on trapezoidal shaped basin are
simulated using Finite Difference algorithm. The analysis of FDM simulation for
3D SCI reveals a reduction in spectral amplitude at natural frequency of building
which increases with the number of buildings. The reduction of spectral amplitude
is of the order of 65% in the case of cluster 2 with 25 buildings as compared to
standalone building located at the center of closed 3D trapezoidal basin. The S-wave
response of buildings shows much larger value of spectral amplitude (approx. 180)
for standalone building for 3D analysis in comparison with that of 2D analysis for
SH and SV-wave and also there is larger reduction of Spectral Amplification Factor
(SAF) at fundamental frequency with the splitting of bandwidth for both Cluster 1
and Cluster 2. These findings call for the urgent need of 3D Site-City Interaction
studies in urban environment in development of the seismic resilient and sustainable
city.
Keywords Site-city interaction · Basin effects · S-wave response
1 Introduction
The effects of Site-City interaction were initially pragmatic while investigating

the complex response of buildings and the long duration of ground motions with
beating phenomenon in the Mexico City during the Michoacan 1985 earthquake
[1–3]. The past studies on the site-city interaction (SCI) effects were carried out for
N. Kumar (B) · P. Lohchab

Central University of Haryana, Mahendergarh, Haryana 123031, India
e-mail: neerajlohchab@cuh.ac.in
J. P. Narayan · S. Kumar
Indian Institute of Technology Roorkee, Roorkee, Uttrakhand 247667, India
https://doi.org/10.1007/978-981-99-1459-3_9
102 N. Kumar et al.
two-dimensional site-city models in double resonance condition and reported that

SCI significantly modifies the seismic response of buildings and free-field ground
motion as well as splits the frequency band at natural mode of vibration of building
[4–10]. The response of a building depends on the sediment parameters of the basin
and its geometry. The role of number of the buildings, the city-density and hetero-
geneity of city on the response of buildings and motions without any structures over
ground surface are studied in details [6, 8, 11–13]. The SCI effects are significant
when two conditions are satisfied: (1) buildings rest on soft soil and (2) buildings
and underlying layer of sediment are in double resonance. The structural parame-
ters (especially mass and stiffness) also play crucial role in dynamic response of
buildings during earthquake [8, 13, 14]. Most of the studies are limited to the two-
dimensional simulations of site-city models and limited literature is available on the
three-dimensional SCI effects on free-field motion and response of buildings. There
are two major issues with this specific case (1) increase in the natural frequency and
amplification of basin due its 3D nature and (2) reduction in the dominant frequency
of building as its rest on the very soft sediment resulting in rocking motion. Recently,
the study carried by Kumar and Narayan [14, 15] and Chandra and Gueguen [16]
reveals that the response of buildings is altered by the presence of other nearby build-
ings. Kumar and Narayan [12] have simulated the response of buildings situated on
closed 2D basins considering the two cities of 3 buildings and 5 buildings with SH-
and SV-wave. Lu et al. [17] have simulated a 3D site-city model having 9 building
group located on 3D trapezoidal basin using vertically propagating S-wave (Ricker
wavelet) and also performed a large-scale non-linear time history analysis of build-
ings at Tsinghua University campus, Beijing on a regional scale. However, a very
limited literature is available on 3D SCI effects considering the double resonance
situation buildings [14]. Present paper focuses on the response of cluster of buildings
situated in 3D trapezoidal basin in double resonance situation and the reduction in
spectral amplification of buildings with splitting.
2 Model Parameters
In order to comprehend the three-dimensional site-city interaction effects on the

behavior of buildings, two groups of buildings: Cluster 1 (comprising 9 buildings)
and Cluster 2 (comprising 25 buildings) are considered on trapezoidal basin (Fig. 1).
The size of trapezoidal basin in the horizontal plain is taken as 351 m (both NS- and
EW-directions) (Fig. 2). The utmost depth of basin is fixed as 51 m. The dimension
of flat part of basement of basin is 303 m in both the lateral directions. Generally, the
shapes of the ponds/lakes filled with soil are semi-circular and trapezoidal in nature.
The parameters and size of trapezoidal basin and building blocks are kept same as
that considered in the Kumar and Narayan [13] to compare the results of 3D SCI
with that of 2D SCI on the response of building. The buildings are incorporated as a
3D homogenous visco-elastic block with 5% damping [18], instead of real building
[12, 14, 19]. The lateral dimensions of the buildings are fixed to 33 m × 33 m and the
Influence of Site-City Interaction on the Response of Buildings … 103
building’s height is accustomed so that the natural frequency of building coincides

with that of the underlying basin. The plans of Cluster 1 and Cluster 2 are shown in the
Fig. 1. Three-dimensional SCI effects on the S-wave responses of basin and cluster
of buildings are studied using three-dimensional staggered grid visco-elastic finite
difference algorithm developed by Narayan and Sahar [16]. Time-domain simulation
is executed including frequency-dependent damping based on GMB-EK rheological
model [20]. To avoid reflection from the edges of FD domain, sponge absorbing
boundaries are implemented [21] at bottom and side-faces and a modified vacuum
formulation is used for free surface boundary. To simplify the nomenclature, the
buildings are named as ‘A’, ‘B’, ‘C’, ‘D’ and ‘E’ in both the directions, where ‘C’
is at the center of the city, as shown in Fig. 1. The gap between two constitutive
buildings is kept as 9 m. Recorders are positioned at the top of building blocks,
which are located in the NS- and EW-arrays crossing each other at the center of
the basin. The rheological parameters for building blocks, soil sediment and bedrock
such as P-and S-wave velocity, density, quality factor and unrelaxed moduli are given
in Table 1 [10, 13, 14]. The site-city models are excited with a North–South (NS)
polarized vertically propagating plane S-wave front. A plane wave front at a depth
of 125 m is generated in form of Ricker wavelet using shear stress σxz [14]. To
comprehend the SCI effects on the response of cluster of buildings, the response of a
standalone building (C1) located at the center of the trapezoidal basin is considered
as a reference.
Fig. 1 Plan of Cluster 1 (Left) and Cluster 2 (Right) comprising of 9 buildings and 25 buildings,
respectively
104 N. Kumar et al.
Fig. 2 Sketch of Cluster 2

comprising 25 buildings
located in trapezoidal basin
Table 1 The considered parameters for building block, basin and bedrock models
Materials V S (m/s) V P (m/s) QS QP Density Unrelaxed Moduli (GPa)
(kg/m3 ) μu Ku λu
Building block 120 204 10 17 350 0.0064 0.0168 0.0039
Basin 350 860 35 86 2000 0.2628 1.5218 0.9962
Bedrock 1800 3120 180 312 2500 8.2104 24.526 8.1058
3 S-Wave Response of 3D Trapezoidal Basin
In order to study the response of cluster of buildings located in closed 3D trapezoidal

basin in double resonance situation, first spatial variation of spectral amplifications
at different modes of vibrations is required, since spectral amplifications at higher
modes can be larger than the natural mode of vibration near the edges. The recorders
are placed in two arrays oriented in the East–West (EW) and North–South (NS)
directions crossing perpendicular to each other at the center of basin. The spacing
between two recorders is kept as 21 m to record the EW, NS and UD (Up-Down)
components of the motion at the surface of basins.
3.1 Seismic Response of Basin
The S-wave responses of the trapezoidal basin model are recorded in both the NS- and
EW-array (Fig. 2). Figure 3a, b describes the NS-components and UD-component
of S-wave in the NS-array, respectively, and Fig. 3c describes the NS-components
in EW-array of the recorded S-wave in basin model. An increase in the amplitude
and the duration of recorded signals toward the center of 3D basins is observed
in NS-array. This is possibly due to the interference of basin generated Rayleigh
waves, diffracted waves from basin edges and the entrapped bidirectional seismic
waves which act in both the horizontal and vertical directions [22]. The obtained
signal in the UD-components in NS-array may be due to the bending of the incident
S-wave, which acts as an SV-wave in the NS direction and the Rayleigh waves
generated by basin. Though, the amplitude of Rayleigh wave is small as compared
to the amplitude of the incident S-wave. The reverse polarity of the diffracted body
wave at the opposite edges of basin and basin generated Rayleigh waves has resulted
in zero amplitude of the recorded signal in the vertical component at the center
of basin. Whereas in EW-array, an increase in amplitude and duration of recorded
signals toward the center of 3D basins is due the interference of basin generated
surface waves, diffracted waves from basin edges and the trapping of seismic waves
[22]. The recorders in EW-array do not record signal in the UD-components, as the
incident S-wave has NS-polarization. Further, along this plane, the incident S-wave
will behave as SH-wave so only generated Love waves or diffracted SH-waves will
be recorded in EW-array.
3.2 Spectral Amplification in Basin
The spatial spectral amplifications of NS-component of ground motion in the NS-

and EW-array in the trapezoidal basin are presented in Fig. 4. The figure reveals
a number of modes of vibrations of the basin in the considered frequency band.
Among these modes of vibrations, the fundamental (F 0 ) and first (F 1 ) modes are
the dominant one and have highest values of spectral amplifications. Further, the
frequency consequent to fundamental and first modes are matching in the NS- and
EW-arrays in the respective basin. But the spatial spectral amplification pattern is
only same for the fundamental mode and not for the first mode in the EW- and NS-
arrays. However, other higher modes observed in basins are not matching in the both
the arrays.
Furthermore, the spatial spectral amplifications corresponding to the first mode
(F 0 ) of vibration in the EW-array is larger than that at fundamental mode of vibration
near the edges in the basin. The complex interference of the multiples wave like
incident S-wave, basin generated surface waves and diffracted waves make it highly
complex to recognize that the combination of which modes (horizontal or vertical
modes) are responsible for the creating the higher modes of vibrations in basins
B
[13]. The obtained Spectral Amplification Factor (SAF) and F03D (natural frequency
of closed 3D basins) of trapezoidal basins is 19.99 and 1.90 Hz, respectively. The
B
SAF at F03D is largest and decay toward edges of basin. Further, the obtained SAF
patterns subsequent to the fundamental and first mode of vibrations in the NS- and
EW-arrays in the basin very much match with those obtained in a 2D basin for the
SV- and SH-waves, respectively.
106 N. Kumar et al.
Fig. 3 a and b The NS- and UD-components of S-wave responses in the NS-array and c NS-
component of S-wave responses in EW-array are shown for Trapezoidal Basin model
4 Site-City Interaction Effects in Basin
The responses of buildings of Cluster 1 and Cluster 2 located on basin and their SCI
effects in double resonance condition are studied. The parameters of building blocks,
F0 F1 F0 F1
Fig. 4 The spectral amplification of NS-components for the NS-array and EW-array corresponding
to trapezoidal basin model
basin and bedrock are given in Table 1. The spatial geometries of buildings are taken
as 33 m × 33 m in lateral direction. The height of building is considered as 11.1 m
using empirical relation given by Narayan and Kumar [14], so that natural frequency
S
of building blocks (F03D ) matches with that of basin. To simplify the investigation as
well as to address the responses of different buildings located in different basins and
arrays, a unique name is proposed for each building according to its location. The
proposed name for the responses of a standalone building at the center of considered
trapezoidal basin is ‘C1’. The proposed nomenclature for the responses of buildings
108 N. Kumar et al.
Table 2 The 3D natural B = F S (Hz) B S (Hz)

S ) Building F03D 03D F03D SAF
frequency of building (F03D
B S ), C1 1.90 1.88 183.88
at rock and in basin (F03D
and largest SAF in the Cluster 1 (9 buildings)
frequency band of F03DB S of the C9EW 1.90 1.81 84.14
S-waves SCI models in basin B9EW 1.81 71.31
B9NS 1.81 75.66
Cluster 2 (25 buildings)
C25EW 1.90 1.77 64.63
B25EW 1.77 58.08
A25EW 1.77 40.64
B25NS 1.77 59.27
A25NS 1.77 43.97
‘C’, ‘B’ and ‘A’ of the Cluster 2 (25 buildings) in the EW-array are C25EW, B25EW
and A25EW, respectively. Similarly, proposed names for the responses of buildings
‘B’ and ‘A’ of the Cluster 2 (25 buildings) in the NS-array are B25NS and A25NS,
respectively. In the case of Cluster 1 (9 buildings) located in the trapezoidal basin,
the proposed names for the responses of buildings ‘C’ and ‘B’ in the EW-array are
C9EW and B9EW, respectively, and proposed names for the responses of building
B in the NS-array is B9NS (Table 2).
4.1 Response of Standalone Building
The response of standalone building (C1) at the center of trapezoidal basin is shown
in Fig. 5a. The maximum value of amplitudes is 28.59 cm/s of building C1. The
spectral amplifications of standalone building C1 is shown in Fig. 5b. The analysis
SB
of this figure depicts that the F03D of standalone building in basin is 1.88 Hz and
SB
the corresponding SAFs are 183.88 of C1 building. A minor reduction of F03D of
standalone building at the center of basins can be inferred as compared to that on the
rock [14]. The obtained ASA value is 9.40 in the standalone buildings.
4.2 SCI Effects in Cluster 1 (9 Buildings)
The S-wave responses of all the nine buildings of the Cluster1 located in the trape-
zoidal basin are computed at the center of roof of individual buildings. Figure 5a
present the responses of the buildings of Cluster 1 located in the trapezoidal basin.
The obtained largest amplitude of the buildings C9EW, B9EW and B9NS of the
Cluster1 in the trapezoidal basin are −19.82 cm/s, −17.96 cm/s and 18.63 cm/s,
Fig. 5 a The NS-component of the standalone building, Cluster 1 buildings and Cluster 2 buildings
located in trapezoidal basin model b, c a comparison of spectral amplifications of the NS-component
of buildings of Cluster 1 (9 Buildings) and Cluster 2 (25 Buildings) located in the trapezoidal basin
with that of a standalone building ’C’ located at the center of the trapezoidal basin
respectively. A comparison of response of standalone building of C1 model with the

respective building C9EW of the Cluster 1 depicts a large reduction of amplitude
in the Cluster1 due to SCI effects. In addition to this, larger amplitude of building
‘C’ as compared to the building ‘B’ in both EW- and NS-array of Cluster 1 is also
observed, that may be due to the larger SAF at FB03D of basin at the location of
building ‘C’ in the site-city models. In other words, the 3D resonance phenomenon
in the considered basin with a large shape-ratio may be responsible for this obser-
vation. Figure 5b present the spectral amplifications of buildings of the Cluster 1
SB
located in trapezoidal basin. The obtained SAFs at F03D (1.81 Hz) of the buildings
110 N. Kumar et al.
C9EW, B9EW and B9NS are 84.14, 71.31 and 75.66, respectively. The obtained
SB
F03D as 1.81 Hz of the buildings in the trapezoidal basins show that SCI effect due
to City1 has reduced the natural frequency of building by about 4.73%, respectively,
as compared to that of building on rock. The SCI effects due to the Cluster1 have
reduced the value of largest SAF of the order of 54.24%, 61.21% and 58.85% for the
buildings C9EW, B9EW and B9NS, respectively, in the trapezoidal basin. The ASA
values are 6.95, 6.23 and 6.32 in the case of buildings C9EW, B9EW and B9NS in
Cluster 1, respectively, being 26.0%, 33.7% and 32.7% lesser compared with that
of the C1 model (9.40). Even though, it is apparent that the SCI effect is least for
building located at the center of the Cluster 1, but it is occurring because of the used
common reference transfer function corresponding to the standalone building at the
SB
center of basin. A splitting of the spectral bandwidth of F03D of the buildings in the
city can be inferred.
4.3 SCI Effects in Cluster 2 (25 Buildings)
The S-wave responses of all the twenty-five buildings of the Cluster 2 located in the
trapezoidal basin are computed at the center of roof of individual building. Figure 5a
presents the responses of buildings C25EW, B25EW A25EW, B25NS and A25NS
of the Cluster 2 in the trapezoidal basin, wherein largest amplitude is 15.81 cm/s,
−15.69 cm/s, −12.57 cm/s, 14.87 cm/s and 11.89 cm/s, respectively. The obtained
amplitudes of buildings in the EW-array are larger than that in the NS-array at the
same distance from the center of the cluster. Finally, it may be hypothesized that the
SCI effects is responsible for the reduction of level of vibration of buildings and this
minimization in shaking is increasing with the number of buildings. The spectral
amplifications of buildings C25EW, B25EW, A25EW, B25NS and A25NS of the
Cluster 2 in the trapezoidal basin is shown in Fig. 5c, wherein the obtained largest
SB
SAFs at F03D (1.77 Hz) are 64.63, 58.08, 40.64, 59.27 and 43.97, respectively. The
SCI effects due to Cluster 2 have reduced the value of largest SAF of the order of
64.85%, 68.41%, 77.89%, 67.76% and 76.08% for the buildings C25EW, B25EW,
A25EW, B25NS and A25NS, respectively, of the C25EW, B25EW, A25EW, B25NS
SB
and A25NS in the trapezoidal basin (Table 2). The obtained F03D as 1.77 Hz for
SB
the trapezoidal basin depicts that SCI effects have reduced the F03D of building by
6.84%, as compared to the standalone building on rock. A splitting of the spectral
SB
bandwidth of F03D of the buildings in the Cluster 2 can also be inferred. It appears
that this splitting is occurring due to the very large drop of spectral amplification at
S
F03D of building on rock. This drop may be due to the inertial effects of the buildings.
S
The spectral amplification drops at F03D of building on rock in the case of Cluster
2 is much larger than that in the case of Cluster 1. The ASA values are 5.87, 5.49,
4.85, 5.52 and 4.61 in the case of buildings C25EW, B25EW, A25EW, B25NS and
A25NS in Cluster 2, respectively, being 37.5%, 41.5%, 78.4%, 41.2% and 59.5%
lesser compared with that of the C1 model (9.40).
5 Discussion
SB
A considerable reduction of F03D of buildings is inferred due to the SCI effects as
S
compared to F03D of standalone building on rock and this reduction is proportional
to the number of buildings in the city. For example, the percentage reduction of
SB
F03D of building were of the order of 5% and 7% in the Cluster 1 (9 buildings) and
Cluster 2 (25 buildings), respectively, located in the 3D trapezoidal basin (Table 3).
SB
However, the % reduction of F02D of building were of the order of 12% and 14% for
the SH-wave in the City1 (3 buildings) and City2 (5 buildings), respectively, located
in the trapezoidal 2D basin (Table 3).
The obtained SAF at natural frequency of standalone 3D building in 3D basin
(183.8) in double resonance case is also much larger than that of standalone 2D
building in the 2D basin (51.94 and 71.51 for the SH- and SV-waves, respectively).
But, at same time, the obtained decrease of largest SAF 3D building of the Cluster
2 (64.85%) wherein 25 buildings are located in 3D trapezoidal basin is much larger
than that of 2D building of a city with 5 buildings located in the 2D trapezoidal
basin (40 and 40.2 for the SH- and SV-waves, respectively) [13]. Further, this drop
of largest SAF of buildings of the city is proportional to the number of buildings in
the city. As in practice, the obtained % reduction of largest SAF of building of the
Cluster 1 and Cluster 2 were of the order of 54.24% and 64.85%, respectively, in the
case of city located in the trapezoidal basin (Table 3).
A splitting of frequency band at natural mode of vibration (F02D SB
and FSB
03D ) is
obtained in the cases of SH-wave and SV-wave responses of buildings of city located
in the 2D basins as well as the S-wave responses of buildings of Cluster 1 and Cluster
2 located in the 3D basins. The splitting and a drop in the SAF at the frequency
S
F02D /F03D
S
due to the SCI effects can be explained in terms of inertial effects of the
S SB
buildings. The splitting is due to very large drop of SAF at F02D and F03D of standalone
building on rock. This drop in SAF at natural frequency of standalone building on
rock is proportional to the number of buildings in the city. For example, the SAF drop
S
at F03D of standalone building on rock in the case of Cluster 2 located in 3D basin is
larger than that in the case of Cluster 1 located in 3D basin. Furthermore, the SAF
S
drop at F03D of standalone building on rock in the case of Cluster 2 (25 buildings)
located in 3D basin is larger than that in the case of City 2 (5 buildings) located in
SB
2D basin. The obtained splitting of spectral bandwidth of F02D of buildings of city
SB
located in the 1D and 2D basins and F03D of buildings of city located in 3D basin
due to the SCI effects corroborates with that reported by Kham et al. [6], Semblat
et al. [8] and Kumar and Narayan [13] based on the analysis of 2D SCI studies.
6 Conclusions
A comparison of the observed 3D SCI effects on the S-wave response of buildings

of the Cluster 1 and Cluster 2 reveals that the SCI reduced the amplitude of shaking
112
S B ) in basin and in largest SAF in the

Table 3 A comparison of 2D and 3D response of the building ‘C’, % age reduction in natural frequency of building (F03D
frequency band of natural mode in the different SCI models
Comparison of 2D and 3D models
SCI models 2D SH-wave 2D SV-wave
Building “C” s
F02D = B
F02D = 1.70 Hz s
F02D (Hz) % age SAF % age s
F02D B = 1.85 Hz
= F02D s
F02D (Hz) % age SAF % age
Standalone building 1.57 7.6 51.9 – 1.72 7.02 71.5 –
Cluster 1 1.50 11.7 36.1 30.5 1.65 10.8 48.4 32.3
Cluster 2 1.47 13.5 31.2 40.0 1.62 12.4 42.7 40.2
Comparison of 2D and 3D models
SCI models 3D SV-wave
Building “C” s
F02D B = 1.90 Hz
= F02D s
F02D (Hz) % age SAF % age
Standalone building 1.88 1.05 183.8 –
Cluster 1 1.81 4.73 84.14 54.24
Cluster 2 1.77 6.84 64.63 64.85
N. Kumar et al.
to greater extent in all the buildings of the city and this reduction is larger at the
outermost buildings of the city and least at the center of city for 3D closed basins as
observed in case of 2D closed basin [13]. Further, the reduction effects of SCI are
larger on the response of buildings of the Cluster2 than that on the buildings of the
Cluster1 in the respective basin, which indicated that the SCI effects increases with
the number of buildings of the city, but the effects are more prominent in case of 3D
basins as compared to 2D basins. These findings call for the urgent need of 3D SCI
studies in urban environment especially in the areas of high seismic hazard due to
basin effects.
References
1. Chávez-García, F.J., Bard, P.Y.: Site effects in Mexico City eight years after the September
1985 Michoacan earthquakes. Soil Dyn. Earthq. Eng. 13(4), 229–247 (1994)
2. Wirgin, A., Bard, P.Y.: Effects of buildings on the duration and amplitude of ground motion in
Mexico City. Bull. Seismol. Soc. Am. 86(3), 914–920 (1996)
3. Gallipoli, M.R., Mucciarelli, M., Castro, R.R., Monachesi, G., Contri, P.: Structure, soil–struc-
ture response and effects of damage based on observations of horizontal-to-vertical spectral
ratios of microtremors. Soil Dyn. Earthq. Eng. 24(6), 487–495 (2004)
4. Stewart, J.P., Seed, R.B., Fenves, G.L.: Seismic soil-structure interaction in buildings. II:
empirical findings. J. Geotech. Geoenviron. Eng. 125(1), 38–48 (1999)
5. Guéguen, P., Bard, P.Y.: Soil-structure and soil-structure-soil interaction: experimental evidence
at the Volvi test site. J. Earthquake Eng. 9(5), 657–693 (2005)
6. Kham, M., Semblat, J.F., Bard, P.Y., Dangla, P.: Seismic site–city interaction: main governing
phenomena through simplified numerical models. Bull. Seismol. Soc. Am. 96(5), 1934–1951
(2006)
7. Groby, J.P., Wirgin, A.: Seismic motion in urban sites consisting of blocks in welded contact
with a soft layer overlying a hard half-space. Geophys. J. Int. 172(2), 725–758 (2008)
8. Semblat, J.F., Kham, M., Bard, P.Y.: Seismic-wave propagation in alluvial basins and influence
of site-city interaction. Bull. Seismol. Soc. Am. 98(6), 2665–2678 (2008)
9. Guéguen, P., Colombi, A.: Experimental and numerical evidence of the clustering effect of
structures on their response during an earthquake: a case study of three identical towers in the
city of Grenoble, France. Bull. Seismol. Soc. Am. 106(6), 2855–2864 (2016)
10. Sahar, D., Narayan, J.P.: Quantification of modification of ground motion due to urbanization in
a 3D basin using viscoelastic finite-difference modelling. Nat. Hazards 81(2), 779–806 (2016)
11. Tsogka, C., Wirgin, A.: Simulation of seismic response in an idealized city. Soil Dyn. Earthq.
Eng. 23(5), 391–402 (2003)
12. Sahar, D., Narayan, J.P., Kumar, N.: Study of role of basin shape in the site–city interaction
effects on the ground motion characteristics. Nat. Hazards 75(2), 1167–1186 (2015)
13. Kumar, N., Narayan, J.P.: Quantification of site–city interaction effects on the response of
structure under double resonance condition. Geophys. J. Int. 212(1), 422–441 (2018)
14. Kumar, N., Narayan, J.P.: Quantification of Fundamental Frequencies of 3D Basins and Struc-
tures and Site-City Interaction Effects on Responses of Structures. Pure Appl. Geophys. 176,
4477–4502 (2019)
15. Kumar, N., Narayan, J.P.: Effects of site-city-interaction and polarization of the incident wave
on the transfer function and fundamental frequency of structures. Nat. Hazards 97(2), 747–774
(2019)
16. Narayan, J.P., Sahar, D.: Three-dimensional viscoelastic finite-difference code and modelling of
basement focusing effects on ground motion characteristics. Comput. Geosci. 18(6), 1023–1047
(2014)
114 N. Kumar et al.
17. Lu, X., Tian, Y., Wang, G., Huang, D.: A numerical coupling scheme for nonlinear time history
analysis of buildings on a regional scale considering site-city interaction effects. Earthquake
Eng. Struct. Dynam. 47(13), 2708–2725 (2018)
18. Indian Standard IS–1893:2002 (Part 1), Criteria for earthquake resistant design of structures—
Part 1: General provision and buildings, Bureau of Indian Standards, New Delhi (2000)
19. Bard, P.Y., Chazelas, J.L., Gueguen, P., Kham, M., Semblat, J.F.: Site–City Interaction in
Assessing and Managing Earthquake Risk, Chapter 5. Springer (2005)
20. Emmerich, H., Korn, M.: Incorporation of attenuation into time-domain computations of
seismic wave fields. Geophys. 52(9), 1252–1264 (1987)
21. Israeli, M., Orszag, S.A.: Approximation of radiation boundary conditions. J. Comput. Phys.
41(1), 115–135 (1981)
22. Kamal, N.J.: 3D basin-shape ratio effects on frequency content and spectral amplitudes of basin-
generated surface waves and associated spatial ground motion amplification and differential
ground motion. J. Seismol. 19, 293–316 (2015)
Assessment of Double Resonance
from Microtremor Observations
for Jammu Region in India
Abdullah Ansari , Falak Zahoor , K. S. Rao , A. K. Jain, Aashi Pal,

Neeraj Kumar, Sakib Majid Hajam, Pallavi Shukla, Krishna Sharma,
Faizan Fayaz, Mir Akhtar Yousuf, Shakir Riyaz, and Umer Altaf Khan
Abstract The Jammu region (JR) in the union territory of India is located in the
northwestern Himalayas, affected by moderate to large magnitude earthquakes. To
assess the possibility of double resonance effects (DRE) during significant earthquake
events, this study examines the influence of Reinforced Concrete (RC) building
vibrations in the JR in the context of local seismic response investigations. Using
microtremor, a single station seismic ambient noise investigation was conducted at
242 locations to assess the predominant frequency (f p ) and H/V spectra for subsurface
soil. During the field survey, building data is collected, and an empirical relationship
between building height and vibration period is developed. The findings show that the
fundamental frequency (f f ) of 17.35% of structures overlaps with the predominant
frequency (f p ) of the site locations in the study area where microtremor testing was
done. The vulnerability map divides the city into three discrete zones with high,
medium, and low resonance levels, allowing for better hazard mitigation and town
planning.
Keywords H/V spectra · Double resonance · Predominant frequency · Jammu ·

Himalayas
1 Introduction
Jammu and Kashmir is located in the northwestern Himalayas, one of the world’s
most seismically active areas, with major earthquakes, triggered in 1555, 1828, 1885,
A. Ansari (B) · F. Zahoor · K. S. Rao

Department of Civil Engineering, Indian Institute of Technology Delhi, Hauz Khas, New
Delhi 110016, India
e-mail: aamomin183@gmail.com; ansariaa@civil.iitd.ac.in
F. Zahoor · A. K. Jain · A. Pal · N. Kumar · S. M. Hajam · P. Shukla · K. Sharma · F. Fayaz ·
M. A. Yousuf · S. Riyaz · U. A. Khan
Department of Civil Engineering, National Institute of Technology Srinagar, Jammu and Kashmir
190006, Srinagar, India
https://doi.org/10.1007/978-981-99-1459-3_10
116 A. Ansari et al.
1905, 2005, and 2019 [1–3]. The 2005 Kashmir earthquake (M w = 7.6) was one
of the most devastating, killing over one lakh people and destroying infrastructure
such as bridges, retaining walls, dams, and building structures. Recent moderate
earthquakes in this region include the 2013 Kishtwar earthquake (M w = 5.7) and
the 2019 Mirpur earthquake (M w = 5.6). The 2019 Mirpur earthquake in Pakistan
was felt in the Jammu region (JR), Punjab region, Uttarakhand, and Delhi killing
50 people and causing property damage in villages surrounding Mirpur city such
as Jatlan, Manda, and Afzalpur [4–7]. The slope failure and ground shaking were
observed in the Rajouri and Poonch districts due to the 2019 Mirpur earthquake. The
post-effect of these earthquake events created the problem of liquefaction in different
parts of Jammu and Kashmir [1, 8, 9].
The bridge construction, building development schemes, tunnelling, and national
highway expansion-related infrastructure projects as well as urbanisation-oriented
construction projects and schemes have advanced at a higher rate in this area in
recent [10–12]. In the map shown in Fig. 1, ongoing infrastructure projects in Jammu
and Kashmir are highlighted. The Udhampur Srinagar Baramulla Rail Link Project
(USBRL) includes the 11.2 km long Pir Panjal railway tunnel, which is located north
of Banihal town in Ramban district and travels through the Pir Panjal Range in the
central Himalayas. The historical record of earthquakes and seismotectonic settings,
as well as urbanisation and industrial growth within the JR, drive the need to quantify
local seismic response and investigate the implications of double resonance for the
safe design of RC building structures.
Fig. 1 Map of Jammu Region (JR) in Jammu and Kashmir. Red circles indicate the test locations
Assessment of Double Resonance from Microtremor Observations … 117
2 Single Station Microtremor Testing
There are twenty districts in Jammu and Kashmir, ten in the Jammu region (JR), and
ten in the Kashmir valley (KV). The Chenab, Ravi, Tawi, and Poonch are the major
rivers passing through the JR covering a total area of 26,293 km2 . The near-field
as well as far-field earthquakes throughout the Himalayan region, extending up to
Hindukush in Afghanistan, have been flicked by both JR and KV [13, 14]. Natural
(earthquake, wind, sea waves, etc.) and manmade (traffic, industrial sound, people
walking on the street, etc.) sources contribute to ambient seismic noise. Nogoshi [15]
developed the Horizontal-to-Vertical Spectral Ratio (HVSR) analysis of microtremor
recordings, which was emphasised by [16]. This approach is used to calculate the
amplitude spectrum by performing a Fourier transform on the raw data. To exclude
the influence of site effects, propagation route, and source effects, the ratio of the
squared average of N-S and E-W amplitude spectrum to the vertical amplitude spectra
is examined. The maxima for HVSR curves relate to seismic wave amplification
induced by a strong impedance difference between underlying strata highlighting
the predominant frequency (f p ) for foundation soil. These frequencies represent the
approximate values at which the soil might undergo resonance if an earthquake of
similar frequency occurs in the region [13, 17].
In this study, a Tromino [18] with a frequency range of 0.1–1000 Hz was employed
to record ambient noise and microtremors. The ambient noise measuring instruments
provides the level of constant vibration of the Earth interior. This is referred to
as seismic noise, and it is produced as a result of both high-frequency manmade
events and low-frequency natural occurrences. Nakamura [16] proposed that site
response can be anticipated from Horizontal-to-Vertical Spectral Ratio (HVSR) in
the frequency domain of seismic microtremors. The amplitude of the H/V curve can
be estimated using the mathematical relationship pointed out in Eq. (1).
/
HN2 S (ω) + HE2 W (ω)
HVSR(ω) = (1)
VU P−D O W N (ω)
By leaving the equipment at the testing position undisturbed for around 30 min,
passive noise was captured at a sampling rate of 128 Hz (Fig. 2). Readings were
obtained at 242 unique locations covering the whole JR, during peaceful times to
avoid influence from undesired anthropogenic noise. The geology of Jammu and
Kashmir divides the region into three structural zones: the Pir Panjal, Zanskar, and
Tertiary groups. Depending on the local geology and topography, generally clear,
multiple, and flat types of curves are observed. For high, medium, and low-risk levels
of DRE for sites exhibiting clear peaks for HVSR curves are illustrated in Fig. 3.
The field observations for sites showing the clear peak during extensive geophysical
testing are reported in Table 1.
Fig. 2 Single station HVSR microtremor field setup showing all accessories needed to perform the
geophysical testing
Fig. 3 HVSR function

derived from microtremor
for three site locations (43,
159, and 217) indicating the
various resonance modes.
HVSR curves for site
locations 43 and 159 present
a high and moderate risk for
double resonance,
respectively. The magenta
colour curve is shown for
site location 217 which is
subjected to low-level double
resonance effect (DRE)
3 Field Survey and Structural Integrity of Building

Scenario
During the field survey in the Jammu region (JR), RC building data was collected
for all test locations where microtremor readings were measured. The southwestern
part of JR is flat while the rest is made up of the Jammu hills, with elevations ranging
between 330 m at the Jammu to 1638 m at the Kishtwar. Jammu is the winter capital
of Jammu and Kashmir; hence, the maximum population is attached to Jammu and
neighbouring districts like Udhampur, Doda, Samba, and Kathua. The topography of
the northern part of Ramban and Doda, as well as Kishtwar which shares the border
Table 1 Frequency and amplitudes of HVSR curves with a clear peak in the Jammu region (JR)
Site no. f p (Hz) H/V Site no. f p (Hz) H/V Site no. f p (Hz) H/V
1 6.15 4.57 74 0.86 2.08 157 5.74 3.34
2 2.1 1.45 75 0.96 2.56 158 6.08 5.09
3 6.05 4.66 76 5.78 3.12 159 5.32 4.78
4 6.26 4.25 77 6.05 6.02 160 7.49 1.92
5 5.94 3.16 78 5.31 7.22 161 6.33 3.55
6 5.74 3.34 79 5.33 6.29 162 6.31 5.51
7 7.58 1.98 81 5.59 3.79 163 7.54 1.87
8 5.64 6.61 82 5.94 3.16 164 7.38 2.48
9 0.89 2.03 84 5.64 6.61 165 7.76 1.85
10 1.18 2.86 85 5.55 5.31 166 7.45 1.95
11 0.84 2.96 86 1.23 1.68 167 7.89 1.82
12 0.85 2.29 91 3.44 2.19 168 7.58 2.15
13 0.92 2.02 93 1.45 3.36 170 7.71 1.98
14 0.86 2.62 94 1.26 2.98 173 7.37 1.71
15 0.94 1.74 96 1.41 2.11 174 7.58 1.92
17 0.38 1.88 97 1.85 1.92 175 7.33 2.11
18 0.87 4.01 100 2.21 2.15 176 7.63 1.66
19 4.28 2.99 103 7.64 2.11 177 0.84 2.07
20 0.51 1.93 104 1.14 2.21 178 0.95 1.93
22 0.99 2.91 105 3.32 2.22 179 0.99 2.91
24 1.02 2.41 108 0.37 1.25 181 5.33 6.29
25 1.26 2.88 111 0.94 2.25 182 6.15 4.23
26 0.63 1.81 112 7.64 1.88 184 1.14 1.16
29 3.53 1.74 113 1.41 2.11 187 1.49 2.13
31 7.64 1.95 114 1.26 2.88 194 5.49 4.69
32 3.18 2.23 115 7.71 1.93 195 1.84 2.05
35 4.09 3.31 120 5.79 2.89 199 7.58 1.82
37 2.52 1.62 121 6.32 4.64 200 7.53 1.99
38 1.18 2.55 122 6.32 4.25 201 7.32 2.39
39 7.51 1.96 123 7.5 1.92 202 6.09 5.11
40 7.45 1.99 124 8.03 1.75 203 6.22 4.43
42 7.58 2.15 128 7.76 1.85 204 6.25 4.88
43 5.32 6.25 129 9.78 3.52 205 6.35 5.51
44 7.58 1.96 132 7.54 2.28 207 7.77 1.91
45 7.64 2.13 133 6.38 4.15 208 7.68 1.98
48 7.38 1.97 134 7.5 2.67 209 7.94 2.55
(continued)
Table 1 (continued)
Site no. f p (Hz) H/V Site no. f p (Hz) H/V Site no. f p (Hz) H/V
50 7.38 1.91 135 6.54 4.49 211 8.19 2.31
52 7.45 2.04 136 7.58 1.92 217 5.32 5.48
53 7.71 2.26 137 7.33 2.39 218 1.86 1.92
54 8.03 1.75 138 7.33 2.11 220 2.21 2.16
56 7.71 1.61 140 0.85 2.29 221 1.26 2.98
57 9.79 3.52 142 1.19 2.86 222 1.28 2.58
58 7.64 1.74 143 5.57 3.71 223 1.24 1.67
60 1.41 2.12 144 1.28 2.59 227 1.14 2.21
62 1.12 1.18 147 5.35 6.29 228 1.18 2.55
63 1.25 2.11 148 6.31 5.51 231 1.02 2.41
66 1.28 2.58 150 5.52 6.11 232 5.55 5.31
68 0.9 4.71 151 5.31 7.22 234 1.15 1.68
71 0.67 2.26 153 6.21 5.34 237 6.05 6.02
72 0.94 2.25 154 6.21 4.78 238 1.25 2.11
Table 2 Parametric count taken under consideration while doing field survey in Jammu region (JR)
Construction No. of Functionality No. of No. of floors No. of
year buildings buildings (N f ) buildings
Before 1920 37 Public 46 2 105
1920–1950 42 Residential 185 3 62
1950–1980 43 No information 11 4 31
1980–2005 48 5 29
After 2005 72 5+ 15
with Ladakh, is extremely harsh and attracts less population. Data for construction
year, functionality, and structure height for buildings taken under study are major key
components selected while doing the field survey. Details for all these components
are enlisted in Table 2.
4 Evaluation of Empirical Relationship Between Vibration

Period (T f ) and Building Height (H)
The building scenario data collected during the field survey was used to evaluate the
empirical relationship between build height (H) and vibration period (T f ) using the
regression analysis (Fig. 4). The inverse of vibration period gives the fundamental
frequency (f f ) of selected RC buildings in the study area. The fundamental frequency

(f f ) of important building structures (Fig. 5) is mentioned in Table 3.
Fig. 4 Variation of vibration period with respect to the building height for selected site locations
during the field survey
Fig. 5 Buildings in different districts of the Jammu region (JR) highlight the structural integrity
and built environment
Table 3 Fundamental frequency (f f ) of important Reinforced Concrete (RC) buildings for public
domain usage in the Jammu region (JR)
Building name Address District No. of floors Building height Fundamental
(H) in m frequency (ff)
in Hz
Government Chak Bhalwal Jammu 3 13.71 2.98
College of
Engineering
and Technology
(GCET)
(32.8508°N,
74.7787°E)
Euro Kids Balour Samba 3 11.88 3.44
School
(32.5627°N,
75.0991°E)
University of Janglote Road Kathua 4 18.28 2.23
Jammu (Kathua
Campus)
(32.4161°N,
75.5190°E)
Hotel Patnitop Padora enclave, Udhampur 3 12.78 3.19
Heights Patnitop
(33.0847°N,
75.3354°E)
Jamia Masjid Madina Colony Rajouri 2 8.52 4.79
(33.3768°N,
74.3102°E)
Hazrat Allama Nai Basti Poonch 4 15.84 2.58
Anwar Shah
Kashmiri
Islamic
International
Academy
(33.7678°N,
74.0956°E)
Shri Mata Katra Reasi 2 8.52 4.79
Vaishno Devi
Narayana
Superspeciality
Hospital
(33.0308°N,
74.9490°E)
(continued)
Table 3 (continued)
Building name Address District No. of floors Building height Fundamental
(H) in m frequency (ff)
in Hz
Government Khari Ramban 3 13.71 2.98
High School
(GHS)
(33.3831°N,
75.1560°E)
Government Near Doda Bus Doda 5 22.85 1.79
Medical stand
College (GMC)
(33.1415°N,
75.5831°E)
Dargah Shah Chowgan Kishtwar 2 7.92 5.15
Asrar-ud-Din
Baghdadi
(33.3229°N,
75.7593°E)
5 Double Resonance Effect (DRE) in Jammu Region (JR)
A building oscillates during an earthquake and may sustain significant damage if

the fundamental period of oscillation is comparable to the predominant frequency
(f p ) of the foundation soil on which it is built [19–22]. The “double resonance
effect” is an untoward coincidence that can induce collapse even under minor seismic
stimulation. In the present study, the predominant frequency (f p ) of foundation soil
is cross-checked with the fundamental frequency (f f ) of the RC buildings in a way to
assess the double resonance effect (DRE) [14, 23]. The analysis results showed that
more than 50% area falls under the high to moderate level risk zone for the double
resonance effect (DRE). For the southwestern towns of Jammu and Kathua and the
maximum portion of Samba, complete overlapping of predominant frequency (f p )
of foundation soil with the fundamental frequency (f f ) of the RC buildings were
observed (Fig. 6). These buildings show a high risk of DRE and are located in the
Indo-Gangetic plain surrounded by young alluvial deposits. The sites located on
the right bank of Tawi River specifically in the northern end, Udhampur, Rajouri,
and a few sites in Kathua adjacent to Ravi River exhibit an intermediate level of
overlapping. A total of 129 sites out of 242 reveal no overlapping and show low
risk of DRE (Table 4). In Reasi, Poonch, Doda and Kishtwar, chances of DRE are
extremely low and can be declared a safe zone with no chances of DRE.
Fig. 6 Integrated map dividing the Jammu region (JR) into high, moderate, and low-level risk
zones for double resonance effect (DRE) based on predominant frequency (f p ) of foundation soil
and fundamental frequency (f f ) of Reinforced Concrete (RC) building structures
Table 4 Zonation of sites to

Chances of No. of buildings % of total sites Risk zone
highlight Double Resonance
overlapping
Effect (DRE) during
earthquake events in the Complete 42 17.35 High
Jammu region (JR) Intermediate 71 29.33 Moderate
No 129 53.30 Low
6 Conclusion
Jammu and Kashmir is located in the northwestern Himalayas, one of the world’s
most seismically active region, with major earthquakes occurring in 1555, 1828,
1885, 1905, and 2005. In the present study, the resonance frequency (f 0 ) of foundation
soil is cross-checked with the fundamental frequency (f f ) of the Reinforced Concrete
(RC) buildings in a way to assess the double resonance effect (DRE). More than
50% area falls under the high to moderate level risk zone for the DRE. For the
southwestern towns of Jammu and Kathua and the maximum portion of Samba,
complete overlapping of f 0 of foundation soil with the f f of the RC buildings was
observed. These buildings show a high risk of DRE and are located in the Indo-
Gangetic plains surrounded by young alluvial deposits. The sites located on the right
bank of Tawi River specifically in the northern end, Udhampur, Rajouri, and a few
sites in Kathua adjacent to Ravi River exhibit an intermediate level of overlapping.
A total of 129 sites out of 242 reveal no overlapping and show a low risk of DRE.
In Reasi, Poonch, Doda, and Kishtwar, chances of DRE are extremely low and can
be declared a safe zone with no chances of DRE. These results suggest that soil-
structure interaction plays a significant role in expected damage scenarios and must
be considered while doing seismic microzonation for urban areas.
Acknowledgements Tromino used for geophysical testing is provided by the Geotechnical Divi-
sion of National Institute of Technology Srinagar, Jammu and Kashmir. The authors thank for the
technical and logistical assistance offered by the National Institute of Technology Srinagar. Abdullah
Ansari is thankful to Prof. GM Bhat (University of Jammu, Jammu), Dr. Sanjeev Gupta (GCET,
Jammu), and Tanzeel Ur Riyaz (RIMT University, Punjab) for their support during data acquisition.
The authors are also grateful to the Divisional Commissioner Office of Jammu and Kashmir for
granting special permission for fieldwork in Jammu and Kashmir during the COVID-19 pandemic.
References
1. Zahoor, F., Ansari, A., Rao, K. S., Jain, A. K.: Seismic hazard assessment of Kashmir region
using logic tree approach: Focus on sensitivity of PSHA results on seismicity declustering and
attenuation relationships. Pure Appl. Geophys. 180(1), 1–29 (2023). https://doi.org/10.1007/
s00024-023-03239-5
2. Mahajan, A.K., Thakur, V.C., Sharma, M.L., Chauhan, M.: Probabilistic seismic hazard map
of NW Himalaya and its adjoining area, India. Nat. Hazards 53(3), 443–457 (2010)
3. Nazari, Y.R., Saatcioglu, M.: Seismic vulnerability assessment of concrete shear wall buildings
through fragility analysis. J. Build. Eng. 12, 202–209 (2017)
4. Ansari, A., Rao, K. S., Jain, A. K.: Seismic vulnerability of tunnels in Jammu and Kashmir
during post-seismic functionality. Geotech. Geolog. Eng. 40(11), 1–26 (2022). https://doi.org/
10.1007/s10706-022-02341-0
5. Ameen, A. M. M. A.: Modeling of 2004 Andaman-Sumatra Tsunami and historical tsunamis
from Andaman and Nicobar Islands: Towards Estimation of Tsunami Hazard along the
adjoining areas of Indian Ocean. Master’s Thesis, Indian Institute of Technology Kanpur,
India (2018)
6. Putti, S.P., Satyam, N.: Ground response analysis and liquefaction hazard assessment for
Vishakhapatnam city. Innovative Infrastruct. Solutions 3(1), 1–14 (2018)
7. Rao, K.S., Rathod, G.W.: Seismic microzonation of Indian megacities: a case study of NCR
Delhi. Indian Geotech. J. 44(2), 132–148 (2014)
8. Ansari, A., Zaray, A. H., Rao, K. S., Jain, A. K., Hashmat, P. A., Ikram M. K., Wahidi, A. W.:
Reconnaissance surveys after June 2022 Khost earthquake in Afghanistan: Implication towards
seismic vulnerability assessment for future design. Innov. Infrastruct. Sol. 8(3), 1–15 (2023).
https://doi.org/10.1007/s41062-023-01077-x
9. Satyam, D.N., Rao, K.S.: Liquefaction hazard assessment using SPT and VS for two cities in
India. Indian Geotech. J. 44(4), 468–479 (2014)
10. Ansari, A., Rao, K.S., Jain, A.K., Ansari, A.: Deep learning model for predicting tunnel damages
and track serviceability under seismic environment. Model. Earth Syst. Environ. 9, 1349–1368
(2022). https://doi.org/10.1007/s40808-022-01556-7
11. Ansari, A., Zahoor, F., Rao, K.S., Jain, A.K.: Deterministic approach for seismic hazard assess-
ment of Jammu region, Jammu and Kashmir. In: Geo-Congress 2022: Geophysical and Earth-
quake Engineering and Soil Dynamics, GSP 334, pp. 590–598 (2022). https://doi.org/10.1061/
9780784484043.057
12. Zhong, Z., Shen, Y., Zhao, M., Li, L., Du, X., Hao, H.: Seismic fragility assessment of the
Daikai subway station in layered soil. Soil Dyn. Earthq. Eng. 132, 106044 (2020). https://doi.
org/10.1016/j.soildyn.2020.106044
13. Ansari, A., Zahoor, F., Rao, K. S., Jain, A. K.: Seismic response and vulnerability evaluation
of Jammu Region (Jammu and Kashmir). Indian Geotech. J. 52(6), 1–14 (2022). https://doi.
org/10.1007/s40098-022-00694-0
14. Lister, G., Kennett, B., Richards, S., Forster, M.: Boudinage of a stretching slablet implicated
in earthquakes beneath the Hindu Kush. Nat. Geosci. 1(3), 196–201 (2008)
15. Nogoshi, M., Igarashi, T.: On the amplitude characteristics of microtremor, Part 2. J. Seismol.
Soc. Jpn 24, 26–40 (1971)
16. Nakamura, Y.: A method for dynamic characteristics estimation of subsurface using
microtremor on the ground surface. Railway Tech. Res. Inst. Q. Rep. 30(1), 25–33 (1989)
17. Bhandary, N.P., Paudyal, Y.R., Okamura, M.: Resonance effect on shaking of tall buildings in
Kathmandu Valley during the 2015 Gorkha earthquake in Nepal. Environ. Earth Sci. 80(13),
1–16 (2021)
18. Micromed s.p.a.: The short Tromino® how to, Ver. 1.1, pp. 1–26 (2009)
19. Adhikari, A., Rao, K.R.M., Gautam, D., Chaulagain, H.: Seismic vulnerability and retrofitting
scheme for low-to-medium rise reinforced concrete buildings in Nepal. J. Build. Eng. 21,
186–199 (2019)
20. Alam, M.N., Alam, M.S., Tesfamariam, S.: Buildings’ seismic vulnerability assessment
methods: a comparative study. Nat. Hazards 62(2), 405–424 (2012)
21. Ansari, A., Rao, K.S., Jain, A.K.: Damage analysis of seismic response of shallow tunnels in
Jammu. In: Das, B.B., Hettiarachchi, H., Sahu, P.K., Nanda, S. (eds.) Recent Developments
in Sustainable Infrastructure (ICRDSI-2020)—GEO-TRA-ENV-WRM. Lecture Notes in Civil
Engineering, vol. 207, pp. 611–619. Springer, Singapore (2022). https://doi.org/10.1007/978-
981-16-7509-6_47
22. Ameen, A. M. M. A.: Seismic Hazard Evaluation of Jammu Region and Risk Assessment of
Tunnels in the Himalayas. Ph.D. Thesis, Indian Institute of Technology Delhi, India (2023)
23. Ansari, A., Satake K., Malik, J. N.: Modelling the 2004 Indian Ocean Tsunami to estimate
tsunami heights and its amplitude and to study its effects on coastal areas. Proceedings of the
ERI Earthquake Conference, Tokyo, Japan (2017)
Influence of Epistemic Uncertainty
on the Seismic Vulnerability of Indian
Code-Compliant RC Frame Building
Kaushik Gondaliya , Vishisht Bhaiya, Sandip Vasanwala, Atul Desai,

and Jignesh Amin
Abstract The study’s primary goal is to examine the presence of epistemic uncer-
tainty in the RC frame when subjected to lateral force. The seismic response of
the RC frame and the impact of uncertainties, namely material (γc , f c , and f y ) and
geometrical (Db and bc ) nonlinearity, on the structural elements are determined using
nonlinear static analysis. A fragility curve for a 4-storey RC frame structure designed
using the most recent IS 1893 for seismic zone-V was created using a simulation-
based method. The traditional deterministic technique was also utilised to compare
the effect of the structure’s lateral response uncertainty. The classifications for the
various damage levels, including slight, moderate, severe, and complete, were estab-
lished. The random variables are the strength properties of the concrete (f ck ) and
steel (f y ) materials, the depth of the beam (Db ), the width of the column (bc ), and the
weight density of concrete (γc ). The Monte Carlo technique is used to determine the
structural randomness of the RC frames. The gathered data demonstrated a signifi-
cant increase in the fragility uncertainty and structural responsiveness compared to
the deterministic technique. A fundamental conclusion of this study is the critical
relevance of tackling issues about the building’s fragility and predicted damage state
from a stochastic perspective.
K. Gondaliya (B) · V. Bhaiya · S. Vasanwala · A. Desai

Department of Civil Engineering, Sardar Vallabhbhai National Institute of Technology, Surat,
Gujarat 395007, India
e-mail: d19am007@amd.svnit.ac.in
V. Bhaiya
e-mail: Vishisht@amd.svnit.ac.in
S. Vasanwala
e-mail: sav@amd.svnit.ac.in
A. Desai
e-mail: akd@amd.svnit.ac.in
J. Amin
Department of Civil Engineering, Graduate School of Engineering and Technology, Gujarat
Technological University, Chandkheda, Gujarat 382424, India
e-mail: prof_jignesh@gtu.edu.in
https://doi.org/10.1007/978-981-99-1459-3_11
128 K. Gondaliya et al.
Keywords RC frame · Epistemic uncertainty · Seismic vulnerability · Pushover

analysis
1 Introduction
With advancements in Earthquake engineering and computing application, it is

possible to conduct multiple numbers of analyses to check the wide range of lateral
response possibilities. Two forms of uncertainty are involved in determining the
fragility curves for every selected RC frame structure [1]. Aleatory uncertainty is
the natural phenomenon presenting the randomness in the ground motion in terms
of direction, intensity, magnitude, and distance from the site location. The aleatory
form of uncertainty is difficult to reduce. However, knowledge of the structure’s mate-
rial and geometry nonlinearity can reduce the epistemic type. Epistemic uncertainty
occurred due to the assumption of material strength and geometry nonlinearity in the
analysis [2–5]. It has been observed that epistemic uncertainty can highly influence
the RC frame building’s lateral stiffness and nonlinear response.
Incomplete scientific knowledge is the source of epistemic uncertainty, which
may be addressed in principle by further evidence or more sophisticated scien-
tific concepts. Many researchers have performed stochastic analysis approaches to
minimise epistemic uncertainty [6–8].
Seismic fragility analysis is highly influenced by the consideration of epistemic
uncertainty during nonlinear static analysis [9, 10]. It has been observed from past
literature that various statistical approach is available to conduct the seismic fragility
analysis. Capacity-spectrum based analytical approach is the most efficient method
for performing the fragility analysis [11]. It is frequently assumed that the following
lognormal probability density function can adequately describe the fragility curve:

1 Sd
ρk (Sd ) = Φ ln (1)
βk Sd,ds
where S d represents the spectral displacement, S d, ds represents the median value

of spectral displacement, and S d represents the threshold limit for a given damage
state, which depicts the point at which the probability of equaling or surpassing the
damage state is 50%. β k is the standard deviation of the natural logarithm of S d for
a given damage state, which provides the notion of scattering, and Φ corresponds to
the standard normal cumulative distribution function.
This study focuses on determining the seismic vulnerability assessment (SVA) of
the 4-storey RC frame building by employing deterministic and stochastic analyt-
ical techniques. Before using nonlinear static analysis, a Monte Carlo technique is
performed [12]. Utilising the capacity spectrum approach, the seismic vulnerability
of the selected RC frame building is determined.
Influence of Epistemic Uncertainty on the Seismic Vulnerability … 129
2 Description of the Study RC Frame
IS 456:2000, IS 1893:2016, and IS 13920:2016 serve as a baseline of the 4-storey RC

special moment-resisting frame building structure [13–15] (see Fig. 1). According
to IS 875:1987 (Part-1) and IS 875:1987 (Part-2) [16, 17], the gravity loads are
estimated to be 1 kN/m2 for dead load and 3 kN/m2 for live load. The internal and
external masonry walls are each given a thickness of 115 and 230 mm, respectively,
while the slab is assigned a thickness of 150 mm. The lateral rigidity of infill walls
is not taken into account.
2.1 Material and Geometry Nonlinearity
Spatial Variation. Random variables are used to determine the probabilistic seismic
vulnerability of structures. The random variables that explicitly affect the epistemic
uncertainty of the response were considered in the present study. Table 1 shows the
selected five random variables for the stochastic analysis of the 4-storey RC frame
building. Here, the characteristic strength of concrete (f c ) is a random variable, and
√ (E c ) is an explicitly random variable. E c is considered
the elastic modulus of concrete
by the equation E c = 5000 f ck . While young’s modulus (E s ) of the steel was taken
at the standard rate 2 × 105 N/mm2 [13] guidelines, and the only random variable
was the yield strength value of the rebar Fe500.
Four groups, one for each floor, make up each floor’s structural component on the
building’s floor. The total correlation matrix is employed to consider the correlation
between the samples [18]. The sample created for the column or beam belonging to
Fig. 1 a–b Geometry configuration of the selected 4-storey RC frame building with rebar detailing
of c beam and d column
Table 1 Details of selected random variables for the simulation

Variable Mean SDa CVb Distribution
fc 25 MPa 3.11 0.12 Normal
fy 500 MPa 17.26 0.04 Lognormalc
γc 25 kg/m3 2.50 0.10 Normal
Db 450 mm 9.38 0.02 Normal
bc 350 mm 7.89 0.02 Normal
a SD = Standard deviation
b Coefficient of variation
c Lognormal random number generated [19]
the same group is thought to have a strong correlation, whereas the sample of different
groups is believed not to correlate. It also accounts for a certain degree of spatial
fluctuation of unpredictability. For each iteration of the Monte Carlo algorithms, a
random trial of variable rate is created for individual elements of every group [7–9].
2.2 SVA of Selected 4-Storey RC Frame
Capacity-Spectrum Method (CSM). SVA of the RC frame building is derived using

the CSM. The nonlinear static analysis is commonly referred to as the pushover
analysis (PA). Probabilistic analysis is adopted based on the Monte Carlo approach
using a random variable of the selected parameters for the RC frame structures. The
choice of load distribution is essential when assessing the seismic performance of a
structure using PA since it determines how an earthquake’s inertia force will affect a
building. The equation calculates the load distribution of seismic force at each floor
(Qi ) [14].
Wi h i2
Qi = Vd (2)
Σ
n
Wi h i2
i=1
where Qi represents the lateral load at the ith floor, lateral loads generate flexural
deformations near element borders; hence, flexural hinges are provided to member
ends throughout the modelling phase. While hinges M3 accommodating flexure
are provided for the beams, hinges P-M2 -M3 accommodating axial and bending
moments are specified for the columns. Additionally, the pushover analysis considers
the P-delta effect of evaluating the actual reaction of a structure. In beam/column
elements, the relative locations of the hinges are established by default SAP2000
software provides [20]. Figure 2 shows the derived PA curves per the deterministic
and stochastic approach for the selected 4-storey RC frame building.
Fig. 2 Comparison of the

pushover analysis as per the
deterministic and
probabilistic approach
In the present work, seismic fragility evaluations are conducted by establishing

damage states for each frame and assessing the likelihood of exceeding each damage
state by HAZUS criteria. Consideration is given to four damage states: slight,
moderate, severe, and complete. The damage state threshold is shown in Fig. 3
to be determined from the bilinear capacity spectrum.
Uncertainty (β k ) is calculated in the present study by applying the approximation
of normal Assuming that when S d is equal to Sdk , the probability of equaling or
exceeding the damage state is 50%, the probabilities of the other damage states are
estimated from the corresponding binomial probability distribution [21–23]. Seismic
fragility analysis is carried out to find the most probable likelihood of damage under
Fig. 3 Damage threshold of capacity spectrum and fragility curve fitting sample of a 4-storey RC
frame structure
seismic excitation. The most likely damage condition of a building can be expressed
by the weighted mean damage index matrix, DSm :
Σ
4
DSm = k pk [N , d] (3)
i=1
where pk [N, d] is the likelihood that a particular damage state will occur and k is
one of the following values: 0, 1, 2, 3, or 4, a zero value of d implies non-damage
to a structure, whereas d = 1 denotes destruction. The seismic demand specification
calculates the chance of exceeding (Pk ). It is accomplished via a fragility graph by
determining the likelihood of exceeding a spectral displacement (S d ) appropriate to
the performance point (S d, ds ). Consequently, the probability of occurrence (pk ) of
discrete damage grade is derived as shown in Eq. 4.
ρk = Pk (k + 1) − Pk (k) (4)
Furthermore, the mean damage condition of the structure is determined using the
probability of occurrence (ρk ). Figure 4 shows the fragility curves produced from
the deterministic and stochastic technique for slight, moderate, severe, and complete
damage states.
Fig. 4 Fragility graph is derived for material and geometry nonlinearity using the various damage
states of the 4-storey RC frame building
Fig. 5 Comparison of the fragility curve derived using the a deterministic and b stochastic approach
of the selected RC frame building
This part discusses the most significant outputs of the stochastic viewpoint. The
following are the most important findings of the present study of the 4-storey RC
frame.
• Comparing the model analysed using a deterministic methodology to a compa-
rable model analysed using a stochastic method, the deterministic technique
reveals a 33.38% decrease in the elastic behaviour.
• The stochastic analysis of the proposed building indicated considerable changes
in the nonlinear response of the structure under seismic force, which considerably
impacted the failure mechanism of structural elements.
• In a deterministic analysis, an RC frame construction sustains minimal damage.
However, moderate damage is observed in a stochastic analytic technique.
Figure 5 shows the fragility graph generated per the deterministic and stochastic
analysis technique for the sample study building. As the performance point values
grow, it was noted that RC frame building evaluation using a deterministic approach
produces a conservative estimate of likelihood.
4 Conclusion
The 4-storey RC frame building is analysed using the five random variables based on
the deterministic and stochastic approach. A stochastic PA technique reveals up to a
10–13% increment in the likelihood of exceedances for complete damage states in
the 4-storey RC frame structure. Here, the lateral stiffness of the masonry infill wall
is not addressed, which could result in a more realistic behaviour of the RC frame
building under lateral force. This study can be utilised to build a machine learning
model to forecast the mean damage state.
References
1. Bulleit, W.M.: Uncertainty in structural engineering. Pract. Period. Struct. Des. Constr. 13(1),
24–30 (2008)
2. Iervolino, I.: Estimation uncertainty for some common seismic fragility curve fitting methods.
Soil Dyn. Earthquake Eng. 152, 107068 (2022)
3. Choudhury, T., Kaushik, H.B.: Seismic response sensitivity to uncertain variables in RC frames
with infill walls. J. Struct. Eng. 144(10), 04018184 (2018)
4. Bai, J.W., Hueste, M.B.D. and Gardoni, P.: Probabilistic assessment of structural seismic
damage for buildings in mid-America. In: AIP Conference Proceedings, Vol. 1020, No. 1,
pp. 1685–1692. American Institute of Physics (2008)
5. Cremen, G., Baker, J.W.: Variance-based sensitivity analyses and uncertainty quantification for
FEMA P-58 consequence predictions. Earthquake Eng. Struct. Dyn. 50(3), 811–830 (2021)
6. Gondaliya, K.M., et al.: Seismic vulnerability assessment of Indian code compliant RC frame
buildings. J. Vibration Eng. Tech. 1(25), (2022)
7. Gondaliya, K., Bhaiya, V., Vasanwala, S., Desai, A.: Probabilistic seismic vulnerability of
Indian code-compliant RC frame. Pract. Period. Struct. Des. Constr. 27(3), 04022028 (2022)
8. Alzate, V., Felipe, Y., et al.: An efficient methodology to estimate probabilistic seismic damage
curves. J. Struct. Engi. (New York, NY) 145(4), 04019010–04019011 (2019)
9. Vargas-Alzate, Y.F., et al.: Seismic risk assessment using stochastic nonlinear models. Sustain.
12(4), 1308 (2020)
10. Rosowsky, D.V., Stewart, M.G.: Probabilistic construction load model for multistory
reinforced-concrete buildings. J. Perf. Const. Facil. 15(4), 145–152 (2001)
11. Barbat, A.H., Pujades, L.G., Lantada, N.: Performance of buildings under earthquakes in
Barcelona, Spain. Computer-Aided Civil Infra. Eng. 21(8), 573–593 (2006)
12. Hurtado, J.E., Barbat, A.: Monte Carlo techniques in computational stochastic mechanics.
Archiv. Comp. Methods Eng. 5(1), 3–29 (1998)
13. BIS IS 456.: Plain and Reinforced Concrete-code of practice. Bureau of Indian Standards, New
Delhi (2000)
14. BIS IS 1893.: Criteria for earthquake resistant design of structures, Part-1 General Provisions
and Buildings. Bureau of Indian Standards, New Delhi (2016)
15. BIS IS 13920.: Ductile detailing of reinforced Concrete-code of practice. Bureau of Indian
Standards, New Delhi, India (2016)
16. BIS 875 Part 1.: Design loads (other than earthquake) For buildings and structure. Dead loads—
code of practice. Bureau of Indian Standards, New Delhi (1987)
17. BIS 875 Part 2.: Design loads (other than earthquake) for building and structure. Imposed
loads—codes of practice. Bureau of Indian Standards, New Delhi (1987)
18. Alzate, V.Y.F., et al.: Probabilistic seismic damage assessment of reinforced concrete buildings
considering directionality effects. Struct. Infrastruct. Eng. 14(6), 817–829 (2018)
19. WESSA Homepage: http://www.wessa.net/, last accessed 2022/02/28.
20. Structural Analysis Software (SAP2000), Advance, static and dynamic finite element analysis
of structures, Computer and structures, Inc., Berkeley, CA (2015)
21. Lantada, N., Pujades, L.G., Barbat, A.H.: Vulnerability index and capacity spectrum-based
methods for urban seismic risk evaluation: a comparison. Nat. Hazards 51(3), 501 (2009)
22. Barbat, A.H., Pujades, L.G., Lantada, N.: Seismic damage evaluation in urban areas using
the capacity spectrum method: application to Barcelona. Soil Dyn. Earthq. Eng. 28(10–11),
851–865 (2008)
23. Manafpour, A.R., Moghaddam, P.K.: Probabilistic approach to performance-based seismic
design of RC frames. In: Second International Conference on Vulnerability and Risk Anal-
ysis and Management (ICVRAM) and the Sixth International Symposium on Uncertainty,
Modeling, and Analysis (ISUMA), Vulnerability, uncertainty, and risk: Quantification, mitiga-
tion, and management, 1736–1745. American Society of Civil Engineering (ASCE), Liverpool,
UK (2014)
Spatial Distribution
of the Gutenberg-Richter Parameters
and Fractal Dimension and Their
Correlations in Northeast India and Its
Vicinity
R. B. S. Yadav, P. Chauhan, M. Sandhu, R. Kumar, and V. Singh
Abstract The study aims to estimate the Gutenberg-Richter parameters (a and b)

and fractal dimension (Dc) using the maximum likelihood estimation (MLE) method
in 18 shallow (≤70 km) and 5 intermediate (>70 km) depth seismic zones in northeast
India and its vicinity. Scaling relations have been developed among the estimated
hazard parameters. A unified and comprehensive earthquake catalogue spanning the
period from 1897 to 2016 is used for the purpose. The regions associated with the
low b-value and high Dc value have been considered the utmost potential regions
for the incidence of big events in the examined area. The b-values in the examined
region vary from 0.59 to 1.31 in shallow zones and from 0.88 to 0.98 in intermediate
zones. Similarly, the Dc values vary from 1.81 to 2.65 in shallow zones and 2.22
to 2.71 in intermediate zones. The low b-values (less than 0.9) and the high Dc
values (greater than 2.0) are related to shallow zones 3 (Arunachal Himalaya), 6
(Eastern Himalayan syntaxis), 13 (Burmese region), 17 (south of Shillong Plateau)
and all intermediate zones in Indo-Burmese regions. This makes these zones the
most vulnerable to high earthquake hazards. The associations between Dc and b,
and Dc and a/b illustrate a positive and negative correlation, respectively. The spatial
variations of these parameters can be used as important parameters of earthquake
hazard levels in the region.
Keywords b-value · Fractal dimension · Earthquake hazard · Northeast India
R. B. S. Yadav (B) · P. Chauhan · M. Sandhu · V. Singh

Department of Geophysics, Kurukshetra University, Kurukshetra, India
e-mail: rbsykuk@gmail.com
R. Kumar
National Centre for Seismology, New Delhi, India
https://doi.org/10.1007/978-981-99-1459-3_12
138 R. B. S. Yadav et al.
1 Introduction
Northeast India and its vicinity are the utmost peculiar seismic dynamic region in the
Indian subcontinent, whose seismotectonic is mainly governed by the interaction of
three tectonic plates, viz. the Indian, Eurasian and Burmese plates. It has generated
the two deadliest earthquakes in 1897 (Mw 8.1), named Shillong earthquake and
1950 (Mw 8.6), named the Assam earthquake. The Bureau of Indian Standards,
2002 [1] demarcated four seismic zones (II, III, IV and V) in the Indian region with
the equivalent zonal factors labelled as 0.10, 0.16, 0.24 and 0.36 g, respectively that
can unleash an earthquake of magnitudes exceeding the values of 5, 6, 7 and 8,
respectively [2–4]. The study region is situated in zone V (highest hazardous zone),
which is capable to generate an earthquake of magnitude 8.0 or more. This region can
experience peak ground acceleration (PGA) in the range of 0.35–0.40 g for 475 years
of return periods [5].
A number of statistical studies have been performed by different researchers in
the study region, which include estimation of earthquake hazard indicators, return
periods and occurrence probabilities [3, 4, 6–11, etc.]. These studies are based on
different statistical and physical models and power-law relations. Two very important
earthquake hazard parameters, the b-value of Gutenberg-Richter recurrence relation
[12] and fractal dimension (Dc ), have been considered to evaluate the level of earth-
quake potential in many regions. The b-value is indicative of the comparative share
between the number of small and big events in a certain area. A high b-value speci-
fies a growth in the volume of small events and a reduction in the volume of strong
events and vice versa. Dc value is a scale-invariant parameter, and it is represented as
a well-organized statistical factor to assess the dimensional dispersal of seismicity.
The grade of heterogeneity in a region is generally quantified by the measure of
fractal dimension. It is governed by the distribution of the differential stress along
with the prevailing geological and structural heterogeneity [13].
In the present study, a quantitative approach is adopted to characterize the seismic
activity of northeast India in terms of the G-R frequency-magnitude recurrence
relation (a- and b-values) and fractal dimension (Dc ). The spatial distribution of
these parameters in the study region is also correlated with seismic complexity and
tectonics to identify potential regions.
2 Seismotectonics of Study Region
The examined area (northeast India and adjacent regions) lies between 20° and 32° N
latitude and 87° and 100° E longitude. These areas consist of the Himalayan thrusts
(Main Boundary Thrust, MBT and Main Central Thrust, MCT), eastern syntaxis,
Arakan-Yoma subduction belt, Shillong Plateau and Bengal basin (Fig. 1). Due to the
interaction of Eurasian and Indian plates, the east–west trending Himalayan system
has been developed, while the subduction of the Indian plate with the Burmese plate
Spatial Distribution of the Gutenberg-Richter Parameters and Fractal … 139
generated the north–south trending Arakan-Yoma subduction belt. The part of the
Himalayan system in northeast India consists of a number of north-oriented thrust
faults, which include mainly MBT, MCT and Himalayan Frontal thrust. The Arakan-
Yoma subduction area is consisting southeast heading thrusts (e.g. Naga and Disang
Thrusts). The pop-up tectonic process has generated Shillong Plateau (SP), which
has peculiar tectonics. The eastern syntaxis of the Himalayan zone comprises mainly
Lohit and Mishmi Thrusts.
The northeast Himalaya is a highly seismic active area in the Himalayan system,
and its earthquake activity is primarily governed by the interaction of the Indian
plate with the Eurasian and Burmese plates [14]. This region has generated two
catastrophic events of Mw ≥ 8.0. June 12, 1897, (Mw 8.1) earthquake occurred on
SSE dipping Oldham reverse fault in the Shillong Plateau. The August 15, 1950,
Fig. 1 Tectonic map of northeast India and adjacent region depicting main faults and lineaments.
Two great earthquakes occurred in 1897 (Shillong) and 1950 (Assam) earthquakes are shown along
with the 2016 Manipur earthquake (res star)
Fig. 2 Map depicts the seismicity of the study region. The intermediate to deep depth seismicity
is observed in the Arakan-Yoma subduction zone, which is the collisional boundary of the Indian
ad Burmese plates
Assam earthquake (the sixth largest earthquake of the twentieth century) has an
instrumentally determined magnitude of 8.6 that occurred approximately the India-
China border. The entire region is characterized by shallow depth earthquakes except
for the Arakan-Yoma subduction zone, which is characterized by the intermediate to
depth earthquake (Fig. 2).
3 Earthquake Data Used and Seismic Zonation
The earthquake catalogue compiled by [2] for the period from 1897 to 2007 has
been updated up to 2016. The initial catalogue of [2] has been prepared with the
Fig. 3 Delineation of shallow depth (a) and intermediate depth (b) source zones
help of various published literature and earthquake catalogues provided by national

and international seismological agencies viz. the India Meteorological Department
(IMD), India; International Seismological Centre (ISC), National Earthquake Infor-
mation Centre of the United States Geological Survey (USGS) and Global Centroid
Moment Tensor Catalogue (GCMT). For the period between 2008 and 2016, the
catalogue is updated with the help of ISC data. A full description of compilation,
unification of magnitude scales into moment magnitude (Mw), declustering of depen-
dent events (foreshocks and aftershocks) and completeness analysis with magnitude
and time of catalogue is given by [2]. The updated catalogue contains 3650 events
during the period from 1997 to 2016.
The ultimate demarcation of seismic source zones needs comprehensive infor-
mation on historical and modern seismicity, tectonics, geology and neo-tectonic
activities of the examined region. Depending upon the focal depth variation, the
seismogenic source zones in the study area have been divided into two parts: 18
shallow focal depth (≤70 km) source zones (Fig. 3a) and 5 intermediate focal depth
(>70 km) source zones (Fig. 3b).
4 Methods Applied
4.1 Frequency-Magnitude Recurrence Relation
A maximum likelihood method is used to evaluate the parameters a and b of the

Guttenberg-Richter frequency-magnitude recurrence relation [12]
Log10 N = a − bM (1)
where N is the cumulative count of events with size ≥ M. The parameter b is the slope
of the frequency-size distribution, and a is the number of events ≥ 0. In most cases, a
low b-value is interpreted as the increased frequency of big events, large differential
crustal stress and decreased heterogeneity [15, 16]. Generally, its average value is 1.0,
which can range between 0.6 and 1.4 [17]. The area associated with high geological
complexity and heterogeneity in rock material or crack density, generally, shows
increased b-values demonstrating the significance of the multi-fracture system. On
contrary, a low b-value is associated with a low grade of heterogeneity, increased
stress or strain and a high strain rate.
In the present study, an MLE method proposed by [18] is applied to estimate the
b-value.
1
b= [ ( )] , (2)
log10 M − Mmin −Δm/2
where M min is the lower cut-off magnitude (magnitude of completeness, M c ), M is

the average magnitude and Δm is the bin size of magnitude.
4.2 Fractal Dimension
In the present study, Dc is calculated using a method proposed by [19], in which the
correlation integral of the distribution of N events yields the following form:
2
C(r ) = N (R < r ) (3)
N (N − 1)
where N(R < r) is the total event couples placed by a distance R less than r. The
correlation integral is related to the standard correlation function
C(r )r (Dc ) (4)
where Dc is a fractal dimension or the correlation dimension. It is calculated from the

log–log plot of C(r) versus r as a slope of the best–fitted straight line. The Dc value
decreases if the earthquake events are more clustered. If all the events are clustered
on a single point, the Dc value attains zero, while line sources provide Dc close to
1 and the earthquake fractures filled up a crustal volume providing Dc close to 3
[16, 20, 21].
The seismicity characteristics of northeast India have been evaluated in terms of

G-R parameters (a- and b-values) and fractal dimension (Dc ) using a uniform and
comprehensive earthquake catalogue spanning the period from 1897 to 2016. For
this purpose, 18 shallow and 5 intermediate depth source zones have been identified
and variations of these parameters have been correlated in it.
5.1 a- and b-values Variation
The G-R frequency-magnitude recurrence relation parameters (a-, a-annual and b-

values) have been estimated using three methods, namely the maximum curvature
method, 90% probability and Entire Magnitude Range (EMR) [22]. Among these,
the maximum curvature method gives a reliable estimate. The estimated values are
listed in Tables 1 and 2 for shallow and intermediate depth source zones, respectively.
Some examples of the frequency-magnitude distributions are shown in Fig. 4 for the
shallow and intermediate depth zones. The a-value (annual) is also called the activity
rate, i.e. the number of earthquakes per year in a given area or zone.
The computed b-value ranges between 0.59 and 1.31 in shallow zones with the
lowest in Sh-zone-4 (Himalaya Frontal Thrusts region) and the highest in Sh-zone
18 (Shillong plateau). The low b-values (<0.80) are observed in shallow zones 2, 3,
4 (Himalaya Frontal Thrusts), 6, 7, 8, 12, 13 (Arakan-Yoma subduction), 16 and 17
(Shillong Plateau). On the other hand, high b-values (>1.0) are observed in Sh-zone 1,
5 (Himalaya Frontal Thrust), 9 (Arakan-Yoma subduction) and 18 (Shillong Plateau).
For the intermediate depth, all the seismic zones show a b-value less than 1.0. A high
seismic activity rate (a-annual > 4.0 events/per year) is observed in shallow zones 1,
5, 9, 11 and 18, and intermediate depth zones 3, 4 and 5.
5.2 Fraction Dimension Variation
Fractal dimension (Dc ) has been calculated for 15 shallow zones out of 18 and all 5
zones of intermediate depth that are listed in Tables 1 and 2, respectively. Some zones
show scattered seismicity, and therefore, we were unable to calculate Dc (e.g. Sh-
zone 5, 11 and 16). Some examples of the graphs of correlation integral vs. distance
are shown in Fig. 5 for shallow and intermediate depth zones. The calculated Dc
values vary between 1.81 and 2.65 with the lowest Dc value found in Sh-zone 12 of
the Arakan-Yoma subduction belt, and the highest Dc value was observed in Sh-zone
15 of the Shillong Plateau region. Dc value less than 2.0 was estimated in Sh-zone 12
of the Arakan-Yoma subduction belt region suggesting the source is linear to 2D, and
seismicity is clustered. All other zones show a Dc value between 2.0 and 3.0, which
Table 1 G-R recurrence parameters (a- and b-values), magnitude of completeness (M c ) and fractal
dimension (Dc ) for shallow zones 1–18 in the study region
S. No Seismic Methods Mc b-value a-value a-value Fractal
zones (annual) dimension
1 Sh-zone 1 Maximum 4.8 1.10 ± 0.2 6.96 4.93 2.28 ± 0.02
curvature
Mc90 4.7 1.01 ± 0.1 6.54 4.51
EMR 4.8 1.10 ± 0.1 6.96 4.93
method
2 Sh-zone 2 Maximum 4.6 0.71 ± 0.1 4.61 2.71 2.28 ± 0.04
curvature
EMR 4.5 0.70 ± 0.1 4.60 2.70
method
3 Sh-zone 3 Maximum 4.3 0.77 ± 0.06 5.30 3.40 2.4 ± 0.01
curvature
Mc90 4.8 0.98 ± 0.1 6.36 4.46
EMR 4.3 0.77 ± 0.06 5.3 3.4
method
4 Sh-zone 4 Maximum 4.6 0.59 ± 0.1 4.07 2.14 2.20 ± 0.03
curvature
Mc90 4.1 0.56 ± 0.08 3.93 2.0
EMR 4.6 0.59 ± 0.1 4.07 2.14
method
5 Sh-zone 5 Maximum 4.8 1.01 ± 0.3 6.39 4.57
curvature
EMR 5.0 0.94 ± 0.3 6.00 4.18
method
6 Sh-zone 6 Maximum 4.4 0.74 ± 0.06 5.29 3.26 2.49 ± 0.01
curvature
Mc90 4.4 0.74 ± 0.06 5.29 3.26
EMR 4.6 0.81 ± 0.08 5.67 3.63
method
7 Sh-zone 7 Maximum 4.6 0.89 ± 0.1 5.67 3.77 2.43 ± 0.02
curvature
Mc90 4.5 0.79 ± 0.1 5.18 3.29
EMR 4.6 0.89 ± 0.1 5.67 3.77
method
8 Sh-zone 8 Maximum 4.5 0.87 ± 0.09 5.84 3.89 2.37 ± 0.02
curvature
Mc90 4.3 0.74 ± 0.06 5.22 3.26
EMR 4.8 0.88 ± 0.1 5.93 3.98
method
(continued)
Table 1 (continued)
9 Sh-zone 9 Maximum 4.5 1.05 ± 0.1 6.73 4.75 2.65 ± 0.02
curvature
Mc90 4.3 0.85 ± 0.07 5.76 3.78
EMR 4.5 1.05 ± 0.1 6.73 4.75
method
10 Sh-zone Maximum 4.5 0.90 ± 0.1 5.58 3.90 2.22 ± 0.02
10 curvature
EMR 4.7 1.06 ± 0.2 6.37 4.70
method
11 Sh-zone Maximum 4.5 1.00 ± 0.3 5.59 4.08
11 curvature
12 Sh-zone Maximum 4.4 0.65 ± 0.1 4.51 2.61 1.81 ± 0.02
12 curvature
Mc90 4.3 0.64 ± 0.09 4.45 2.54
EMR 4.3 0.64 ± 0.09 4.45 2.54
method
13 Sh-zone Maximum 4.4 0.75 ± 0.08 5.11 3.15 2.38 ± 0.01
13 curvature
EMR 4.4 0.75 ± 0.08 5.11 3.15
method
14 Sh-zone Maximum 4.5 0.89 ± 0.1 5.71 3.79 2.45 ± 0.03
14 curvature
Mc90 4.4 0.86 ± 0.1 5.51 3.59
EMR 4.6 0.87 ± 0.1 5.55 3.63
method
15 Sh-zone Maximum 4.4 0.98 ± 0.2 5.85 3.98 2.65 ± 0.03
15 curvature
Mc90 4.2 0.82 ± 0.1 5.07 3.21
EMR 4.5 0.90 ± 0.2 5.48 3.62
method
16 Sh-zone Maximum 4.2 0.66 ± 0.07 4.70 2.62
16 curvature
Mc90 4.0 0.57 ± 0.05 4.28 2.21
EMR 4.6 0.68 ± 0.1 4.82 2.75
method
17 Sh-zone Maximum 4.3 0.63 ± 0.07 4.45 2.46 2.34 ± 0.02
17 curvature
Mc90 4.2 0.63 ± 0.07 4.46 2.47
(continued)
Table 1 (continued)
EMR 4.7 0.69 ± 0.1 4.76 2.77
method
18 Sh-zone Maximum 4.8 1.31 ± 0.5 7.50 5.71 2.38 ± 0.02
18 curvature
EMR 4.6 1.10 ± 0.3 6.44 4.65
method
Table 2 G-R recurrence parameters (a- and b-values), magnitude of completeness (M c ) and fractal
dimension (Dc ) for intermediate zones 1–5 in the study region
S .No Seismic Methods Mc b-value a-value a-value Fractal
1 Int-zone 1 Maximum 4.2 0.90 ± 0.1 5.40 3.36 2.41 ± 0.02
curvature
Mc90 4.7 0.75 ± 0.9 4.70 2.66
EMR 4.1 0.81 ± 0.1 4.97 2.94
method
2 Int-zone 2 Maximum 4.5 0.92 ± 0.2 5.53 3.61 2.22 ± 0.02
curvature
Mc90 4.2 0.80 ± 0.1 4.92 3.00
EMR 4.3 0.78 ± 0.1 4.81 2.90
method
3 Int-zone 3 Maximum 4.5 0.88 ± 0.07 6.13 4.18 2.57 ± 0.01
curvature
Mc90 4.3 0.77 ± 0.5 5.57 3.63
EMR 4.5 0.88 ± 0.07 6.13 4.18
method
4 Int-zone 4 Maximum 4.4 0.98 ± 0.08 6.52 4.63 2.68 ± 0.04
curvature
Mc90 4.2 0.87 ± 0.05 5.98 4.09
EMR 4.7 1.14 ± 0.1 7.28 5.39
method
5 Int-zone 5 Maximum 4.5 0.92 ± 0.1 5.74 4.03 2.71 ± 0.02
curvature
Mc90 4.3 0.77 ± 0.09 5.00 3.28
EMR 4.6 0.87 ± 0.1 5.50 3.78
method
Fig. 4 Some examples of graphs of frequency-magnitude distribution for shallow depth zones a
Sh-zone 1, b Sh-zone 4, c Sh-zone 6, d Sh-zone 15, e Sh-zone 16 and f Sh-zone 17; and intermediate
depth zones g int-zone 1, h int-zone 2 and i int-zone 3
indicates a uniform or random spatial distribution of seismicity. For the intermediate

depth zones, the calculated Dc values lie between 2.22 and 2.71 suggesting that the
faults are spatially distributed.
Fig. 5 Some examples of graphs of fractal dimension for shallow depth zones a Sh-zone 1, b Sh-
zone 4, c Sh-zone 6, d Sh-zone 15, e Sh-zone 16 and f Sh-zone 17; and intermediate depth zones
g int-zone 1, h int-zone 2 and i int-zone 3
5.3 Correlation Between G-R Parameters and Dc Values
The ratio between b and Dc is an effective indicator of seismic hazards. We correlated

seismic b-value with Dc estimated for only shallow depth zones (Fig. 6a) and obtained
the following relation:
Dc = 0.41b + 2.0 (5)
Another relationship between a/b-values (i.e. modal values) and Dc (Fig. 6b) for
shallow depth zones depicts the following relation:
Fig. 6 Graphs show the relationship between a b versus Dc values and b a/b versus Dc value. The
straight line is the least square linear regression
Dc = −0.21(a/b) + 3.74 (6)
6 Conclusions
The following conclusions can be drawn based on the results obtained in this study:
1. A low b-value (<0.80) observed in shallow zones 2, 3, 4 (Himalaya Frontal
Thrusts), 6, 7, 8, 12, 13 (Arakan-Yoma subduction), 16 and 17 (Shillong Plateau)
shows the occurrence of a large proportion of higher magnitude earthquakes and
high differential crustal stress that can unleash in the form of large events.
2. Most of the seismic sources in shallow depth exhibit high Dc values indicating
uniform or random spatial distribution of seismicity.
3. A positive correlation is observed between Dc vserus b-value and a negative corre-
lation between Dc versus a/b-value (modal), which indicates higher Dc values
(random characteristics) are well correlated with lower magnitude earthquakes.
4. A non-uniform variation of seismicity parameters from one zone to another zone
suggests the presence of structural heterogeneity and complex tectonics in the
region.
Acknowledgements The first author is thankful to Kurukshetra University for providing financial
support in the form of a Seed Money Grant (SMG). Authors are also acknowledged the support
provided by RUSA2.0. The authors are thankful to the reviewers and their positive responses.
References
1. BIS, IS 1893–2002 (Part 1): Indian standard criteria for earthquake resistant design of structures,
part 1–general provisions and buildings. Bureau of Indian Standards, New Delhi (2002)
2. Yadav, R.B.S., Bormann, P., Rastogi, B.K., Das, M.C., Chopra, S.: A homogeneous and
complete earthquake catalog for northeast India and the adjoining region. Seism. Res. Lett.
80(4), 609–627 (2009)
3. Yadav, R.B.S., Tripathi, J.N., Rastogi, B.K., Das, M.C., Chopra, S.: Probabilistic assessment of
earthquake recurrence in Northeast India and adjoining region. Pure Appl. Geophys. 167(11),
1331–1342 (2010)
4. Yadav, R.B.S., Tsapanos, T.M., Burton, P.W., Kumar, R., Sandhu, M.: Seismicity and magnitude
recurrence hazard assessment in Eastern Nepal, Northeast India and Tibet Himalaya. Phys.
Chem. Earth, Parts A/B/C 127, Article no. 103158 (2022)
5. Bhatia, S.C., Kumar, M.R., Gupta, H.K.: A probabilistic seismic hazard map of India and
adjoining regions. Ann di Geofis 42, 1153–1166 (1999)
6. Parvez, I.A., Ram, A.: Probabilistic assessment of earthquake hazards in the north-east Indian
peninsula and Hindukush regions. Pure Appl. Geophys. 149, 731–746 (1997)
7. Shanker, D., Sharma, M.L.: Estimation of seismic hazard parameters for the Himalayas and its
vicinity from complete data files. Pure Appl. Geophys. 152, 267–279 (1998)
8. Sharma, M.L., Malik, S.: Probabilistic seismic hazard analysis and estimation of spectral strong
ground motion on bed rock in northeast India. In: 4th International Conference Earthquake
Engineering, Taipei, Taiwan (2006)
seismicity in Northeast India and its adjoining region. J. Seismol. 12, 107–123 (2008)
10. Yadav, R.B.S., Tripathi, J.N., Shanker, D., Rastogi, B.K., Das, M.C., Kumar, V.: Probabilities
for the occurrences of medium to large earthquakes in northeast India and adjoining region.
Nat. Hazards 56(1), 145–167 (2011)
11. Sandhu, M., Yadav, R.B.S., Kumar, R., Baruah, S., Singh, A.P., Mishra, M.: Spatial variability
of earthquake hazard parameters, return periods and probabilities of earthquake occurrences in
the eastern Himalayan seismic belt. Phys. Chem. Earth, Parts A/B/C, 103194 (2022). https://
doi.org/10.1016/j.pce.2022.103194
34, 185–188 (1944)
13. Öncel, A.O., Wilson, T.: Space-time correlations of seismotectonic parameters and examples
from Japan and Turkey preceding the Izmit earthquake. Bull. Seismol. Soc. Am. 92, 339–350
(2002)
14. Bilham, R.: Himalayan Earthquakes: a review of historical seismicity and early 21st century
slip potential. Geol. Soc. London Special Publ. 483, 423–482 (2019)
15. Singh, A.P., Mishra, O.P., Kumar, D., Kumar, S., Yadav, R.B.S.: Spatial variation of the after-
shock activity across the Kachchh Rift Basin and its seismotectonic implications. J. Earth Syst.
Sci. 121(2), 439–451 (2012)
16. Yadav, R.B.S., Gahalaut, V.K., Sumer, C., Shan, B.: Tectonic implications and seismicity trig-
gering during the 2008 Baluchistan, Pakistan Earthquake Sequence. J. Asian Earth Sci. 45(2),
167–178 (2012)
17. Udias, A.: Principles of seismology. Cambridge University Press, New York (1999)
18. Aki, K.: Maximum likelihood estimate of b in the formula logN = a-bM and its confidence
limits. Bull. Earthquake Res. Inst., Tokyo University 43, 237–239 (1965)
19. Grassberger, P., Procaccia, I.: Characterization of strange attractors. Phys. Rev. Lett. 50, 346–
349 (1983)
20. Yadav, R.B.S., Papadimitriou, E.E., Karakostas, V.G., Rastogi, B.K., Shanker, D., Chopra,
S., Singh, A.P., Kumar, S.: The 2007 Talala, Saurashtra, western India earthquake sequence.
Tectonic implications and seismicity triggering. J. Asian Earth Sci. 40(1), 303–314 (2011)
21. Chamoli, A., Yadav, R.B.S.: Multifractality in seismic sequences of NW Himalaya. Nat.
Hazards 77(1), S19–S32 (2015)
22. Woessner, J., Wiemer, S.: Assessing the quality of earthquake catalogues: estimating the
magnitude of completeness and its uncertainty. Bull. Seismol. Soc. Am. 95(2), 684–698 (2005)
Seismic Hazard Analysis Considering
the Effect of the Shape, Size, and EQ
Distribution of Seismic Sources
for Different Locations in Sikkim, NE
India
Borah Niranjan , Mondal Joy Kumar , and Kumar Abhishek
Abstract For a seismically active region, ensuring the safety of the structures against
future earthquakes (EQs) is very important. This seismic safety is mainly ensured
by designing the structures to withstand the probable EQ hazard that may occur in
future. This future probable EQ hazard level can be estimated by performing seismic
hazard analysis (SHA) for the region. The two types of SHA, deterministic (DSHA)
and probabilistic (PSHA), strongly depend on the presence of seismic sources near
the region, its past EQ history, and the site condition of the region. In the present
work, DSHA and PSHA are used to obtain the seismic hazard level (SHL) of different
locations in Sikkim, NE India. This is to be noted here that some faults near Sikkim
are not linear, and different segments of some faults show different seismic activity
rates. The effect of these types of shapes is considered for the present SHA. As SHA
depends on the type of seismic sources, two types of seismic sources (linear and
areal) with multiple source models (SMs) are considered. These SMs are based on
geotectonics and EQ distribution. For DSHA, highest SHL is obtained when SM with
large areal sources is considered. However, for PSHA, hybrid SM, which contains
both linear and areal sources, gives the highest SHL. In DSHA and PSHA, signif-
icantly lower SHL is obtained when linear SM is considered. Thus, SHL obtained
from more than one SM should be examined before reaching a judgement about SHL
of a specific region.
Keywords Seismic hazard · Seismicity parameters · GMPE · NE India
B. Niranjan (B) · M. J. Kumar · K. Abhishek

IIT Guwahati, Guwahati, Assam, India
e-mail: niran521152@gmail.com
K. Abhishek
e-mail: abhiak@iitg.ac.in
https://doi.org/10.1007/978-981-99-1459-3_13
152 B. Niranjan et al.
1 Introduction
The development of infrastructure in a particular region is critical as it is the key

to the progress of the people living there. At the same time, while designing and
constructing any infrastructure/structures, care should be taken so that the structure
can prevail against any future natural disasters. Natural disasters such as earthquakes
(EQs) may lead to losses of lives due to the failure of the structure. Therefore,
for regions with high seismic activity, structures are designed as EQ-resistant. An
EQ-resistant structure is designed and constructed to withstand a region’s probable
future EQ load. The forces a structure may be subjected to during a future EQ, can
be determined based on the seismic hazard level for the region. The seismic hazard
level for a region can be estimated by performing a detailed seismic hazard analysis
(SHA). For performing SHA, information on the region’s past EQs (magnitude,
location, and time of occurrence), EQ sources of the region, regional ground motion
prediction equations (GMPEs), etc., are required. It has to be highlighted here that
the outcome from an SHA study is strongly influenced by the appropriate selection of
the above-mentioned regional parameters/equations. However, in previous seismic
hazard studies, ambiguity is seen in the selection of the type of sources, methodology,
GMPEs. As a result, seismic hazard levels for the same region may vary from one
study to another based on the selected regional parameters. In the present study, the
influence of different types of sources, their shapes, and EQ distribution within each
of them on the outcome of SHA is studied. Both deterministic SHA (DSHA) and
probabilistic SHA (PSHA) are performed for different locations in Sikkim, a state
of India.
2 Study Area and Its Past Earthquake Scenario
Sikkim, a state in North East (NE) India, is situated in the highly seismically active
Himalayan region. As per IS-1893 (2016) part 1 [1], Sikkim falls in seismic zone
IV. Active faults such as MFT and MBT are present in the vicinity of the region.
Additionally, several other active faults, such as MCT, Teesta lineament, and Gangtok
lineament, are directly passing through Sikkim. Due to the presence of active faults
and the Indo-Eurasian plate boundary, the region has the potential to experience
EQs highly destructive in future. The region has already witnessed a number of past
damaging EQs, i.e. Nepal-India border EQ of 1988 (Mw = 6.9), 1943, Assam EQ
(Ms = 7.2), 1930 Dhubri EQ (Ms = 7.1), 1923 Meghalaya EQ (Ms = 7.1), Assam
EQ of 1897 (Mw = 8), and 1869 Cachar EQ (Mw = 7.5). These EQs caused losses
of lives and the destruction of properties all over Sikkim.
Figure 1 shows EQ events that took place in and around Sikkim. This past EQ data
is collected from NDMA [2], ISC [3], and USGS [4]. It has to be mentioned here that
the magnitudes of the collected EQ data were in different units. Thus, all magnitudes
of the EQs are converted to the Mw using the relation proposed by Sitharam and Sil
Seismic Hazard Analysis Considering the Effect of the Shape, Size, … 153
Fig. 1 Past EQ events in and around Sikkim
[5]. In this catalogue, the number of EQs having magnitudes Mw < 4, 4 ≤ Mw < 5,
5 ≤ Mw < 6, 6 ≤ Mw < 7, 7 ≤ Mw < 8, and 8 ≤ Mw are 515, 3304, 526, 81, 17, and
5, respectively (Fig. 1). The location of the great EQs (EQ events having magnitude
greater than or equal to 8Mw) near Sikkim is depicted in Fig. 1 by red star markers.
The distribution of EQs suggests that East, South, and West Sikkim are seismically
very active.
3 Methodology
Present work performs DSHA and PSHA for some important locations of Sikkim.
The source models (SMs) adopted for the study and the methodology used are
described in the following subsections.
3.1 Seismic Source Models Used for the Analyses
There is always an uncertainty exist regarding the location, shape, size, etc., of seismic
sources [6]. Thus, one type of SM may not sufficient to correctly estimate the hazard
level of a particular region. Mostly, the types of seismic sources considered for hazard
analysis are linear, areal, and hybrid. Thus, the present study considers three types
of SMs to determine the variations in the final seismic hazard values obtained from
considering different source types. These SMs are areal SM (SM1), linear SM (SM2),
Fig. 2 SMs considered for SHA: a SM1a (Za1, Za2, etc., are the areal sources considered in the
SM), b SM1b (Zb1, Zb2, etc., are the areal sources considered in the SM), c SM2 (a and b), and d
SM3 (SA1, SA2, etc., are the areal sources considered in the SM)
and hybrid SM (SM3). Figure 2 shows all these SMs. Further details about these SMs
are described below.
Areal SM. Normally, EQs occur in faults, however, it is also seen that EQs occur
far away from known faults. Thus, to account for the occurrence EQs in area far
from known faults or past EQ events, areal sources are considered. The present study
utilises two areal SMs (SM1a and SM1b) with different dimensions and locations to
deal with the uncertainty associated with the delineation of the areal sources. Both
these two SMs are based on the EQ distribution and seismotectonics of the region.
SM1a is based on the seismic activity and tectonics observed in Figs. 1 and 2c,
respectively. On the other hand, SM1b is based on Gupta [7], and Kolathayar and
Sitharam [8]. It has to be mentioned here that SM1a is a large-scale seismic SM with
areal sources having larger dimensions than SM1b (Fig. 2a, b). In Fig. 2, the areal
sources of SM1a are Za1, Za2, Za3, Za4, and Za5 (Fig. 2a). The areal sources of
SM1b are Zb1, Zb2, Zb3, Zb4, Zb5, Zb6, Zb7, Zb8, Zb9, Zb10, Zb11, Zb12, Zb13,
and Zb14 (Fig. 2b).
Linear SM. Normally, EQs occur on faults and faults are mostly linear. Thus, for
SHA, linear sources are considered. Normally, it is considered that the seismicity
is uniform throughout a fault and EQs may occur randomly in any location on the
faults. However, sometime it is also observed that distribution of EQs is not uniform
throughout a linear seismic source. Thus, in the present study, the variation due
to consideration of these two cases (uniform seismicity and nonuniform seismicity
throughout a fault) is determined. The present study utilises two linear SMs (SM2a
and SM2b). The faults/linear source information for linear SM is compiled from
Bhukosh [9], Kundu and Gahalaut [10], and Chen [11] and is shown in Fig. 2c.
In both cases, linear sources considered for the analyses are the same (Fig. 2c).
However, in SM2a, the seismic activity is considered uniform throughout the fault
length. In SM2b, on the other hand, seismic activity for the different parts of the
fault is considered different based on the EQ distribution throughout the fault. It has
to be highlighted here that SM2b is considered in the present study to account for
nonuniform seismic activity within a long fault. Further, it is also observed that for
the region, some faults are changing their directions from location to location and are
not entirely linear. Due to this, for the analyses, those faults are considered segment
wise such that each segment is linear. Figure 2c shows the SM2 considered for the
study.
Hybrid SM. In hybrid SM, both linear and areal sources are considered together
for SHA (SM3). Figure 2d shows the sources considered in SM3 for the analyses.
Areal sources are considered for the areas with no faults information despite having
many EQs, locations with a complex tectonic setting, and locations where EQs are
forming clusters. These areal sources are referred as SA1, SA2, SA3, SA4, SA5,
SA6, SA7, SA8, SA9, SA10, SA11, SA12, SA13, SA14, SA15, SA16, and SA17 in
Fig. 2d.
3.2 Deterministic Seismic Hazard Analysis of the Locations
In DSHA, firstly, for all the SMs, EQs are separated for each source. Based on the
Mmax observed in each source, the maximum potential magnitude (Mp) is calculated
by considering an increment of Mw value of 0.5 to the observed Mmax value. It is to
be mentioned here that there are many methods available for obtaining the Mp value
[12, 13]. The Mp values obtained from different methods together can be utilised
for the determination of seismic hazard level. However, as the present work mainly
focuses on the estimation of the variation of hazard level due to the consideration of
different SMs, the current method is only used considering the increment value of
0.5 Mw, from the work done by Bahuguna and Sil [14]. Further, minimum source
to the site distances is calculated for all the SMs. Utilising the above computed Mp
of the sources and minimum source to site distances, the maximum seismic hazard
level is estimated for all the locations following GMPE developed by NDMA. Thus,
for different SMs, different DSHA response spectra are generated.
3.3 Probabilistic Seismic Hazard Analysis of the Locations
In PSHA, firstly, declustering is done to obtain an EQ catalogue having only main EQ

events. The declustering is carried out by the methodology proposed by Uhrhammer
[15]. The number of EQs remains in the declustered catalogue having magnitudes
Mw < 4, 4 ≤ Mw < 5, 5 ≤ Mw < 6, 6 ≤ Mw < 7, 7 ≤ Mw < 8, and 8 ≤ Mw are 370,
1878, 317, 55, 15, and 4, respectively. After obtaining the declustered EQ catalogue,
EQs are then distributed source-wise. For the areal sources of SM1b, which have few
EQs and are close to each other, combined EQs are used for the determination of the
seismicity parameters. Before determining the seismicity parameters, completeness
analyses with respect to magnitude (magnitude of completeness Mc) and time are
carried out using maximum curvature and CUVI methods. Based on the complete
EQ catalogue, seismicity parameters for each source of the SMs are calculated. It
has to be mentioned here that the least square method (LSM) is utilised in the present
study to estimate the seismicity parameters (a and b of Gutenberg-Richter recurrence
relation [16] Eq. 1). In the LSM approach, seismicity parameters are determined by
fitting a straight line to the frequency-magnitude distribution plot [17]. The procedure
followed in the above discussion can be found in Borah et al. [18] and Borah and
Kumar [19]
log N (M) = a − bM. (1)
For SM2 and SM3, SM1a sources are considered as seismic source zones. Firstly,
seismicity parameters are estimated for these source zones. Later, for each linear
source/fault, the seismicity parameters are distributed based on the method suggested
by Iyengar and Ghosh [20]. This distribution is based on the number of EQs that took
place within the fault and the length of the fault. In the case of areal sources, only
the number of EQs within a source is considered for the distribution of seismicity
parameters.
The seismicity parameters thus obtained in the present study are used to determine
the probability of exceedance of ground motion level ‘z’, PE (z)
PE (z) = 1 − e−vz .t , (2)
where vz is the annual frequency of exceedance of ground motion level z. vz can be

calculated by using the Eq. (3) [21]
⎡ ⎤
i =m r j =rmax

N m
u

v(z) = λn (m i )⎣ Pn R = r j |m i P Z > z|m i r j ⎦. (3)
n=1 m i =m 0 r j =rmin
The term λn (m i ) in Eq. (3) is the annual frequency of occurrence of EQs of

magnitude mi on seismic source n. Pn (R = r j |m i ) is the probability of an EQ of
magnitude mi on source n occurring at a certain distance r j from the site. P(Z >
z|m i r j ) is the probability that ground motion z will be exceeded, given an EQ of

magnitude mi on source n at a distance r j from the site. The procedure to obtain
these terms can be found in Baro et al. [22]. In the present study, utilising Eq. (2) and
Eq. (3), the hazard level for a particular location is calculated for 2% probabilities in
50 years.
For 6 locations (Soreng, Pelling, Namchi, Mangan, Padamchen, and Gangtok. see
Fig. 3) distributed in Sikkim, SHA is carried out using all the earlier mentioned SMs.
The results obtained are presented in the following subsections.
4.1 DSHA Results
Figure 4a–f shows Response Spectra (RS) obtained based on DSHA for Soreng,
Pelling, Namchi, Mangan, Padamchen, and Gangtok, respectively. Further, RS based
on each of the SMs is also shown individually in each of these figures (Fig. 4a–f).
RS obtained considering SM1a, SM1b, SM2a, and SM3 clearly shows minimal
variations in Sa values at all the time periods. However, in the case of SM2b, where
seismic activity is considered different throughout the faults, the Sa values for all 6
locations are significantly low compared to the values obtained from considering the
other SMs (see Fig. 4a–f).
Fig. 3 Locations in Sikkim

for which SHA are
performed (blue lines
represent the faults presents
in and around Sikkim)
1.40 1.40
1.20 (a) Soreng SM1a 1.20 (b) Pelling SM1a
SM1b SM1b
1.00 1.00
SM2a SM2a
0.80 SM2b 0.80 SM2b
0.60 SM3 0.60 SM3
Mean Mean
0.40 0.40
0.20 0.20
0.00 0.00
0 1 2 3 4 0 1 2 3 4
1.40 1.40
1.20 SM1a 1.20 SM1a
(c) Namchi SM1b (d) Mangan SM1b
1.00 1.00
SM2a SM2a
0.80 SM2b 0.80 SM2b
0.60 SM3 0.60 SM3
Mean Mean
0.40 0.40
0.20 0.20
0.00 0.00
0 1 2 3 4 0 1 2 3 4
1.40 1.40
1.20 SM1a 1.20 SM1a
(e) Padamchen SM1b (f) Gangtok SM1b
1.00 1.00
SM2a SM2a
0.80 SM2b 0.80 SM2b
0.60 SM3 0.60 SM3
Mean Mean
0.40 0.40
0.20 0.20
0.00 0.00
0 1 2 3 4 0 1 2 3 4
X-axis: Time periods (s) and Y-axis: Sa (g)
Fig. 4 Response spectra obtained from DSHA for the 6 locations in Sikkim
Further, mean RS is also calculated at each of the six locations based on RS

obtained from different SMs and also shown in Fig. 4a–f. Table 1 presents the PGA
values obtained based on different SMs. Further, mean PGA and coefficient of vari-
ation for all the locations are also given in Table 1. The coefficient of variation is the
ratio of standard deviation and mean of any distribution. It shows the variation of
the estimated PGA values obtained from considering different SMs. It can be seen
from Table 1 that the highest mean PGA value (0.6 g) is obtained for Namchi loca-
tion, whereas the minimum PGA value (0.56 g) is obtained for Mangan. Similarly,
the highest variability in estimated PGA is seen in case of Mangan and the lowest
variability is seen in case of Namchi. In other words, the variation in the estimated
PGA value is more when different SMs are used in case of Mangan.
Table 1 PGA values for 6 location obtained from DSHA

Locations SMs Mean Standard Coefficient of
SM1a SM1b SM2a SM2b SM3 deviation variation
Soreng 0.67 0.60 0.64 0.44 0.60 0.59 0.08 0.14

Pelling 0.67 0.60 0.64 0.40 0.60 0.58 0.10 0.17
Namchi 0.67 0.60 0.63 0.49 0.60 0.60 0.06 0.10
Mangan 0.67 0.59 0.64 0.31 0.60 0.56 0.13 0.23
Padamchen 0.67 0.56 0.64 0.40 0.60 0.58 0.10 0.17
Gangtok 0.67 0.57 0.64 0.36 0.60 0.57 0.11 0.19
4.2 PSHA Results
Figure 5a–f shows uniform hazard spectra (UHS) for hazard levels of 2% probabil-
ities in 50 years for Soreng, Pelling, Namchi, Mangan, Padamchen, and Gangtok,
respectively. In each of these figures, Sa obtained based on different SMs is also
shown separately. The Sa values obtained based on different SMs shows variations
at all time periods. This variation is less for some time periods (for time periods more
than 0.3 s), and the variation is more significant for others (for time periods from 0 s
to 0.2 s). Table 2 shows PGA values for all the sites considering different SMs for
hazard levels of 2% probabilities in 50 years. From the result, it is found that for the
large-scale SM (SM1a), the spatial variation in PGA values is very less. As we can
observe (from Table 2) that for all the 6 locations, the PGA value is almost the same
(≈0.43 g). It may be attributed to the consideration of uniform seismicity parameters
of large areal sources. For the other SMs, however, a significant difference in PGA
is observed at different locations. For these SMs, the difference between the PGA
values of two locations is as large as 0.1 g (see Table 2). For each of the five SMs, it
is observed that the maximum PGA is obtained at distinct locations. The maximum
PGA is observed at Mangan, Pelling, Soreng, Soreng, and Pelling for SM1a, SM1b,
SM2a, SM2b, and SM3, respectively. Further, the PGA values obtained while consid-
ering SM3 are the highest for any particular location. While considering the mean
value from all the SMs, Soreng has the highest PGA (0.44 g), and Padamchen has the
lowest PGA value (0.36 g). The standard deviation and coefficient of variation values
for PGA for all the cities suggest that the variability in PGA value is significant for all
the cities. The highest variability is obtained for Mangan, and the lowest variability
is seen in case of Soreng. All these observations suggest that the size, shape, type of
sources, and activity of past EQs significantly affect the spatial distribution of hazard
values in Sikkim region.
1 1
(a) Soreng SM1a (b) Pelling SM1a
0.8 SM1b 0.8 SM1b
SM2a SM2a
0.6 SM2b 0.6 SM2b
SM3 SM3
0.4 Mean 0.4 Mean
0.2 0.2
0 0
0 1 2 3 4 0 1 2 3 4
1 1
(c) Namchi SM1a (d) Mangan SM1a
0.8 SM1b 0.8 SM1b
SM2a SM2a
0.6 SM2b 0.6
SM2b
SM3
SM3
0.4 Mean 0.4
Mean
0.2 0.2
0 0
0 1 2 3 4 0 1 2 3 4
1 1
SM1a SM1a
0.8 (e) Padamchen 0.8
(f) Gangtok
SM1b SM1b
SM2a SM2a
0.6 SM2b 0.6 SM2b
SM3 SM3
0.4 Mean 0.4 Mean
0.2 0.2
0 0
0 1 2 3 4 0 1 2 3 4
X- axis: Time periods (s) and Y-axis: Sa (g)
Fig. 5 UHS obtained from PSHA for hazard levels of 2% probabilities in 50 years
Table 2 PGA values for 6 location obtained from PSHA

Locations SMs Mean Standard Coefficient of
SM1a SM1b SM2a SM2b SM3 deviation variation
Soreng 0.43 0.49 0.39 0.35 0.52 0.44 0.07 0.16

Pelling 0.43 0.49 0.36 0.31 0.53 0.43 0.09 0.21
Namchi 0.43 0.46 0.36 0.32 0.52 0.42 0.08 0.19
Mangan 0.44 0.41 0.30 0.24 0.49 0.38 0.10 0.27
Padamchen 0.43 0.32 0.34 0.24 0.45 0.36 0.09 0.24
Gangtok 0.43 0.38 0.33 0.28 0.49 0.38 0.08 0.21
5 Conclusion
The present study performs DSHA and PSHA for 6 locations of Sikkim, considering
different seismic SMs. The SMs considered are areal SMs, linear SMs, and hybrid
SM. It is found that for the locations of Sikkim, large areal SM (SM1a) is generating
the highest value of seismic hazard for all the locations in the case of DSHA. The
lowest hazard value is observed in the case of linear SM (SM2b), where seismic
activity is considered different throughout a linear source. For a long fault/linear
source, it is usually seen that the seismicity is nonuniform throughout the fault.
As a result, if the historical EQ information obtained throughout the fault is not
adequately captured, the potential of the fault may be underestimated in the case
of DSHA. Based on the results obtained from PSHA, it is found that the hazard
level is highest for all locations when the hybrid source model is considered. Here,
linear SMs give the lowest hazard level for all the locations. In the case of areal SM,
for some locations, SM1a is giving higher values, and for other locations, SM1b
is giving higher values. However, in the case of linear SM, SM2a gives the highest
values for all the locations. The present results show that the selection of seismic SMs
significantly affects the outcome of SHA (both DSHA and PSHA) for locations in
Sikkim. The hazard value may vary significantly based on the site’s location, type of
seismic sources, and distribution of seismicity. Thus, while determining the seismic
hazard potential of Sikkim or other nearby locations, it is better to consider more than
one SMs during SHA. Besides taking the mean value obtained from multiple hazard
values, each value from different SMs should also be examined while determining
the seismic hazard level of a location. In other words, variations in hazard level
much also examine for a particular location. This will give a better picture of the
seismic potential of a particular region. The results of the present SHA can be further
improved by incorporating other types of epistemic uncertainties (uncertainty in the
selection of methodologies to determine seismicity parameters, selection of GMPE,
selection of probability models, etc.) associated with SHA.
References
1. BIS.: IS-1893 2016 part 1: Criteria for Earthquake Resistant Design of Structures Sixth Revision
(2016)
2. NDMA.: Development of Probabilistic Seismic Hazard Map of India Technical Report (2010)
3. ISC.: Event CatalogueSearch [Internet]. [cited 2019 May 1]. Available from: http://www.isc.
ac.uk/iscbulletin/search/catalogue/
4. USGS.: Search Earthquake Catalog [Internet]. [cited 2019 May 1]. Available from: https://ear
thquake.usgs.gov/earthquakes/search/
5. Sitharam, T.G., Sil, A.: Comprehensive seismic hazard assessment of Tripura and Mizoram
states. J. Earth Syst. Sci. 123, 837–857 (2014)
6. Roe, P., Georgsen, F., Abrahamsen, P.: An uncertainty model for fault shape and location. Math.
Geosci. 46, 957–969 (2014)
7. Gupta, I.D.: Delineation of probable seismic sources in India and neighbourhood by a compre-
hensive analysis of seismotectonic characteristics of the region. Soil. Dyn. Earthq. Eng. 26,
766–790 (2006)
8. Kolathayar, S., Sitharam, T.G.: Characterization of regional seismic source zones in and around
India. Seismol. Res. Lett. 83, 77–85 (2012)
9. Bhukosh.: Bhukosh. Available from: https://bhukosh.gsi.gov.in/Bhukosh/MapViewer.aspx
10. Kundu, B., Gahalaut, V.K.: Tectonic geodesy revealing geodynamic complexity of the Indo-
Burmese arc region, North East India. Curr. Sci. 104, 920–933 (2013)
11. Chen, B.: Stress-induced trend: the clustering feature of coal mine disasters and earthquakes
in China. Int. J. Coal. Sci. Technol. [Internet]. 7, 676–92(2020). Springer Singapore. Available
from: https://doi.org/10.1007/s40789-020-00334-z
12. Kijko, A., Sellevoll, M.A.: Estimation of earthquake hazard parameters from incomplete data
files, part I: utilization of extreme and complete catalogues with different threshold magnitudes.
Bull. Seismol. Soc. Am. 79, 645–654 (1989)
13. Kijko, A., Graham, G.: Parametric-historic procedure for probabilistic seismic hazard analysis.
Part I: estimation of maximum regional magnitude m(max). Pure Appl. Geophys. 152, 413–42
(1998)
14. Bahuguna, A., Sil, A.: Comprehensive seismicity, seismic sources and seismic hazard assess-
ment of Assam, North East comprehensive seismicity, seismic sources and seismic. J. Earthq.
Eng. Taylor Francis 00, 1–44 (2018)
15. Uhrhammer, R.A.: Characteristics of Northern and Central California seismicity. Earthq. Notes.
(1976)
185–8 (1944)
17. Guttorp, P.: On least-squares estimation of b values. Bull. Seismol. Soc. Am. 77, 2115–2124
(1987)
18. Borah, N., Kumar, A., Dhanotiya, R.: Seismic source zonation for NE India on the basis of
past EQs and spatial distribution of seismicity parameters. J. Seismol. (2021)
19. Borah, N., Kumar, A.: Assessing if foreshocks and aftershocks are dependent events of the main-
shock for North–East India. 17th World Conference Earthquake Engineering 17th. (Sendai,
2020)
20. Iyengar, R.N., Ghosh, S.: Microzonation of earthquake hazard in Greater Delhi area. Curr. Sci.
87, 1193–1202 (2004)
21. EM.: Response spectra and seismic analysis for concrete hydraulic structures. Washington,
DC, Department of the Army U.S. Army Corps of Engineers (1999)
22. Baro, O., Kumar, A., Ismail-Zadeh, A.: Seismic hazard assessment of the Shillong plateau
using a probabilistic approach. Geomatics, Nat. Hazards Risk. [Internet]. Taylor Francis 11,
2210–38 (2020). Available from: https://doi.org/10.1080/19475705.2020.1833989
A Comparative Study on Application
of Machine Learning Algorithms
in Ground Motion Prediction Equations
A. Ahmed and M. Gade
Abstract Ground motion prediction equations (GMPEs) play an important role in

the field of earthquake engineering and seismic hazard analysis. With the advance-
ments in the area of artificial intelligence (AI) and machine learning (ML), it becomes
critical to have a comparative study on the new methods of obtaining ground motion
parameters for a region. Traditional GMPEs developed using parametric regression
assume a fixed functional form, and in doing so, it becomes challenging to draw
the complex and nonlinear characteristics of the data completely. Adopting nonpara-
metric regression methods, which are independent of predefined equations, helps
us to overcome these limitations. This paper investigates a comparative study of
some common machine learning algorithms to forecast peak ground acceleration
(PGA) associated with ground motion data of NGA West. The prediction model
considers nonparametric types of regression and tries to manifest the advantages of
these models over conventional GMPEs, derived using parametric regression. The
present work explores algorithms like decision trees, random forests, support vector
regression, and weighted average ensemble method (WAEM). Also, the performance
of the developed models is compared with that of conventional methods. In this study,
the earthquake magnitude, rupture distance, average shear wave velocity, and the type
of faulting mechanism are used as predictor variables. The algorithms are trained
using a database of 13,555 ground motions, with magnitude ranging from 3.0 to 7.9
recorded over the rupture distance range of 0–500 km. Finally, an efficient model is
identified using evaluation metrics like mean squared error (MSE), R-squared value,
and standard deviation.
Keywords Machine learning · GMPEs · Support vector regression · Decision

trees · Random forest
A. Ahmed (B) · M. Gade

Indian Institute of Technology Mandi, Kamand, Himachal Pradesh, India
e-mail: t20012@students.iitmandi.ac.in
M. Gade
e-mail: maheshreddy@iitmandi.ac.in
https://doi.org/10.1007/978-981-99-1459-3_14
164 A. Ahmed and M. Gade
1 Introduction
The ground motion prediction equations (GMPEs) are essential in quantifying

seismic hazards and the design of earthquake-resistant structures. Typically, GMPEs
use different regression-based methods to predict the ground motion intensity
measures (IMs), by giving input variables such as source and path characteristics
along with local site conditions [1]. Traditional regression models have been the
most classical approach in predicting these IMs, where the dependant variable is a
function of various input variables like magnitude M w , distance R, source, and site
characteristics, like fault mechanism and Vs30 [2]. These models extract the first-order
aspects of the input variables but restrict themselves in drawing the complex nonlinear
behaviour of the data. To overcome this limitation, researchers have tried to add more
regression coefficients to the models, incorporating different physical phenomenons.
This brings into the picture another drawback, called overfitting, where the predic-
tion model ‘overfits’ itself to the existing data and results in an increased variance
failing to fit the additional data aptly. Hence, it becomes a challenging task to select
such a predefined functional form that incorporates complex source, site, and path
effects, without compromising the quality of the fit. Adopting nonparametric regres-
sion paves the way for a data-driven technique where the predictor variable does
not take a predefined mathematical form but regresses according to the information
derived from the data. Various nonparametric prediction models have been devel-
oped by researches [3–6], which require a relatively large number of data points as
compared to their parametric counterparts. Tezcan and Cheng [7] used a supervised
machine learning algorithm called support vector regression (SVR) to estimate PGA,
PGV, and spectral accelerations (SAs) using the PEER-NGA database consisting of
2754 records from 135 events. Thomas et al. [8], Tao et al. [9], and Hu and Zhang
[10] also used the SVR method to forecast the ground motion parameters for different
parts of the world. Researchers have also tried deep learning (DL) models like artifi-
cial neural networks (ANNs), Güllü and Erçelebi [11], Kerh and Ting [12], Ahmad
et al. [13], Derras et al. [14], Dhanya and Raghukanth [15], and Gandomi et al.
[16] to predict PGA, PGV, and SAs using various databases available across the
world. Hamze-Ziabari and Bakhshpoori [17], Asencio-Cortés et al. [18], Trugman
and Shearer [19], and Kong et al. [20] used other ML algorithms like regression trees
and random forests to develop GMPEs. Lately, researchers Fayaz et al. [21] and Asim
et al. [22] developed hybrid models in which the prediction parameters were obtained
using SVR and a hybrid neural network (HNN)-based classification system. Also,
Kubo et al. [23] presented a hybrid strategy that used machine learning technology
with a physical model that produced better predictions than either method alone.
The present study aims at developing prediction models using three learning
algorithms, namely support vector regression, decision trees, and random forests, to
forecast PGA. The study also uses the weighted average ensemble method, which
results in a better prediction power as compared to the individual use of ML algo-
rithms. The same database is used to train all the models, which is explained in the
A Comparative Study on Application of Machine Learning Algorithms … 165
next section. The models are implemented in an open-source programming language,

Python 3.7.12.
The subsequent portions of the paper are organized in the following order. The
earthquake data utilized in the study is described in Sect. 2, which also includes a
discussion of input and output variables. A brief discussion on different ML algo-
rithms utilized in this study is done in Sect. 3. The performance of developed models
is investigated using different criteria presented in Sect. 4. A comparative assess-
ment is made with the help of existing relations in Sect. 5. Finally, the summary and
conclusions from the study are presented in Sect. 6.
2 Dataset
In the present study, we have used the updated PEER-NGA-West2 database [24],
available at https://peer.berkeley.edu/research/data-sciences/databases. The data is
pre-processed as per the elimination criterion followed by Dhanya and Raghukanth
[15], and the final database obtained has a total of 13,555 data points recorded
from 288 earthquakes. The earthquake magnitude (Mw ), rupture distance (Rr up ),
average shear velocity (Vs30 ), and faulting mechanism (FM) are considered as input
parameters to develop the models. Here, Vs30 represents the shear wave velocity in
the top 30 m of the soil profile in m/s and is one of the most significant features
influencing the ground motion parameters. Table 1 lists the classification criteria for
defining values of FM, as per the documentation provided by NGA-West2. Vs30 ,
Rr up , and PGA are converted to their respective logarithmic values before building
the models.
Hence, the input and output vector, before further pre-processing is

log10 (PGA) = Mw log10 Rr up log10 (VS30 ) FM .
The database is divided into training and testing data, with 70% for the training
part and 30% for testing. Thus out of 13,555 points, the training subset has 9,488 and
the testing subset has 4067 data points. Variables with values at different scales may
not contribute equally to the model fitting, which might lead to bias. As a result, we
have used feature-wise standardization before model fitting to address this potential
problem. Here, the values of a particular feature are centred around the mean with a
unit standard deviation. Let x be the variable and μ and σ be the mean and standard
Table 1 Values of FM on the basis of rake angle

Faulting style FM Range of rake angle (λ)
Strike slip 1 −180° < λ < −150°, −30° < λ < 30°, 150° < λ < 180°
Normal and normal oblique 2 −150° < λ < −30°
Reverse and reverse oblique 3 30° < λ < 150°
deviation of that particular variable. After applying the above-mentioned scaling

technique, we get a standardized value z, defined as.
z = (x − μ)/σ. (1)
3 Machine Learning Algorithms
3.1 Support Vector Regression (SVR)
Vapnik [25] came up with a major breakthrough in machine learning, by introducing a

very efficient yet computationally simpler algorithm called support vector machines
(SVMs). In contrast to other classification approaches such as logistic regression,
which takes a probabilistic approach to problem solving, SVM takes a statistical
approach. Let us first look at the classification done by SVM before understanding
the notion underlying SVR. In an N-dimensional space, SVM introduces a concept
of a ‘hyperplane’, which acts as a decision boundary to optimally divide the classes,
similar to a line and a plane in two-dimensional and three-dimensional space, respec-
tively. The SVR model employs a number of kernel functions, but in our study we
will be using three kernel functions: linear kernel, polynomial kernel, and radial basis
function (RBF) Kernel. We have built and evaluated the model’s performance using
these three kernel functions, as reported in Sect. 4.
3.2 Decision Trees (DT)
Decision trees are general-purpose prediction and classification mechanisms that

have evolved into extremely cross-disciplinary, computationally intensive methods
for prediction and classification, machine learning, and AI problems. The main
feature of decision trees is that they create partitions and associated descendent data
subsets (called leaves) at any given level of the tree based on the values of associated
input variables. Asim et al. [22, 26], Kong et al. [20], and other researchers have made
used of this technique to arrive at the prediction of ground motion parameters. The
algorithm iterates through each subset, taking into account only those attributes that
were not chosen in prior rounds. As a result, a decision tree-based learning model
is built, in which the path from the root to the leaf depicts the values of the input
variables, and each leaf represents the value of the target variable.
3.3 Random Forest (RF)
Ensemble methods in statistics and machine learning combine many learning algo-
rithms to achieve greater predictive performance than either of the individual algo-
rithms. Ho [27] and Amit and German [28] made use of such an ensemble approach
called random forests (RFs) in which multiple decision trees are utilized at a time,
and their respective outputs are aggregated to give a final output. RFs works on the
concept of ‘Law of Large Numbers’ where randomness in the model is induced by
working with multiple DTs. This random selection of rows and columns in the form
of individual DTs helps to reduce the chance of model overfitting and hence the
generalization error. As the number of DTs increases, although the computational
complexity rises, but the generalization error reduces significantly and the model
works better on the testing data.
4 Performance Analysis
In this section, prediction models for PGA are developed using SVR, DT, and RF, and
the performance of the individual models is compared to identify the efficient model.
Further, the attenuation characteristics of the developed models are also investigated
to verify the models’ ability to predict the PGA values as per physical laws for the
given input parameters.
4.1 Evaluation Metrics
Firstly, the coefficient of determination (R 2 ) is calculated for each developed model

and listed in Table 2. An R 2 value above 0.7 indicates a strong ‘goodness of fit’, which
we obtained for all models. Another criteria considered to access the performance
is mean square error (MSE), which is the average of the squares of the errors, and
a lower value represents a better model. The standard deviation of the residuals (σ )
of log10 PGA also represents the total variability in the ground motion prediction
model. The lesser the value of σ , the better the model’s predictive power. Out of all
the considered models, the σ of residuals for RF is the least, equal to 0.2426 (log10
PGA units).
4.2 Physical Observed Trends
The formation of conventional ground motion prediction equations relies upon the
physical relationships between the input and output variables. Their formulation
Table 2 Evaluation metrics

Algorithm R2 MSE σ
for different ML algorithms
SVR: linear kernel 0.8453 0.1767 0.4263
SVR: polynomial kernel 0.7225 0.3296 0.5694
SVR: RBF 0.9048 0.1147 0.3205
DT 0.9043 0.1153 0.3359
RF 0.9188 0.0964 0.2426
is primarily based on seismological relationships. The data-driven techniques do

not need any predefined physical constraints and mechanisms, but they must be
investigated for these physical phenomena to check whether the proposed methods
are feasible to use or not. The physical trends of predicted PGA values from the
SVR model with RBF kernel, for various magnitude and site classes, are presented
in Figs. 1 and 2.
It is evident from Fig. 1 that even in the absence of any predefined assumption of
functional dependence, the results show a significant magnitude reliance, which is
consistent with physical observations. As the rupture distance increases, the value of
PGA decreases smoothly. The smooth nature of plots adequately depicts the seismo-
logical laws, and it should also be noticed that for short rupture distances (R Rup <
6 km), the intensity values are almost constant. It is due to the source dominance in
the near-field region. When we move away from the source, the path characteristics
start taking dominance over the source, and there is a significant drop in the PGA
values. In addition to this, since PGA is a high-frequency ground motion character-
istic, it attenuates faster at large distances but does not show much change when the
sites are closer to the source. Moreover, it can also be seen from Fig. 1 that as the
value of magnitude increases, values of PGA increase, and the degree of attenuation
Fig. 1 Variation of SVR model predictions (PGA) with respect to Rrup and M w with V s30 =
1130 m/s, FM = 1
Fig. 2 Variation of SVR model predictions (PGA) with respect to Rrup and V s30 with M w = 6, FM
=1
also shifts at a higher value of rupture distance. The ‘Local site effects’ significantly
influence the intensity of earthquake. The effect of different ‘Soil Classes’ on PGA
is illustrated in Fig. 2. It can be observed from the figure that the model is able
to capture the amplification of PGA values with respect to soil type. From these
observations, we can conclude that the developed data-driven PGA prediction model
based on SVR is efficient in following the physics of the GMPEs.
A similar analysis was also performed for prediction models developed using DT
and RF algorithms, but it was noticed that these relations were not smooth and a
significant amount of undulations were present. This can be attributed to the fact that
DT and RF algorithms are tree-based in nature and do not involve any mathematical
equations in their formulations. Instead, they rely on a discrete tree-based method,
which assigns same value of PGA for different values of input parameters. The
phenomenon of overfitting takes place in case of these two algorithms; hence, we
have not presented the physical trends for these two models. However, recall that
the RF model gave an excellent analytical performance, with a very high value of
R 2 and the lowest value of the total standard deviation of residuals. In order to take
advantage of these observations, we have explored a method known as the ‘weighted
average ensemble method’ (WAEM) and the corresponding details are presented in
the following sub-section.
4.3 Weighted Average Ensemble Technique
It is observed earlier that the support vector regression (SVR) algorithm’s predictions
follow the attenuation characteristics of the IMs, but the analytical performance is
decent. On the other hand, the random forest (RF) gives a better analytical perfor-
mance than the SVR, but the physical trends do not follow the attenuation character-
istics due to overfitting problems. In order to take advantage of these two algorithms,
a weighted average of the predictions given by SVR and RF is proposed. Different
weights are assigned to SVR and RF predictions in a hit and trial fashion, and their
performance has been assessed. A combination of 70% SVR and 30% RF gives satis-
factory results, both analytically as well as attenuation characteristics. We noticed
that the R 2 value in this weighted average technique comes out to be better than
the one given by the individual implementation of SVR model. We proceed with
the calculation of other evaluation metrics, and the values of R 2 , MSE, and σ come
out to be equal to 0.9057, 0.1137, and 0.2962, respectively. It should be highlighted
that the value of total standard deviation, which plays a vital part in the analysis of
seismic hazard comes out to be lesser than the individual SVR model, which gives
this technique an edge. Now it will be interesting to observe the attenuation patterns
of the proposed weighted average ensemble algorithm.
The predicted PGA values variation with input parameters are presented in Figs. 3
and 4. It can be seen from Figs. 3 and 4 that the PGA attenuates smoothly as the
rupture distance increases. Further, the figures also demonstrate the magnitude and
site class dependency on PGA predictions. From these plots, we can conclude that the
present ensemble prediction model is superior to the SVR model in terms of analytical
performance and ability to capture the physical influence of all input parameters.
The above figure shows that the predicted PGA values at the near-field region
are very close for different soil classes. This can be attributed to the fact that in
the near-field region, the source characteristics are dominant compared to the site
characteristic (V s30 ), and hence, these curves, for different site conditions, give almost
Fig. 3 Variation of WA ensemble model predictions (PGA) with respect to Rrup and M w with V s30
= 1130 m/s, FM = 1
Fig. 4 Variation of WA ensemble model predictions (PGA) with respect to Rrup and V s30 with M w
= 6, FM = 1
the same intensity values in near-field regions. Now it will be interesting to compare
the performance of the present models with the prediction models available in the
literature. In this regard, an attempt is made, and the results are presented in the
following section.
5 Comparison with Existing Models
Here, we have selected Campbell and Bozorgnia [29] and Dhanya and Raghukanth
[14] equations for the comparison purpose, as both equations employed the same
database as ours (NGA-WESt2). Here, we have selected the classical parametric
regression-based equation (Campbell and Bozorgnia [29]), and another one (Dhanya
and Raghukanth [14]) uses the ANN technique to compare our results. A comparison
of mean PGA predictions is presented in Fig. 5. It is evident from the figure that the
SVR and WA ensembled predictions are comparable with the selected equations.
A comparison of the analytical performance of these models is performed, and
the results are presented in Table 3. It can be observed from the table that the R 2 and
standard deviation values of the SVR model are comparable with selected equations.
On the other hand, the standard deviation of the WA ensemble model is considerably
lesser than the selected models. Hence, it can be concluded that the WA ensemble
model can predict the PGA for a given set of input parameters with lesser variability.
Fig. 5 Comparison of the PGA predictions of SVR and WA ensemble models with selected
equations (M w = 6, V s30 = 1130 m/s, FM = 1)
Table 3 Evaluation metrics given by different models

Model R2 Total standard deviation (σ) (in log10 PGA units)
CB_2014 Not reported 0.3649
DR_2018 0.9005 0.3270
SVR 0.9048 0.3205
WA ensemble 0.9057 0.2962
6 Summary and Conclusions
The present study explores three different machine learning (ML) algorithms to
develop efficient data-driven GMPEs. These algorithms, SVR, DT, and RF, employ
unique learning architectures and extract the complex behaviour of data directly from
it. From the initial studies, it is observed that only the SVR algorithm is able to capture
the attenuation characteristics of the ground motion parameter, PGA. However, it is
observed that the RF model is superior in terms of analytical performance. In order to
take advantage of these two ML techniques, a ‘weighted average ensemble method’
is introduced. It is observed from the results that the WA ensemble model’s analytical
performance has improved compared to the SVR model and is able to capture the
attenuation patterns as well. Further, a comparison study with the existing equation
is also attempted to check the ability of the present ML-based prediction equations.
The mean predictions of the present two (SVR and WA ensemble) models are compa-
rable with the selected GMPEs (CB_2014, DR_2018). Further, the comparison of
analytical performance has shown that the SVR model’s R2 and σ are close to the
selected equations. On the other hand, the WA ensemble model’s σ is considerably
lesser than the selected prediction models. Thus, the present study has demonstrated
that the ML algorithms like SVR and RF can be used to develop efficient GMPEs.
Further, the same algorithms can be extended to develop prediction models for other
IMs like PGV, Ia, and spectral acceleration at several time periods.
References

2. Douglas, J.: Ground-motion Prediction Equations 1964–2010. (Pacific Earthquake Engineering
Research Center, 2011)
3. Anderson, J.G.: Nonparametric description of peak acceleration above a subduction thrust.
Seismol. Res. Lett. 68(1), 86–93 (1997)
4. Fajfar, P., Peruš, I.: A non-parametric approach to attenuation relations. J. Earthq. Eng. 1(02),
319–340 (1997)
5. Anderson, J.G., Lei, Y.: Nonparametric description of peak acceleration as a function of
magnitude, distance, and site in Guerrero, Mexico. Bull. Seismol. Soc. Am. 84(4), 1003–1017
(1994)
6. Brillinger, D.R., Preisler, H.K.: An exploratory analysis of the joyner-boore attenuation data.
Bull. Seismol. Soc. Am. 74(4), 1441–1450 (1984)
7. Tezcan, J., Cheng, Q.: Support vector regression for estimating earthquake response spectra.
Bull. Earthq. Eng. 10(4), 1205–1219 (2012)
8. Thomas, S., Pillai, G.N., Pal, K.: Prediction of peak ground acceleration using ∈-SVR, v-SVR
and Ls-SVR algorithm. Geomatics, Nat. Hazards Risk 8(2), 177–193 (2017)
9. Tao, D., Ma, Q., Li, S., Xie, Z., Lin, D., Li, S.: Support vector regression for the relationships
between ground motion parameters and macroseismic intensity in the Sichuan–Yunnan region.
Appl. Sci. 10(9), 3086 (2020)
10. Hu, J., Zhang, H.: Support vector machine method for developing ground motion models for
earthquakes in western part of China. J. Earthq. Eng. 1–16 (2021)
11. Güllü, H., Erçelebi, E.: A neural network approach for attenuation relationships: an application
using strong ground motion data from Turkey. Eng. Geol. 93(3–4), 65–81 (2007)
12. Kerh, T., Ting, S.B.: Neural network estimation of ground peak acceleration at stations along
Taiwan high-speed rail system. Eng. Appl. Artif. Intell. 18(7), 857–866 (2005)
13. Ahmad, I., El Naggar, M.H., Khan, A.N.: Neural network based attenuation of strong motion
peaks in Europe. J. Earthq. Eng. 12(5), 663–680 (2008)
14. Derras, B., Bard, P.-Y., Cotton, F., Bekkouche, A.: Adapting the neural network approach
to PGA prediction: an example based on the KiK-net data. Bull. Seismol. Soc. Am. 102(4),
1446–1461 (2012)
15. Dhanya, J., Raghukanth, S.T.G.: Ground motion prediction model using artificial neural
network. Pure Appl. Geophys. 175(3), 1035–1064 (2018)
16. Gandomi, A.H., Alavi, A.H., Mousavi, M., Tabatabaei, S.M.: A hybrid computational approach
to derive new ground-motion prediction equations. Eng. Appl. Artif. Intell. 24(4), 717–732
(2011)
17. Hamze-Ziabari, S.M., Bakhshpoori, T.: Improving the prediction of ground motion parameters
based on an efficient bagging ensemble model of M5 and cart algorithms. Appl. Soft Comput.
68, 147–161 (2018)
18. Asencio-Cortés, G., Morales-Esteban, A., Shang, X., Martínez-Álvarez, F.: Earthquake predic-
tion in California using regression algorithms and cloud-based big data infrastructure. Comput.
Geosci. 115, 198–210 (2018)
19. Trugman, D.T., Shearer, P.M.: Strong correlation between stress drop and peak ground accel-
eration for recent M 1–4 earthquakes in the San Francisco bay area. Bull. Seismol. Soc. Am.
108(2), 929–945 (2018)
20. Kong, Q., Trugman, D., Ross, Z., Bianco, M., Meade, B., Gerstoft, P.: Machine learning in
seismology: turning data into insights. Seismol. Res. Lett. 90, 11 (2018)
21. Fayaz, J., Xiang, Y., Zareian, F.: Generalized ground motion prediction model using hybrid
recurrent neural network. Earthq. Eng. Struct. Dyn. 50(6), 1539–1561 (2021)
22. Asim, K.M., Idris, A., Iqbal, T., Martínez-Álvarez, F.: Short term earthquake prediction model
using support vector regressor and hybrid neural networks. PloS One 13(7), e0199004 (2018)
23. Kubo, H., Kunugi, T., Suzuki, W., Suzuki, S., Aoi, S.: Hybrid predictor for ground-motion
intensity with machine learning and conventional ground motion prediction equation. Sci.
Rep. 10(1), 1–12 (2020)
24. Ancheta, T.D., Darragh, R.B., Stewart, J.P., Seyhan, E., Silva, W.J., Chiou, B.S-J., Wooddell,
K.E., Graves, R.W., Kottke, A.R., Boore, D.M., et al.: NGA-West2 database. Earthq. Spectra
30(3), 989–1005 (2014)
25. Vapnik, V.: The Nature of Statistical Learning Theory. (Springer Science & Business media,
1999)
26. Asim, K.M., Idris, A., Martínez-Álvarez, F., Iqbal, T.: Short term earthquake prediction in
Hindukush region using tree based ensemble learning, in 2016 International Conference on
Frontiers of Information Technology (FIT) (2016), pp. 365–370. https://doi.org/10.1109/FIT.
2016.073
27. Ho, T.K.: Random decision forests, in Proceedings of 3rd international conference on document
analysis and recognition, vol. 1. (IEEE, 1995), pp. 278–282
28. Amit, Y., Geman, D.: Shape quantization and recognition with randomized trees. Neural
Comput. 9(7), 1545–1588 (1997)
29. Campbell, K.W., Bozorgnia, Y.: NGA-West2 ground motion model for the average horizontal
components of PGA, PGV, and 5% damped linear acceleration response spectra. Earthq. Spectra
30(3), 1087–1115 (2014)
Study of an Anomalous Behavior
of Atmospheric Parameters—As
an Earthquake Precursors
for Himalayan Region Earthquakes
M. Senthil Kumar and Natarajan Venkatanathan
Abstract This research work concentrates on how the atmospheric parameters act
as a precursor of earthquakes occurred in the Himalayan region. In the past decade,
several literature results suggest there is a link between the observation of atmospheric
phenomenon and the earthquake occurrence. In this study, we have chooses to analyze
the outgoing longwave radiation (OLR) and relative humidity (RH) scenario prior
to the occurrence of earthquakes. Eight notable earthquakes (M ≥ 5.0) are selected
that happened in the Himalayan region, along the Main Central Thrust (MCT). The
Himalayas are emphasized by one of the world’s largest continental megathrusts,
which causes the occurrence of massive seismic events that produce significant
destruction. The atmospheric parameters like OLR and RH data sets were derived
from satellite observation, and variations were identified using the Z-factor method.
The temporal analysis of the atmospheric parameters was done for six months before
the occurrence of the earthquake. An abnormal deviation was observed in OLR and
RH before the occurrence of the earthquake. From the observation, all the anomalies
present in the atmospheric parameters were observed ten days prior to the occurrence
of the earthquake except for two events, which shows the heterogeneous nature of
the earthquake process. For all earthquakes, the RH value dropped predominantly
before observing the anomaly rise of the OLR. Similarly, the pattern of these atmo-
spheric parameters indicates sudden drops of RH value followed by the sudden rise
in the OLR, which is proposed by the LAIC model. Thus, the authors concluded that
atmospheric parameters like OLR and RH are the important short-term earthquake
precursor to forecast the earthquakes.
Keywords Outgoing longwave radiation (OLR) · Relative humidity (RH) ·

Earthquake · Earthquake precursors
M. Senthil Kumar · N. Venkatanathan (B)

SASTRA Deemed to Be University, Thanjavur, Tamil Nadu 613401, India
e-mail: physics16972@gmail.com; venkatanathan@eee.sastra.edu
https://doi.org/10.1007/978-981-99-1459-3_15
176 M. Senthil Kumar and N. Venkatanathan
1 Introduction
Considerable variations in relative humidity and temperature in the atmosphere are

detected before the occurrence of earthquakes. The abnormalities in these atmo-
spheric parameters have been identified for prolonged and short-term durations ahead
of the earthquakes. The short-term abnormalities show how the air temperature and
relative humidity patterns have changed before the event [1]. Analysis of meteo-
rological data before the strong earthquakes around the world gives a pathway for
the specific repetitive trend of humidity and air temperature changes [2]. Recent
researches revealed a strong link between radon levels and relative humidity [3].
Surface and near-surface temperature variations were reported before earthquakes
[4]. Relative humidity was observed 10 to 15 days prior to a significant earthquake
strike, according to Pulinets et al., in the zone of earthquake occurrence. The studies
of large earthquakes confirm an enhancement in air temperature and a decrease in
relative humidity observed over the seismic prone zone before the earthquake [5]. In
this study, we investigate an atmospheric phenomenon like outgoing longwave radia-
tion (OLR) and relative humidity (RH) since these parameters have shown anomalous
deviation for a chosen earthquake. For this, the Himalayan belt region earthquakes
were selected for the analysis. Because this region is among the most vulnerable
to continental collisions in the world [6–8]. Significant earthquakes (M ≥ 5.0) that
occurred in the Himalayan region were chosen for the analysis shown in Fig. 1.
Fig. 1 Epicenters of the earthquakes that occurred on the Himalayan belt region and red concentric
circles are represents the epicenter of the earthquakes
Study of an Anomalous Behavior of Atmospheric Parameters—As … 177
2 Methodology
2.1 Atmospheric Parameters
(Outgoing Longwave Radiation and Relative Humidity):

Outgoing longwave radiation is the thermal radiation due to the reflection of incoming
solar radiation on the earth’s surface into the atmosphere (OLR). The energy flux
carried by OLR leaving the earth’s surface into the atmosphere can be measured in
terms of W/m2 . With the help of satellite technology, the OLR has been measured
at the height of 10–12 km from the earth’s surface [9, 10]. An abnormal rise in OLR
flux was found around or over seismically active zones and extending 1000 sq. km
for short periods. Before the occurrence of major earthquakes, several experts have
identified these short-lived thermal anomalies. The unprocessed OLR data is filtered
for the clear sky scenario in the 10–13 μm wavelength range [9, 11]. Anomaly
variations were reported about a few days to months before the deadly earthquakes,
according to multiple studies on OLR flux. The abnormal fluctuations of OLR flux at
earthquake preparation zone has been identified by computing with the mean OLR
flux of preceding ten years are as follows:
Let us consider that “U” is the energy of the OLR left from the earth’s surface.
And “a” is latitude, and “b” is longitude where the energy was calculated, and “t” is
a time.
Σ
m
Ua,b,t
t=1
Ua,b,t = , (1)
m
where “m” is the number of defined past years for which mean OLR flux is measured
for a given location (a, b) and time (t).
Ua,b,t − Ua,b,t
dEindex flux (ΔUabt ) = , (2)
σa,b,t
where
ΔUabt dE_index flux for a given latitude (a), longitude (b) and time (t).
Uabt current OLR and RH flux computed for a given location (a, b) and time (t).
Uabt mean OLR and RH flux calculated for a given location (a, b) and time (t).
To efficiently show the anomalous spikes, we have to use the filtering formula
to filter the index value other than the anomalous one. The anomalous deviation
in energy flux index “[ΔUabt ]∗ ” is recognized by masking the energy flux index
(dE_index flux) value below + 2σ level of mean OLR flux.
If ΔUa,b,t ≥ Ua,b,t + 2,
then
∗
ΔUa,b,t = ΔUa,b,t ELSE ΔUa,b,t = 0, (3)
where
∗
ΔUa,b,t Anomalous energy flux index of OLR and RH observed for a given
location (a, b) & time (t).
Similarly, the above methodology is used for identifying the anomalous drop of
RH.
The impending earthquake zone can be determined using the “anomalous energy
flow index” analysis. In this study, the authors have used OLR data sets with a
resolution of 1° × 1° and RH data sets with a resolution of 2.5° × 2.5° provided by
the National Climatic Data Centre (NOAA) [11].
3 Results
The anomalous variations in relative humidity (RH), and the outgoing longwave
radiation (OLR) of the earthquakes that occurred on the Himalayan belt region were
analyzed and reported in this section.
An earthquake with a magnitude of 6.8 Mw occurred on October 19, 1991, in the
Garhwal Himalayas of Northern India, at the location 30.78N latitude and 78.774E
longitude [12]. The anomalous RH drop was observed on July 22, 1991, which was
89 days prior to the occurrence of the event has a dE_index flux is −3.102 W/m2 ,
and the anomalous variations of OLR was observed on August 12, 1991, i.e., 68 days
before the event with the dE_index flux value 3.0873 W/m2 shown in Fig. 2a. The
duration of the RH and OLR anomaly is 21 days.
The Chamoli Earthquake, which occurred on March 28, 1999, with an epicenter
in 30.512N latitude and 79.403E longitude, which is another significant earthquake
in Northern India from the perspective of the Himalayan region [13]. The anomalous
drop in RH was observed on March 04, 1999, has a dE_index flux of −2.997 W/m2
which was 24 days before the main event, and the anomalous variations in OLR were
recorded on March 19, 1999, it is found nine days prior to the earthquake and has a
dE_index flux was 2.992 W/m2 as shown in Fig. 2c. The duration of these anomalous
variations is 17 days.
On December 14, 2005, a significant earthquake occurred the Garhwal district
with the epicenter of the event being 30.476N in latitude and 79.225E in longitude,
which has been recorded with a magnitude of 5.1 Mw. The anomalous variations
in RH were observed on August 05, 2005, has a dE_index flux is −2.006 W/m2
which was nearly four months prior to the occurrence of the earthquake, and the
maximum OLR anomaly was recorded on August 08, 2005, has a dE_index flux is
4.4262 W/m2 , i.e., just three days after the observation of the RH drop. The last OLR
Fig. 2 Anomalous variations of RH and OLR for the earthquake occurred on a October 19, 1991,
with a magnitude of 6.8, b January 05, 1997, with a magnitude of 5.6, c March 28, 1999, with
a magnitude of 6.6, d December 14, 2005, with a magnitude of 5.1, September 21, 2009, with a
magnitude of e 6.1 and f 5.1, g April 04, 2011, with a magnitude of 5.3, and h February 06, 2017,
with a magnitude of 5. 1
anomaly observation was found on September 06, 2005, which was 99 days before
the main event as shown in Fig. 2d.
Two earthquakes that occurred on September 29, 2009, with different magnitude
lies on the MCT regions. One has occurred in Uttaranchal, India with the latitude
of 30.879°N and longitude of 79.057°E, an earthquake with a magnitude of 5.0Mw,
and another one that happened in Bhutan has a 27.332°N latitude and 91.437°E
longitude with the magnitude of 6.1Mw [14]. The anomaly drop of the RH was
found on June 25, 2009, which was just 88 days before the event, with the value
of −2.536 W/m2 for the earthquake having a magnitude of 6.1Mw (Fig. 2e). The
anomaly index of RH was found on August 28, 2009, which has the dE_index is −
2.624 W/m2 for a magnitude of 5.0 Mw earthquake, which was observed 24 days
before the earthquake. For the 6.1 Mw magnitude earthquake, the OLR anomaly was
found one month after the anomalous drop in RH, i.e., on August 02, 2009, with the
dE_index is 2.1911 W/m2 . The OLR anomalous was found on September 13, 2009,
with the value of 2.2994 W/m2 , which was found 08 days before the occurrence
of the earthquake and 16 days after the anomalous variations in RH for a 5.0 Mw
magnitude earthquake (Fig. 2f).
On April 04, 2011, the earthquake was occurred with the magnitude of 5.3Mw,
which was located on 29.698 latitude and 80.754 longitude [15]. The anomalous
drop in RH was observed on October 20, 2010, which was detected 166 days before
the occurrence of the earthquake, the anomaly drop index is −1.964 W/m2 . The
anomalous variation of OLR was found on November 30, 2010, with the dE_index
of 2.255 W/m2 (Fig. 2g) which observed 41 days after the occurrence of the RH
anomalous drop.
On February 06, 2017, an earthquake happened in Uttarakhand district, India with
a magnitude of 5.1 Mw. The epicenter of the event was reported to be 30.6544 latitude
and 79.1645 longitude. This event occurred, which is recognized as the seismic gap
in Type-1 is located in the Garhwal area of the NW Himalayas with the possibility
to give a rise of a magnitude Mw ~ 8.0 earthquake [16]. The anomalous variations
of RH and OLR were detected between the 12 days duration. The RH anomalous
drop found on October 10, 2016, and the OLR anomaly was observed on October
22, 2016, which have the values of −2.243 W/m2 and 2.0532 W/m2 , respectively
(Fig. 2h).
4 Discussion
Our research into these earthquakes revealed certain connections between atmo-
spheric parameters and seismicity. We were unable to draw any conclusions from
the overall correlation between earthquake parameters and atmospheric parameters
because of the earth’s heterogeneity. The correlation matrix is presented in Table 1,
which illustrates clearly that there is no correlation between atmospheric parameters
and earthquake parameters such as magnitude, depth, and epicenter. Therefore, we
opted for one of the earthquake parameters as a constant to establish a cluster among
these earthquakes. The cluster is separated according to the strike of the fault. The
same strike of the earthquakes followed one another in the same cluster. There are
three clusters among these eight earthquakes, which are detailed in Table 2.
Table 2 is the information and values of the various earthquake clusters’ atmo-
spheric parameters. After analyzing the correlation matrix for each cluster, it is clear
that the earthquake parameters and atmospheric parameters have a strong inverse
relationship.
The relationship between the magnitude of the earthquake and the atmospheric
parameter is shown in Fig. 3a–f. The atmospheric parameters include the maximum
dE index flux of RH & OLR, as well as the day difference. The day difference is
measured by the difference between the day of the earthquake and the last day of
the anomalous variations observed in RH and OLR. The correlation matrix between
earthquake characteristics and atmospheric parameters is presented in Table 3 for
Table 1 Correlation between the magnitude and dE_Index flux value of OLR and RH and time
duration of an anomaly as well as the time gap between the event date from the last observation of
the anomaly of the earthquake precursor for the earthquakes that occurred on the Himalayan belt
region
Latitude Longitude Depth Strike Magnitude
Magnitude −0.04311 0.122678 −0.67553 0.365569 1
dE_Index_OLR 0.043027 −0.09683 0.672893 0.40198 −0.33843
dE_Index_RH −0.22288 0.065747 0.444199 0.436736 −0.05299
EQ_RH −0.13734 0.064785 0.38025 0.653079 −0.42218
EQ_OLR 0.003871 −0.04972 0.296356 0.661857 −0.40744
Table 2 Earthquakes cluster list based on the strike of the fault

EQ_Date Lat Long Strike Mag dE dE EQ_RH EQ_OLR
Index OLR Index RH
Cluster-1
02–06–2017 30.65 79.16 279 5.1 2.540 −3.978 58 46
01–05–1997 29.84 80.53 280 5.6 2.777 −2.442 43 4
Cluster-2
03–28–1999 30.51 79.40 290 6.6 2.992 −2.977 24 7
12–14–2005 30.47 79.25 293 5.1 4.426 −2.835 131 99
09–21–2009 27.33 91.43 297 6.1 3.075 −2.984 88 50
Cluster-3
04–04–2011 29.69 80.75 318 5.3 3.818 −1.964 166 125
10–19–1991 30.78 78.77 320 6.8 3.087 −3.102 89 67
EQ_DATE Earthquake occurrence date
EQ_RH Day difference of the anomaly observed in RH and earthquake date
EQ_OLR Day difference of the anomaly observed in OLR and earthquake date
Fig. 3 Relation between the magnitude of the earthquake and the atmospheric parameters
three different clusters. And it is evident that there is a strong negative correlation
between earthquake magnitude and atmospheric parameters.
Figure 4 describes the day difference between the anomalous observation of RH
and OLR. In this bar diagram, the gray color indicates the day difference between the
anomaly of RH and the event date and the yellow color exhibits the day difference
between the day of the anomaly present in the OLR and the event date. From this
graph, we clearly understand that the anomaly of the RH was found first before the
anomaly was observed in the OLR. Changes in relative humidity and air temperature
are the atmospheric phenomena experienced closest to the surface of the earth. As
a result of the release of gas/radon from the underneath subsurface, the rise in the
temperature is instantly apparent on the ground’s surface. This could be possibly
due to the increased radon, and other gas emissions are caused by air ionization,
which leads to thermal effects such as changes in relative humidity, air temperature
and outgoing longwave radiation. Horizontal air movements, air mixture, a rise in
air temperature, and the temperature difference between the fault’s region and the
region distant from the faults would be the possible triggering effects for originating
the entire earthquake preparation zone [17]. And our model obeys the LAIC model
[18].
Table 3 Correlation between the magnitude and dE_Index flux value of OLR and RH and time
duration of an anomaly as well as the time gap between the event date from the last observation of
the anomaly of the earthquake precursor for the earthquakes that occurred on the Himalayan belt
region with different clusters
Latitude Longitude Depth Strike Magnitude
Cluster_1
Magnitude −1 1 1 1 1
dE_Index_OLR −1 1 1 1 1
dE_Index_RH −1 1 1 1 1
EQ_RH 1 −1 −1 −1 −1
EQ_OLR 1 −1 −1 −1 −1
Cluster_2
Magnitude 1 −1 −1 1 1
dE_Index_OLR −1 1 1 −1 −1
dE_Index_RH −1 1 1 −1 −1
EQ_RH −1 1 1 −1 −1
EQ_OLR −1 1 1 −1 −1
Cluster_3
Magnitude −0.179 0.199 −0.935 −0.249 1.000
dE_Index_OLR 0.446 −0.464 0.997 −0.031 −0.960
dE_Index_RH 0.531 −0.548 1.000 −0.128 −0.929
EQ_RH −0.122 0.102 0.786 0.526 −0.954
EQ_OLR 0.028 −0.048 0.870 0.393 −0.988
RH VS OLR EQ_RH EQ_OLR
300
125
Number of days
200 99
67 50
100 46 166
131
4 89 88
58 43 7 8
24 24
0
Earthquake date
Fig. 4 Reveals the time duration of occurrence of the RH, and OLR before the events for all
earthquakes
5 Conclusion
We all know that the earth is diverse characteristics in each and every place, and
the parameters used to measure the earth’s features vary from place to place due
to the earth’s heterogeneous nature. So, we cannot say for certain that this will be
a factor for this cause. However, in this case, we attempted to have atmospheric
parameters work as one of the precursors for earthquake forecasting techniques for
the entire region but we cannot due to the divergent character of the earth. The
value of the precursors may vary due to the earth’s nature, however, when there is an
anomaly in the atmospheric parameters like RH and OLR in the atmosphere, there is a
cause earthquake near the region. The atmospheric parameters data sets were derived
from satellite observation, and the anomalous variations were identified using the Z-
factor method. The temporal analysis of the atmospheric parameters was done for
six months before the occurrence of the earthquake. The anomalous variations were
detected for RH is below the −2sigma level and for OLR the anomalous variations are
the above 2sigma level. We found there was an anomalous deviation in RH and OLR
before the occurrence of the earthquake. From the observation, for all earthquakes,
the RH value dropped predominantly before observing the anomaly rise of the OLR.
Similarly, in our model, the pattern of these atmospheric parameters indicates sudden
drops of RH value followed by the sudden rise in the OLR, which is proposed by the
LAIC model. The duration period for these two atmospheric parameters is within a
month. Due to the divergent nature of the earth, the overall correlation matrix does
not have any conclude evidence between the atmospheric parameters and earthquake
variables. So, we have divided the earthquakes based on the strike of the fault and
got 3 different clusters. For these three different clusters, we found that there is
relationship using correlation matrix. The magnitude of the earthquake is strong
negative correlation between the dE_index flux value of the RH and OLR as well as
the day difference also negatively correlated with the magnitude of the earthquake.
Thus, the authors concluded that atmospheric parameters like OLR and RH are the
important short-term earthquake precursor to forecast the earthquakes.
6 Funding and Acknowledgments
The authors are thankful to the Ministry of Earth Sciences, India for financial assis-
tance (Project No: MoES/P. O(seismo)/1(343)/2018) to carry-out research on earth-
quake forecasting. We would like to acknowledge SASTRA Deemed to be University
for facilitating and encouraging to do this research work.
References
1. Dunajecka, M.A., Pulinets, S.: Atmospheric and thermal anomalies observed around the time
of strong earthquakes in Mexico. Atmosfera 18, 233–247 (2005)
2. Pulinets, S.A., Dunajecka, M.A.: Specific variations of air temperature and relative humidity
around the time of Michoacán earthquake M8.1 Sept 19, 1985 as a possible indicator of
interaction between tectonic plates. Tectonophysics 431, 221–230 (2007)
3. Prasad, B.S.N., Nagaraja, T.K., Chandrashekara, M.S., Paramesh, L., Madhava, M.S.: Diurnal
and seasonal variations of radioactivity and electrical conductivity near the surface for a
continental location Mysore, India. Atmos. Res. 76, 65–77 (2005)
4. Tronin, A.A., Hayakawa, M., Molchanov, O.A.: Thermal IR satellite data application for
earthquake research in Japan and China. J. Geodyn. 33(2004), 519–534 (2004)
5. Oikonomou, C., Haralambous, H., Pulinets, S., Khadka, A., Paudel, S.R., Barta, V., Muslim, B.,
Kourtidis, K., Karagioras, A., Inyurt, S.: 2021 Investigation of pre-earthquake ionospheric and
atmospheric disturbances for three large earthquakes in Mexico. Geosciences 11, 16 (2021).
https://doi.org/10.3390/geosciences11010016
6. Bilham, R.: Himalayan earthquakes: a review of historical seismicity and early 21st century
slip potential. Geol. Soc., Lond., Spec. Publ. 483 (2019)
7. Mishra, O.P.: Intricacies of the Himalayan seismotectonics and seismogenesis: need for inte-
grated research. Curr. Sci. Assoc. 106(2), 176–87 (2014). http://www.jstor.org/stable/240
99798
8. Kornei, K.: Geodetic data pinpoint earthquake-prone regions of the Himalayas. Eos. 101 (2020).
https://doi.org/10.1029/2020EO141610. Published on 19 March 2020
9. Fu, C.-C., Lee, L.-C., Ouzounov, D., Jan, J.-C.: Earth’s outgoing longwave radiation variability
prior to M ≥ 6.0 earthquakes in the Taiwan area during 2009–2019. Front. Earth Sci. 8 (2020).
https://doi.org/10.3389/feart.2020.00364
10. https://www.weather.gov/jetstream/energy
11. Lee, H.-T.: NOAA CDR Program, 2011: NOAA Climate Data Record (CDR) of Daily Outgoing
Longwave Radiation (OLR), Version 1.2. [indicate subset used]. NOAA National Climatic Data
Center. https://doi.org/10.7289/V5SJ1HH2 [access date]
12. Rajendran, K., Rajendran, C.P., Jain, S.K., Murty, C.V.R., Arlekar, J.N.: The Chamoli earth-
quake, Garhwal Himalaya: field observations and implications for seismic hazard. Curr. Sci.
Bangalore 78(1), 45–51 (2000)
13. Shrikhande, M., Rai, D., Narayan, J.P., Das, J.: The March 29, 1999 earthquake at Chamoli,
India, 12 WCEE 2000: 12th world conference on earthquake engineering; Auckland, New
Zealand, Sunday 30 January–Friday 4 February 2000. Structural engineering 0779–1131, vol.
10 (2000)
14. Kayal, J.R., Arefiev, S.S., Baruah, S., Tatevossian, R., Gogoi, N., Sanoujam, M., Gautam, J.L.,
Hazarika, D., Borah, D.: The 2009 Bhutan and Assam felt earthquakes (Mw 6.3 and 5.1) at the
Kopili fault in the northeast Himalaya region geomatics. Natural Hazards Risk (2010) ISSN
1947-5705 Print/ISSN 1947-5713 Online@2010 Taylor & Francis http://www.tandf.co.uk/jou
rnals https://doi.org/10.1080/19475705.2010.486561
15. Earthquake catalogue NEIC, USGS, https://earthquake.usgs.gov/earthquakes/search/
16. Srivastava, H.N., Verma, M., Bansal, B.K., Sutar, A.K.: Discriminatory characteristics of
seismic gaps in Himalaya. Geomatics Nat. Hazards Risk 6, 224–242 (2013). https://doi.org/
10.1080/19475705.2013.839483
17. Ouzounov, D., Pulinets, S., Hattori, K., Kafatos, M., Taylor, P.: Atmospheric Signals Associated
with Major Earthquakes. A Multi-Sensor Approach, Electromagnetic Phenomena Associated
with Earthquakes, Transworld Research Network
18. Pulinets, S., Ouzounov, D.: Lithosphere–Atmosphere–Ionosphere Coupling (LAIC) model—
an unified concept for earthquake precursors validation. J. Asian Earth Sci. 41(2011), 371–382
(2011). https://doi.org/10.1016/j.jseaes.2010.03.005
Probabilistic Seismic Hazard Assessment
of North East India
C. Lallawmawma, M. L. Sharma, and J. Das
Abstract Probabilistic seismic hazard assessment of the North East Indian region
is carried out considering three models of seismic source, i.e., areal sources, line
sources, and smoothed gridded seismicity model. Between 87°–98°E and 20°–30°N,
earthquake catalogs and various tectonic features are collected from multiple sources
which produce a seismotectonic map. Based on seismicity, tectonic provinces, and
fault rupture mechanism, the study region is divided into six area source zones. Thirty-
two identified fault sources are modeled as linear source models; the gridded seis-
micity model is modeled as a point source model. For each source model, seismicity
parameters are calculated by considering only the completed earthquake catalog.
Four next-generation (NGA) ground motion models are applied to estimate the
hazard at the reference rock condition. The logic tree framework is implemented
in the source models and GMPEs to account for the epistemic uncertainties. Peak
ground acceleration (PGA) and spectral acceleration (Sa) are estimated at different
cities of Northeastern states for a 2% and 10% probability of exceedance in 50 years.
The hazard curves and uniform hazard spectra are also presented.
Keywords PSHA · Logic tree · OpenQuake
1 Introduction
The North East Indian region being overthrust by the Eastern Himalaya in the North-
Northeast and the Burmese arc in the East-Southeast forms one of the most complex
C. Lallawmawma (B) · M. L. Sharma · J. Das

Department of Earthquake Engineering, IIT Roorkee, Roorkee, India
e-mail: clallawmawma1@eq.iitr.ac.in
M. L. Sharma
e-mail: m.sharma@eq.iitr.ac.in
J. Das
e-mail: jddasfeq@iitr.ac.in
https://doi.org/10.1007/978-981-99-1459-3_16
188 C. Lallawmawma et al.
tectonic provinces in the world [1]. With a complex tectonic and geological setup,
this region has experienced several moderate-large earthquakes in the past, including
January 10, 1869, Cachar (Mw 7.4), June 12, 1897, Shillong (Mw 8), August 15,
1950, Assam (Mw 8.7) and September 18, 2011, Sikkim (Mw 6.9). With high popu-
lation, rapid urbanization, and poor construction practices in this high earthquake-
prone region, it becomes imperative to quantify the severity of ground shaking that
can be expected in the area, which incorporates a great deal of earthquake uncertainty
in location, size, and resulting ground motion of future earthquakes.
The PSHA of any region needs to be revisited or updated by accommodating
the latest knowledge about the seismotectonic of the area, incorporating improved
attenuation relations, and taking advantage of the prevailing advanced analysis tech-
niques [2]. In this study, seismicity parameters and seismic hazards are estimated
for seven cities in NE states using different source models and GMPEs. To account
for epistemic uncertainties, logic tree procedure is applied. OpenQuake engine [3] is
used to perform the hazard analysis. The comparison of results obtained from each
source model and GMPEs along with the combined model is also presented.
2 Seismicity and Seismotectonics
2.1 Tectonic Provinces
Northeastern India and its neighborhood comprise a highly seismically active region,
which contains many major tectonic provinces. The seismically active northeast
Indian region falls at the junction of NS trending Burmese arc and EW trending
Himalayan arc resulting in numerous geological structures [4]. As per the seismic
zoning map of India given in the earthquake-resistant design code of India [BIS
1893(Part I) 2016], most of the northeastern states have been placed in seismic
zone V except for Sikkim, which falls in zone IV. The primary tectonic background
includes the eastern Himalayan structures, the Mishmi massif, the Indo-Myanmar
arc, the Surma Basin, and the Shillong plateau [5]. As a collision plate boundary,
the Himalayan tectonic zone is characterized by a series of north dipping thrusts
that are exposed at the surface [6]. The Himalayan structures mainly consist of
the thrust planes namely the Main Central Thrust (MCT), Main Boundary Thrust
(MBT), Main Frontal Thrust (MFT), and their subsidiaries [5, 7]. The MBT separates
the lesser Himalaya from the sub Himalayan belt, whereas the MCT separates the
higher Himalaya from the lesser Himalaya [8]. The eastern syntaxis zone, where the
Burmese arc and the Himalaya arc meet the Mishmi Block, is a significant tectonic
domain. The Indo-Myanmar is formed by the Indian plate subducting beneath the
Sunda plate in the east. The region is characterized by high seismicity. Surma Basin
is an area of folded sediments characterized by westerly convex, sinuous structural
ridges, and valleys [9]. Lineaments/faults running NE-SW and NW–SE are quite
prevalent. Shillong Mikir Plateau, which is considered the plate-boundary zone,
Probabilistic Seismic Hazard Assessment of North East India 189
major geotectonic structures connected in the region are the EW Dauki fault, the
Brahmaputra lineament to the north, NE-SW to the east is the Naga thrust, while to the
west is the Dhubri fault. In this study, the seismotectonic map has been prepared using
ArcGIS [10] mapping tool platform, the tectonic features (faults and lineaments)
given in SEISAT (2000) and from published literature [11–14] were digitized and
imposed with a collected earthquake catalog to prepare seismotectonic map as shown
in Fig. 1.
2.2 Earthquake Database and Treatment of Catalog
Study region ranging between 20°–30° and 87°–98°, earthquake catalog is compiled
from different agencies namely United States Geological Survey (USGS) [15], Indian
Meteorological Department (IMD) [16], International Seismological Center (ISC)
[17], and Global Centroid Moment Tensor (GCMT) [18]. Historical earthquake
records are collected from the published study of Raghukanth [19]. A total of 4792
events are collected from 825 to 2020 with detailed information about the date and
time of occurrence, magnitude, and location with specified coordinates. Compiled
catalogs are declustered and homogenized in a unified moment magnitude scale
using Scordilis [20] and Das [21]. Magnitude of completeness (Mc) is estimated to
assess for completeness in terms of size, and to ensure completeness in terms of time,
Stepp’s [22] approach is used.
3 Seismic Source Models
3.1 Areal Sources
Area sources are the most widely used type of source zone in the PSHA studies. In
many regions, it is known that active faults exist, but it is not possible to characterize
individual faults as sources. In such cases, it is common to define an area source that
aggregates the seismic activity over some spatial region. Area sources are typically a
polygon in which earthquakes can occur anywhere and at any time inside the desig-
nated zone. In terms of time and space, these zones are believed to have consistent
source properties. Based on seismicity, tectonic features, fault rupture mechanism,
and published literature [11–14], the region is grouped into six seismogenic source
zones as shown in Fig. 2.
Source Zone 1 corresponds to the central low lands of Myanmar. The primary
tectonic feature is the Sagaing Fault, which runs for nearly 1000 km from north to
south and is one of Myanmar’s most notable active strike-slip faults. The largest
magnitude observed is Mw 7.8 occurred on June 1, 1787; most of the events show a
strike-slip mechanism.
Fig. 1 Seismotectonic map of the study area showing major fault such as ShanSagaining fault,
Churachanpur Mau fault (CMF), Eastern boundary thrust (EBT), Kabaw fault, F1, Mat fault, Chit-
tagong coastal fault (CCF), Eocene hinge zone (EHZ), Kaladan, Madhupur blind fault, Sylhet fault,
Barapani shear zone, Dauki fault, Dhubri fault, Dudhnoi fault, Bomdila fault, Kopili and F2 fault,
Naga and Disang thrust, Oldham fault, Samin and Chedrang fault, Arun and Purnea fault, BNS
fault, Dudhkosi fault, Indus Suture fault (IST), main boundary thrust (MBT), Main central thrust
(MCT), Lohit thrust, and Mishmi thrust
Fig. 2 Six seismogenic source zones considered
Source Zone 2 represents the Indo-Myanmar subduction zone, about 15 significant

earthquakes (Mw > 6.5) have struck the region in the previous 100 years, making
it a seismically active region. The main tectonic features are the Eastern Boundary
Thrust (EBT) or Kabaw fault and the Churachanpur Mao fault (CMF). Reverse and
Strike-Slip mechanisms are largely observed.
Source Zone 3 corresponds to the Bengal Basin and part of the Mizoram and
Tripura fold belt. Bengal basin is characterized by low seismicity, while Mizoram
and Tripura folded belt shows moderate seismic activity. Srimangal earthquake (Mw
7.6) of July 8, 1918, occurred in this zone. The main tectonic features are Chittagong
Coastal fault (CCF), NE trending Sylhet fault, Kaladan fault, Eastern Hinge zone,
and other NW trending faults. Events show strike-slip and reverse mechanisms.
Source Zone 4 represents Meghalaya Plateau, Mikir Hills, and Naga-Disang
thrust. Meghalaya plateau is bounded to the north by the Arunachal Himalaya,
Bengal basin, and Surma basin to the southern part, and to the western part by
Dhubri fault. Seismic activity is considered plate-boundary zone activity. The Mikir
Hill is surrounded by Bomdila Fault to the east and the Kopili Fault to the west, char-
acterized by a strike-slip mechanism. Large magnitude earthquake Mw 8.1 occurred
on June 26, 1897.
Source Zone 5 represent Eastern Himalaya zone Indo-Tsangpo suture zone. The
main tectonic features are Main Boundary Thrust (MBT) and Main Central thrust
(MCT). Most of the seismicity in the Himalayan region is concentrated around the
MBT and MCT [23]. The maximum magnitude observed in this zone is Mw 7.7.
Source Zone 6 is the Mishmi Massif region, an NW–SE trending feature. The
main tectonic features are Lohit thrust, Mishmi thrust tiding suture, and Pochu fault.
The maximum magnitude observed in this area is Mw 8.5 occurred on August 15,
1950.
3.2 Linear Sources
A fault source is the most prominent representation of a seismic source. Tectonic

features for North East India and surrounding areas are primarily obtained from the
Seismotectonic Atlas (SEISAT 2000) prepared by the Geological Survey of India
(GSI) and published literature [11–14]. The information about fault focal mecha-
nisms is obtained from the literature and catalogs extracted from GCMT [18]. A
total of 32 linear sources are used in this study.
3.3 Gridded Seismicity Sources
To overcome the subjectivity that the traditional area source model introduces and to
model sources in the lack of clear recognized seismic sources, the gridded seismicity
model, Frankel [2] suggested a zoneless spatially smoothing approach based on the
seismic activity rate obtained from the catalog. In this study, this approach is applied.
The study region is divided into grids (in this case, 0.2° × 0.2°). The number of
earthquakes with a magnitude higher than a cutoff magnitude in each grid is counted,
and the activity rate for each grid cell is smoothed utilizing a Gaussian function. The
background smoothed activity rates (10a values) obtained using the Frankel (1995)
approach is shown in Fig. 3.
Fig. 3 Smoothed activity rate 10a values
4 Seismicity Parameters and Maximum Magnitude
The recurrence model parameters determine the number of earthquakes of various

magnitudes that occur each year. The level of seismicity of a region is assessed
from the seismic parameters a, b and Mc. In this study, the Gutenberg–Richter (GR)
magnitude–frequency relationship [24] to obtain the parameters of seismicity ‘a’
value and ‘b’ value. For the area source zone, after determining the completeness of
the catalog reported in each zones, seismicity parameters are estimated. For linear
sources, 30–40 km is measured from the traced fault line, creating a polygon for each
fault. In a similar approach to the area source, seismicity parameters are estimated.
The seismicity rates (10a ) at each position in a grid of 0.2° × 0.2° are obtained
using the Frankel [2] spatially smoothing technique for gridded sources. The value
of ‘b’ is calculated using the complete earthquake catalog within the boundary of the
study region. For background seismicity modeling, the computed value of b = 0.87
is considered to be consistent. Maximum magnitude is computed using the equation
given by KijKo [25] (Tables 1 and 2).
5 Ground Motion Prediction Equation
A ground motion prediction equation (GMPE) is one of the prerequisites for seismic
hazard evaluations, since it transforms event information (such as earthquake loca-
tion and magnitude) into site parameters that characterize the seismic hazard at a site
(e.g., peak ground acceleration, PGA) [26]. The ground motions are estimated using
GMPEs, and essential parameters are the source-to-site distance, magnitude, type
of faulting, and site response. Generally, the GMPEs are empirically derived from
the recorded earthquake ground motion data using statistical regression analysis.
Region specific GMPEs are preferred and are desirable to use for ground motion
estimates. For the regions where strong motion data is not available for such anal-
ysis, the GMPEs developed for other regions are used based on the resemblance of
the characteristics of both regions [27]. Five sets of ground motion prediction equa-
tions (GMPEs) developed by PEER next-generation (NGA) models updated in 2014
implementing more latest and worldwide ground motion data are selected and used
in this study. The GMPEs are Abrahamson, Silva, and Kamai (ASK14) [28], Boore,
Stewart, Seyhan, and Atkinson (BSSA14) [29], Campbell and Bozorgnia (CB14)
[30], Chiou and Youngs (CY2014) [31], and Idriss (I14) [32]. However, Chiou and
Youngs (CY2014) overestimate the hazard. In the study, the other four GMPEs are
used for hazard analysis.
6 Logic Tree Structure
Logic trees provide a convenient form for formal and quantitative treatment of uncer-
tainties [33]. The main uncertainties associated are modeling the seismogenic sources
and the ground motion prediction models used in estimating the hazard. A logic tree
is made of a series of nodes and branches, these branches represent several models
that have been given a pre-determined weight, with the overall weighting factor at
the end branch equaling the total of the weights allocated to individual branches. In
the study, the logic tree comprises of three seismic source models and four GMPEs
Table 1 Seismicity and other required parameters for area sources hazard calculation
Source Mmax Duration of Zmap (maximum likelihood) Source parameters Focal mechanism
(obs) catalog Mc a B Mmax Depth Hypo depth Aspect Dip/strike/rake-probability
range (km)/probability ratio
(km)
SZ1 8 181 4.3 3.108 0.64 ± 0.03 8.5 ± 0.15 0–100 10/0.33 1.0 78/−30/122–0.33
50/0.34 90/74/186–0.34
98/0.33 68/166/2–0.33
SZ2 7.3 140 4.2 4.659 0.85 ± 0.03 7.8 ± 0.15 0–100 10/0.3 1.0 75/178/5–0.33
50/0.34 28/99/44–0.34
90/0.3 54/120/148–0.33
SZ3 7.8 283 4.3 2.925 0.69 ± 0.05 8.30 ± 0.18 0–60 10/0.33 1.0 85/−2/63–0.33
30/0.34 34/−82/179–0.34
55/0.33 37/168/163–0.33
SZ4 8.1 472 4.3 2.856 0.8 ± 0.05 8.6 ± 0.25 0–50 10/0.33 1.0 53/44/89–0.33
Probabilistic Seismic Hazard Assessment of North East India
25/0.34 41/83/91–0.34
46/0.33 55/47/85–0.33
SZ5 7.7 609 4.1 2.704 0.66 ± 0.04 8.2 ± 0.22 0–90 12/0.33 1.0 62/179/109–0.33
50/0.34 45/−90/21–0.34
83/0.33 72/−12/216–0.33
SZ6 8.5 177 4.2 3.029 0.68 ± 0.06 9.0 ± 0.25 0–50 10/0.33 1.0 63/−51/86–0.33
30/0.34 86/−11/35–0.34
50/0.33 24/72/119–0.33
195
Table 2 Rupture percent and source model parameters for considered faults in the study area
196
Name of fault Rupture characteristics Depth range Dip Rake a b

Faulting Observed Mw Total fault length Rupture length Rupture percent (Km)
(Km) (%)
Source 1
ShanSagaining SS 7.8 520 166.72 32.06 0–40 68 166 2.002 0.53 ± 0.04
Source 2
CMF SS 7.2 305 59.98 19.67 0–60 78 42 3.063 0.72 ± 0.06
EBT or Kabaw SS 7.2 280 59.98 21.42 0–80 32 −8 3.81 0.76 ± 0.03
F1 All 6.1 296 9.75 3.29 0–100 89 160 3.07 0.76 ± 0.05
Mat SS 5.6 121 3.93 3.25 0–80 70 175 3.001 0.78 ± 0.10
EBT (portion) Thrust 7 338 35.48 10.41 0–100 54 120 3.497 0.69 ± 0.02
Source 3
CCF SS 7.5 380 100 26.32 0–40 51 87 2.376 0.76 ± 0.10
EHZ All 7.8 320 145.21 45.38 0–60 34 −82 0.717 0.44 ± 0.09
Kaladan Reverse 7.5 424 73.28 17.28 0–100 37 168 2.378 0.63 ± 0.05
Madhupur All 7 68 40.74 59.91 0–50 77 171 1.875 0.68 ± 0.14
Sylhet SS 7.6 158 118.58 75.05 0–50 88 166 2.062 0.57 ± 0.08
Source 4
Barapani Thrust 5.7 50 5.16 10.32 0–40 79 158 2.218 0.82 ± 0.22
Dauki R&SS 7.1 320 41.02 12.82 0–50 55 88 2.444 0.73 ± 0.16
Dhubri SS 7.1 192 50.58 26.34 0–50 87 162 2.125 0.65 ± 0.14
Dudhnoi All 5.4 94 3.21 3.41 0–50 55 88 3.25 0.86 ± 0.14
Bomdila SS 5.7 72 4.66 6.47 0–50 83 7 2.18 0.63 ± 0.07
(continued)
C. Lallawmawma et al.
Table 2 (continued)
Name of fault Rupture characteristics Depth range Dip Rake a b
Faulting Observed Mw Total fault length Rupture length Rupture percent (Km)
(Km) (%)
Kopili and F2 SS 7.2 300 59.98 19.99 0–50 55 47 2.017 0.63 ± 0.08
Naga Disang Reverse 6.5 400 17.18 4.30 0–50 55 88 1.407 0.64 ± 0.14
Thrust
Oldham Reverse 8.1 128 174.98 136.70 0–50 79 167 2.287 0.72 ± 0.15
Samin and Reverse 6.8 22 26.55 120.68 0–40 40 83 2.186 0.71 ± 0.18
Chedrang
Source 5
Arun and All 6 136 8.32 6.12 0–80 52 −62 2.278 0.62 ± 0.06
Purnea
BNS SS 7.7 274 140 51.09 0–40 42 −94 0.874 0.48 ± 0.06
Dudhkosi All 5.7 43 5.16 12 0–60 45 −46 3.578 0.89 ± 0.10
Indus Suture All 7.7 540 123.88 22.94 0–80 76 97 1.421 0.51 ± 0.08
Probabilistic Seismic Hazard Assessment of North East India
MBT and MCT Thrust 7.2 790 47.42 6 0–60 72 −12 2.437 0.60 ± 0.02
Source 6
Lohit Thrust Thrust 8.5 208 312.61 150.29 0–50 24 72 2.279 0.60 ± 0.08
Mishmi Thrust Thrust 7 279 35.48 12.72 0–50 87 99 2.334 0.62 ± 0.04
197
Fig. 4 Logic tree framework used in PSHA for source models and GMPEs
shown in Fig. 4. By allocating proper weightage to each branch, a total of twelve

branches are adopted.
7 Probabilistic Seismic Hazard Analysis and Results
Seismic hazard computes the expected ground motion at a given site considering
the seismicity rate, geology, seismotectonic, and the attenuation characteristics of
the seismic waves in the region. Probabilistic seismic hazard calculation has been
performed based on classical Cornell–McGuire method implementing the total prob-
ability theorem, OpenQuake engine [3] is used to perform the hazard analysis. For
seismic hazard computations, the whole study area is divided at a grid intervals of
size 0.1° × 0.1°, and the values of seismic hazard at the center point of each grid
cell are calculated. The peak ground acceleration (PGA) and spectral acceleration
(SA) at periods of 0.1, 0.2, 0.4, 0.7, 1.0, and 2.0 s at a 5% damping ratio at 10%
(475 years return period) and 2% (2475 years return period) probability of exceedance
in 50 years are estimated. The peak ground acceleration and spectral acceleration
at different periods are calculated using a reference bedrock condition of 760 m/s
average shear wave velocity in the top 30 m (Vs30). This reference site class relates
to a boundary between sites B and C specified by the National Earthquake Hazard
Reduction (NEHRP). Figure 5 presents the hazard map at 2% and 10% probabili-
ties of exceedance in 50 years. Figures 6 and 7 present the mean hazard curve and
uniform hazard spectra at seven cities of North East states, respectively (Table 3).
Fig. 5 Peak ground acceleration 2% and 10% exceedance in 50 years
8 Conclusions
This paper presents a probabilistic seismic hazard assessment of North East India.
The seismic hazard has been computed by performing hazard computations at a
grid interval of 0.1° × 0.1° using three source models. To account for epistemic
uncertainties in source models and GMPEs, the logic tree procedure is adopted. The
calculated PGA for a 10% exceedance in 50 years, which is equivalent to return
period of 475 years is compared with different source models and previous studies.
It is observed that the estimated seismic hazard is higher in Imphal than in other
cities due to the hazard contribution from the Churachanpur Mao fault and eastern
subduction seismic sources. Choice of GMPEs also greatly affects the hazard value
so selecting appropriate GMPEs is crucial in the hazard analysis. The seismic hazard
curve and uniform hazard spectrum have been evaluated for bedrock rock level.
Corresponding to 10% and 2% probability of exceedance in 50 years uniform hazard
spectrum for major cities at the NE states are also presented.
Fig. 6 Mean hazard curve for a Sa (2.0), b Sa (0.1), and c (0) seconds in 50 years
Fig. 7 Uniform hazard spectra at seven cities of NE states at a 10%, b 2% PE in 50 years

Table 3 Estimated peak ground acceleration and comparison with other studies in seven cities
NE cities Present study Previous studies
(location) PGA (in 10% PE in 50 years)
Area Linear Grid Combine Das et al. Nath et al. NDMA Sharma
source [21] [11] [12] and Malik
[34]
Agartala 0.09 0.13 0.16 0.13 0.217 0.5 0.18 0.3
(91.318,
23.785)
Aizawl 0.25 0.15 0.19 0.20 0.114 0.6 0.18 0.3
(92.69,
23.785)
Imphal 0.25 0.46 0.19 0.30 0.144 0.99 0.35 0.4
(93.929,
24.774)
Itanagar 0.09 0.15 0.13 0.12 0.182 0.7 0.3 0.44
(93.569,
27.113)
Shillong 0.08 0.13 0.12 0.11 0.316 1.10 0.25 0.5
(91.884,
25.584)
Kohima 0.20 0.22 0.15 0.19 0.148 0.6 0.2 0.48
(94.083,
25.674)
Despur 0.08 0.13 0.15 0.12 0.2 0.75 0.2 0.4
(91.807,
26.123)
References
1. Das, S., Gupta, I.D., Gupta, V.K.: A probabilistic seismic hazard analysis of Northeast India.
Earthq. Spectra 22, 1–27 (2006)
2. Frankel, A.: Mapping seismic hazard in the central and eastern United States. Seismol. Res.
Lett. 66, 8–21 (1995)
3. Pagani, M., et al.: Openquake engine: an open hazard (and risk) software for the global
earthquake model. Seismol. Res. Lett. 85, 692–702 (2014)
4. Sitharam, T.G., Kolathayar, S., James, N.: Probabilistic assessment of surface level seismic
hazard in India using topographic gradient as a proxy for site condition. Geosci. Front. 6,
847–859 (2015)
5. Kumar Nath, S., Kumar, K., Thingbaijam, S., Raj, A.: Earthquake hazard in Northeast India-A
seismic microzonation approach with typical case studies from Sikkim Himalaya and Guwahati
city. J. Earth. Syst. 117, 809–831 (2008)
6. Choudhary, C., Sharma, M.L.: Global strain rates in western to central Himalayas and their
implications in seismic hazard assessment. Nat. Hazards 94, 1211–1224 (2018)
7. Nandy, D.R.: Geodynamics of Northeast India and the Adjoining Region, 3rd edn. Abc
Publication, Kolkata (2001)
8. Harbindu, A., Sharma, M.L., Kamal.: Stochastic ground-motion simulation of two Himalayan
earthquakes: seismic hazard assessment perspective. J. Seismol. 16, 345–369 (2012)
9. Tiwari, R.P., et al.: No evidence for shallow shear motion on the Mat fault, a prominent strike
slip fault in the Indo-Burmese wedge. J. Earth Syst. Sci. 124, 1039–1046 (2015)
10. ESRI.: ArcGIS Desktop: Release 10. (Environmental Systems Research Institute, Redlands
2011)
11. Nath, S.K., Thingbaijam, K.K.S.: Probabilistic seismic hazard assessment of India. Seismol.
Res. Lett. 83(1), 135–149 (2012)
12. National Disaster Management Authority.: Development of Probabilistic Seismic Hazard Map
of India, Government of India, New Delhi (2011)
13. Baro, O., Kumar, A., Ismail-Zadeh, A.: Seismic hazard assessment of the Shillong Plateau,
India. Geomat. Nat. Haz. Risk 9(1), 841–861 (2018)
14. Bahuguna, A., Sil, A.: Comprehensive seismicity, seismic sources and seismic hazard
assessment of Assam, North East India. J. Earthquake Eng. 24(2), 254–297 (2018)
15. United States Geological Survey (USGS).: https://earthquake.usgs.gov/earthquakes/search/.
Accessed 2020
16. Indian Meteorological Department (IMD).: www.imd.gov.in/, New Delhi, India (Through
Personal communication)
17. International Seismological Center (ISC).: http://www.isc.ac.uk/iscbulletin/search/catalogue/.
Accessed 2020
18. Ekström, G., Nettles, M., Dziewonski, A.M.: The global CMT project 2004–2010: centroid-
moment tensors for 13,017 earthquakes. Phys. Earth Planet. Inter. 200–201, 1–9 (2012)
19. Raghukanth, S.T.G.: Estimation of seismicity parameters for India. Seismol. Res. Lett. 81,
207–217 (2010)
20. Scordilis, E.M.: Empirical global relations converting MS and mb to moment magnitude. J.
Seismol. 10, 225–236 (2006)
21. Das, R., Sharma, M.L., Wason, H.R.: Probabilistic seismic hazard assessment for Northeast
India region. Pure Appl. Geophys. 173, 2653–2670 (2016)
22. Stepp, J.C.: Analysis of completeness of the earthquake sample in the Puget sound area. Contrib.
Seism. Zo. 16–28 (1973)
23. Bajaj, S., Sharma, M.L.: Modeling earthquake recurrence in the himalayan seismic belt
using time-dependent stochastic models: implications for future seismic hazards. Pure Appl.
Geophys. 176, 5261–5278 (2019)
24. Gutenberg, B., Richter, C.F.: Earthquake magnitude intensity energy and acceleration. B.
Seismol. Soc. Am. 46, 105–145 (1956)
25. Kijko, A.: Estimation of the maximum earthquake magnitude, mmax . Pure Appl. Geophys. 161,
1655–1681 (2004)
26. Sharma, M.L., Douglas, J., Bungum, H., Kotadia, J.: Ground-motion prediction equations based
on data from the Himalayan and Zagros regions. J. Earthq. Eng. 13, 1191–1210 (2009)
27. Sharma, M.L.: Attenuation relationship for estimation of peak ground vertical acceleration
using data from strong motion arrays. World Conf. Earthq. Eng., NZ Soc. Earthq. Eng.,
Auckland, NZ 88, 1–8 (2000)
28. Abrahamson, N.A., Silva, W.J., Kamai, R.: Summary of the ASK14 ground motion relation
for active crustal regions. Earthq. Spectra 30, 1025–1055 (2014)
29. Boore, D.M., Stewart, J.P., Seyhan, E., Atkinson, G.M.: NGA-West2 equations for predicting
PGA, PGV, and 5% damped PSA for shallow crustal earthquakes. Earthq. Spectra 30, 1057–
1085 (2014)
30. Campbell, K.W., Bozorgnia, Y.: Campbell-bozorgnia NGA-West2 horizontal ground motion
model for active tectonic domains. NCEE 2014—10th U.S. Natl. Conf. Earthq. Eng. Front.
Earthq. Eng. (2014). https://doi.org/10.4231/D3MS3K235
31. Chiou, B.S.J., Youngs, R.R.: Update of the Chiou and Youngs NGA model for the average
horizontal component of peak ground motion and response spectra. Earthq. Spectra 30, 1117–
1153 (2014)
32. Idrissa, I.M.: An NGA-West2 empirical model for estimating the horizontal spectral values
generated by shallow crustal earthquakes. Earthq. Spectra 30, 1155–1177 (2014)
33. Joshi, G.C., Sharma, M.L.: Estimation of peak ground acceleration and its uncertainty for
northern Indian region. Int. J. Geotech. Earthq. Eng. 2, 1–19 (2011)
34. Sharma, M.L, Malik, S.: Probabilistic seismic hazard analysis and estimation of spectral strong
ground motion on bed rock in north east India. In: 4th International Conference on Earthquake
Engineering (2006)
Experimental Investigation of Seismic
Response of Hybrid Shear Wall
with External Energy Dissipating
Reinforcement
Ankhiparna Guha, S. R. Dash, and Goutam Mondal
Abstract A conventional shear wall has negligible reusability capacity if it gets

severely damaged during any strong earthquake. This forms a major drawback for
many civil installations, which are prone to frequent seismic activities. Often, a post-
tensioned (PT) shear wall having self-centering capacity is used as one of the effective
ways of resisting such large lateral shear forces, however, it has limited energy
dissipation capability. Therefore, in situations having a higher force and limited drift
demand, PT shear walls with internal energy dissipating reinforcements (EDRs) are
that, once they get damaged, they cannot be replaced. Hence to solve this problem, use
of external energy dissipating reinforcement has been introduced due to its merit of
easy replacement. A new configuration of external energy dissipating device has been
introduced in this study, where notch plates are used as external energy dissipating
devices. The suitability and functionality of such external energy dissipating notched
bars (EDNB) have been checked experimentally and reported.
Keywords Post-tensioned shear wall · External energy dissipating notched bars

(EDNBs) · Self-centering
1 Introduction
Adequate seismic resistance and correct seismic design are crucial for ensuring
little structure disturbance and minor or trivial structural damage. One of the most
successful and widely acknowledged methods of such application is the introduction
to structural shear walls. However, the construction is unsuitable due to issues like
residual drifts, localized concrete damage, bar yielding, etc. Using post-tensioned
shear walls can be a workable approach to get around these restrictions. High strength
A. Guha (B)
IIT Kanpur, Kanpur, India
e-mail: ag32@iitbbs.ac.in
S. R. Dash · G. Mondal
IIT Bhubaneswar, Bhubaneswar, India
https://doi.org/10.1007/978-981-99-1459-3_17
206 A. Guha et al.
tendons that remain elastic during the seismic deformation and restore the wall to
its original position once the loads are withdrawn are a component of post-tensioned
shear walls, which exhibit self-centering effects. To preserve the structural form in
good working order, the greatest amount of energy that can be released through
the PT shear wall is restricted to energy release through the elastic behavior of the
PT tendons and pure wall shaking. Therefore, PT shear walls with internal energy
dissipating reinforcements (IEDR) are used in practice in conditions with a greater
force and restricted drift demand. The issue with this internal EDR, however, is that
once it is destroyed, it cannot be repaired. Complete destruction of such an element
necessitates total replacement of the building. Due to its benefit of easy replacement,
external energy dissipating reinforcement (EEDR) has thus been used to alleviate
this issue.
The work of some researchers on numerical analysis, experimental investigation
and the design of PT shear walls is included in this review of the literature. Addi-
tionally, it describes the field research done to improve the performance of PT shear
walls by the use of viscous dampers or steel energy dissipators.
Smith et al. [1] reviewed the lateral load behavior of three precast specimens
intended to mimic monolithic cast-in-place RC shear walls, as well as the perfor-
mance of two 0.40-scale hybrid precast concrete shear wall test specimens. While
the emulative wall only used mild steel bars, the hybrid walls used both mild steel
bars and high strength unbonded post-tensioning (PT) strands to provide lateral
resistance. The hybrid wall with continuous mild steel bars demonstrated superior
restoring, energy dissipation and ductile behavior over larger lateral displacements.
The findings highlight the possibilities for precast walls in seismic areas while also
highlighting crucial detailed issues.
Guo et al. [2] proposed and implied a seismic rehabilitation method by using a
self-compacting (SC) self-centering concrete wall with friction dampers, aiming at
providing self-centering capacity and supplemental energy dissipation to a structure.
The proposed SC wall was applied in the seismic resilience upgrade of a five-story
RC frame building located in a high-seismicity city in China.
Watkins et al. [3] carried conducted an analysis to investigate the influence of PT
wall to floor interaction on global response of the E Defense test facility, taking the
3D numerical modeling one step further. As a result, the modeling took into account
how each structural component will affect the structure as a whole. The two-bay
bonded PT moment frames in the longitudinal direction and the unbonded PT shear
walls in the transverse direction made up the lateral force resisting systems for the
test buildings. The Kobe earthquake of 1995 caused increasingly powerful strong
vibrations to be applied to the test building.
In this current study, experimental investigation has been performed to check
the suitability of two different external energy dissipating device configurations,
namely external energy dissipating reinforcement configuration (EEDR) and external
energy dissipating notched bar configuration (EDNB). The comparison of the effect
of various configurations of energy dissipaters has been represented using finite
element analysis by Guha et al. [4] and Prachi et al. [5]. To get more accurate idea
Experimental Investigation of Seismic Response of Hybrid Shear Wall … 207
about the response of the shear wall, experimental investigation was done in this work.
The best suitable model has been concluded on the basis of their performances.
2 Description of the Experimental Model
The test wall considered in the experimental study was having a height of 2.83 m
and a length of 1 m, at 1/3rd scale, thus satisfying the minimum height to length
ratio requirement of 0.5, according to ACI-ITG 5.2. The design and reinforcement
detailing of the wall is referred in Guha [6]. The thickness of the wall was 0.1 m,
and it was resting on a foundation beam of 0.14 m high. The foundation beam was
consisting of an ISMB 300 section lying horizontally, with two ISMC 100 channel
sections welded to its web. The space between the channel sections was filled with
fiber-reinforced concrete to provide a strong bearing surface for the rocking of the
wall. A depression of 2 cm was provided in fiber-reinforced concrete at the top of the
foundation beam inside which the wall was placed to prevent horizontal sliding of
the wall at the base while rocking. To prevent the uplifting of the foundation beam,
angle sections were placed on the top of the foundation, which was anchored to the
ground with the help of anchor bolts. Two holes were made in the web of the I section
for the connection of PT tendons with the foundation beam. The PT tendons were
placed with a straight profile inside the wall, anchored to the top of the wall at one
end and to the bottom of the foundation beam at the other end.
The main objective of testing the PT hybrid shear wall was to check the feasi-
bility of placing two different external energy dissipating devices which are referred
as external energy dissipating reinforcement bars (EEDRs) and external energy dissi-
pating notched bars (EDNBs) in this study. The devices are fitted externally to facil-
itate its easy replacement without compromising its energy dissipation and self-
centering behavior. For establishing an external connection of the two devices, two
angle sections were attached to the wall at 0.20 m height from the base. The angle
sections were fitted to the wall with the help of bolted connections. The devices were
connected to the angle section at one end and the channel sections of the foundation
beam at the other with the help of bolted connections, thus spanning between the
two with an unbonded length of 0.20 m.
The reinforcement detailing of the PT shear wall along with the schematic diagram
and actual view of the EDNB connection is shown in Fig. 1. The middle half portion of
EDNB having a length of 10 mm was milled to a smaller cross-section of dimensions
8 mm × 10 mm and then tapered to a cross-section of 15 mm × 10 mm to prevent
failure of anchorage and to shift the weak zone to the center of the bar. The two ends
of the bar were threaded so that they could be attached to the channel and the angle
sections with the help of bolted connections.
The connection of EDNB with the angle and channel sections was made in such
a way that it could resist the slippage of angle sections so that the EDNBs can be
subjected to proper tension and compression. The connection made was such that
208 A. Guha et al.
Fig. 1 Reinforcement detailing of PT shear wall along with sketch showing the experimental
arrangement for EDNB connection
the EDNB could be easily replaced after it is damaged by simply removing the bolts
from the angle and channel sections.
The detailing of the EEDR connection is shown in Fig. 2. To prevent anchorage
failure and to move the weak zone to the center of the bar, the middle half part of
EDR is machined to a smaller diameter of 9 mm, as shown in Fig. 3. By removing the
bolt connection from the angle and channel portions, it would be simple to replace
the EDR before it yields. In order to submit the EEDRs to the necessary tension and
compression, the connection of the EDNB with the angle and channel sections was
built in a way that it could resist the slippage of the angle sections. The bolts from the
angle and channel parts could be easily removed, allowing the EDNB to be replaced
once it is damaged.
Fig. 2 Actual connection and schematic diagram of the connection of EEDR

Fig. 3 EEDR with reduced

diameter of the middle
portion
To control the out-of-plane displacement of the shear wall, two numbers of column
supports were given at the front and the backside of the wall (Fig. 4a). Apart from this,
an additional frame arrangement was also attached to the wall which is connected
to the end of the load actuator. To resist the development of additional moments due
to the rigidness of the actuator end, a hinged swivel head has been attached, which
allows rotation at the actuator end while displacement-controlled loading is being
applied.
2.1 Material Properties
Stress–strain properties of the various steel used in the experimental model, based
on tensile testing, are given in Table 1. The average compressive strength of concrete
used in the shear wall model was 40 MPa. To obtain the stress–strain curve of the PT
tendon, a tensile test has been carried out as per ASTM A416. The tendon satisfied
the criteria of the minimum breaking strength of 160 kN and a total elongation value
of greater than 3.5% as required by the code.
2.2 Instrumentation
Two load cells have been placed, one, in front of the hydraulic actuator and another,
at the top of the PT tendon, to measure the load applied to the structure and the
amount of PT stress generated in the tendon during the test. The actuator was fitted
with an lateral variable differential transducer (LVDT) to measure the displacement
210 A. Guha et al.
(a)
(b)
Fig. 4 a PT shear wall experimental setup and b the schematic diagram showing experimental
setup with instrumentation
Table 1 Actual stress–strain properties of reinforcement bars and PT tendon used in the experiment
Reinforcement type Diameter of bar (mm) E (GPa) f y (MPa) f u (MPa) Eu
Distribution steel 6 210 528 622 0.15
EEDR Bar of variable dimensions 206 550 650 0.14
EDNB Plate of variable dimensions 210 550 650 0.14
Tie rod 12 208 502 563 0.14
PT tendon 12.7 200 1678 1750 0.02
of its piston with reference to its body. However, another LVDT has been attached at
the back of the actuator loading frame to correct the actual lateral movement of the
actuator piston with reference to the fixed ground by applying a correction for any
displacement happening to the loading frame at actuator level, although it was very
small. Another LVDT was placed at the bottom of the wall on one side to measure
the horizontal slip of the structure. Two potentiometers have been attached at the
reaction wall side of the PT wall, one horizontally and another vertically, to measure
the horizontal slip and vertical uplift, respectively. All the sensors were calibrated
before fixing to the experimental model. Figure 4 shows the schematic diagram of
the instrumented experimental setup of the hybrid PT shear wall with external energy
dissipating notched bars (EDNB).
2.3 Post-tensioning
The post-tensioning of the tendons was done with the help of a post-tensioning jack.
Due to the physical limitations of the mechanical application of prestress to tendons,
the exact prestress in the tendon could not be precisely controlled and applied as
designed. However, after prestressing, the stress in the tendon was measured to be
about 797 MPa, which was corresponding to 74 kN of prestressing force as compared
to 83 kN of design prestress in each tendon. Safety arrangements are provided at the
top of the anchorage by connecting them firmly to the wall with the help of steel
channel sections.
2.4 Loading
The actuator was placed at the height of 2.3 m from the base of the wall, where
the resultant force of the first modal inertia forces is expected, as per the design.
Displacement controlled slow cyclic loading was applied at the rates of 0.4 mm/sec to
avoid the dyamic effect of the structural response, as shown in Fig. 5. The lateral load
was applied cyclically in the pushing direction to avoid the inconvenient connection
of the bulky swivel head.
3 Experimental Results and Discussions
To check the feasibility of the proposed method of placing external energy dissi-
pating reinforcing bars (EEDR) and energy dissipating notched bars (EDNB) in
a PT shear wall for energy dissipation, experimental testing has been performed.
Various responses of the model were recorded, including backbone curves, stiffness
212 A. Guha et al.
45
40
Displacement (mm)
35
30
25
20
15
10
5
0
0 500 1000 1500 2000
Time (sec)
Fig. 5 Displacement controlled load applied to the structure
degradation and energy dissipation which is analyzed and presented in the following
sections.
3.1 Comparison of Backbone Curves
The backbone curves of the experimental models with two different configurations
namely EEDR and EDNB are plotted in Fig. 6. It has also been observed that in the
initial drift values, the backbone curve of EDNB is steeper as compared to that of
the EEDR model. At design drift 0.384%, the load carried by the EDNB model is
22.5 kN whereas for EEDR model the value is 14.15 kN. But with the increase in the
amplitude of displacement, the backbone curve of EDNB model becomes flatter. For
a drift value of 1.35%, the load carrying capacity of EDNB model becomes equal to
the EEDR model, i.e., 34.25 kN. For drift values greater than 1.35%, the load carrying
capacity of EEDR model becomes higher than the EDNB model. At maximum drift
value, the load carried by the EDNB model is 39.15 kN and for EEDR model the
value is 40.21 kN.
Fig. 6 Comparison of 45
backbone curves of 40
experimental models with 35
EEDR and EDNB
Load (kN)
30
25
20
15
10 EEDR
5 EDNB
0
0 0.5 1 1.5 2
Lateral Drift (%)
1800
1600
1400
Stiffness (kN/m)
1200
1000
800
600
400 EDNB
200 EEDR
0
0.3 0.5 0.7 0.9 1.1 1.3 1.5
Lateral Drift (%)
Fig. 7 Comparison of stiffness degradation curves of EEDR and EDNB models
3.2 Comparison of Stiffness Degradation
The comparison of stiffness of both the models corresponding to their lateral drifts
is shown in Fig. 7. It can be perceived that the initial stiffness of the EDNB model
is relatively higher as compared to the EEDR model (which is evident from the
backbone curve also). At design drift of 0.384%, the stiffness of EDNB model is
1623 kN/m. In case of EEDR mode, the value is 1104 kN/m. Although, as the lateral
drift increases the rate of stiffness degradation is higher in case of EDNB model as
compared to the other one. That’s why, near maximum drift value, the stiffness of
both the model becomes almost similar. At 1.45% lateral drift ratio, the stiffness of
EDNB is 855 kN/m whereas in case of EEDR, the value is 800 kN/m.
3.3 Comparison of Energy Dissipation
The amount of cumulative energy dissipated by the two experimental models has been
compared, as shown in Fig. 8. Both the models are showing almost similar amount
of energy dissipation corresponding to its lateral drift values. Although, if the graph
can be observed closely, it is seen that for lateral drift values up to design drift, i.e.,
0.384%, the amount of energy dissipated by the EDNB model is slightly higher that
of EEDR model. At design drift, the energy dissipated by EDNB is 17.64 kN-m,
whereas in case of EEDR, the value is 11.02 kN-m. At maximum drift, i.e., 1.45%,
the amount of energy dissipation for EDNB is 287.1 kN-m and for EEDR the value
is 292.34 kN-m.
If the behavior of both the models is compared, it can be seen that at same drift
values the EDNB model is showing higher load taking capacity as compared to EEDR
initially. Therefore, while calculating the amount of energy dissipation, despite of
high stiffness values, the EDNB model was able to dissipate almost equivalent amount
of energy as EEDR.
214 A. Guha et al.
Fig. 8 Comparison of 350
Energy Dissipated (kN-m)

energy dissipation of 300
numerical model and
250
experimental model
200
150
100
EEDR
50 EDNB
0
0 0.5 1 1.5 2
Drift Value (%)
The above discussion explains a comparative experimental study of shear wall models
with two different external energy dissipating devices which are external energy
dissipating reinforcing bars (EEDRs) and external energy dissipating notched bars
(EDNBs). Various aspects of the two models have been compared such as compar-
ison of backbone curves, stiffness degradation and energy dissipation. It has been
perceived that although there is not much difference in ultimate load capacity of the
two models, the initial stiffness of the EDNB model is higher as compared to the
EEDR model. This can be observed in the stiffness degradation curve as well. The
initial stiffness of the EDNB model is quite high. But with increase in lateral drift, the
rate of stiffness degradation is high in case of EDNB model, hence at ultimate lateral
drift values both the models showed almost similar stiffness. While comparing the
energy dissipation, both the models are showing almost similar amount of energy
dissipation.
From the above discussions, it can be said that both the models are showing
almost similar behavior, but the EDNB model is showing relatively higher stiffness
as compared to the EEDR model, hence having higher load carrying capacity at initial
small lateral drift values. Therefore, the EDNB model can be more efficient in terms
of lateral load taking capacity as compared to EEDR model. It can be concluded that
the hybrid shear wall model with external energy dissipating notched bar can be a
better option as an energy dissipating device.
References
1. Watkins, J., Sritharan, S., Nagae, T., Henry, R.S.: Computational modelling of a four storey
post-tensioned concrete building subjected to shake table testing. Bull. N. Z. Soc. Earthq. Eng.
50(4), 595–607 (2017)
2. Smith, B.J., Kurama, Y.C., McGinnis, M.J.: Design and measured behavior of a hybrid precast
concrete wall specimen for seismic regions. J. Struct. Eng. 137(10), 1052–1062 (2011)
3. Guo, T., Xu, Z., Song, L., Wang, L., Zhang, Z.: Seismic resilience upgrade of RC frame building
using self-centering concrete walls with distributed friction devices. J. Struct. Eng. 143(12),
04017160 (2017)
4. Guha, A., Taori, P., Dash, S.R., Mondal, G.: Efficient numerical analysis of hybrid shear wall with
energy dissipating reinforcements. In: Proceedings of the Indian Structural Steel Conference-
2022, IIT Hyderabad, Hyderabad, India
5. Taori, P., Dash, S.R., Mondal, G.: Seismic response of post tensioned hybrid shear walls with
external energy dissipating reinforcement (EEDR). J. Earthq. Eng. (2020). https://doi.org/10.
1080/13632469.2020.1778587
6. Guha, A.: Seismic response of post-tensioned hybrid shear walls with external energy dissipating
reinforcement. M. Tech Thesis, IIT Bhubaneswar, India (2020)
1D Velocity Model for NW India
in and Around Delhi
Deepak Kumar , G. Suresh, S. C. Gupta , M. L. Sharma ,

and Hasbi Ash Shiddiqi
Abstract We derive a 1D crustal velocity model based on P- and S-wave residual

time arrivals at the national seismological network maintained by National Center for
Seismology (NCS) for NW India around Delhi between latitude range 22º N to 33º
N and longitude range 73ºE to 80ºE, which separates interplate area in the north and
intraplate area in the south. Available models are mostly from regional studies having
low crustal resolution. We use 413 earthquakes in all, which are chosen based on the
maximum number of stations, minimum azimuthal gap, and root mean square travel
time. We use VELEST code to invert the 1D velocity model, hypocenter locations,
and station delays. The Indo-Gangetic (IG11) model is used as a starting model, and
in this model, we further created 500 initial models by adding random perturbation ±
10%. We also evaluate the effect of different Moho depths (i.e., 38, 43, and 48 km).
We selected the preferred model based on how the models converge and on the travel
time residuals. The minimum residual travel time of P- and S-arrivals indicated
crustal thickness as 43 km. Further, this model can be used for seismicity and 3D
seismic tomographic studies of the area.
Keywords Earthquake location · 1D crustal velocity model
D. Kumar (B) · S. C. Gupta · M. L. Sharma

e-mail: jangradeepak80@gmail.com; dkumar4@es.iitr.ac.in
S. C. Gupta
e-mail: s.gupta@eq.iitr.ac.in; scgeqfeq@iitr.ac.in
M. L. Sharma
e-mail: m.sharma@eq.iitr.ac.in; mukutfeq@iitr.ac.in
G. Suresh
National Center for Seismology, New Delhi, India
e-mail: gsureshimd@yahoo.com
H. A. Shiddiqi
Department of Earth Science, University of Bergen, Bergen, Norway
e-mail: Hasbi.Shiddiqi@uib.no
https://doi.org/10.1007/978-981-99-1459-3_18
218 D. Kumar et al.
1 Introduction
The Delhi falls on the western edge of the Ganga basin, east of the Delhi–Hardwar
Ridge, surrounded by the Himalayas in the north, and the Aravalli ranges in the
south. The rocks of Delhi belong to both the earliest and latest chapters of the geolog-
ical history of the earth: the Pre-Cambrian and the Quaternary. The Pre-Cambrian
rocks belong to the Alwar series of the Delhi System. Geologically, 95% of the
area of the Delhi region is covered by the Quaternary sediments, and the rest 5%
comprises the Meso- to Neo-Proterozoic rocks of the Delhi Supergroup present in
south and southwest intruded by more than one phase of acidic and basic intru-
sives of Neo-Proterozoic age (Post Delhi intrusives) and the tertiary rocks in the
northeast. Numerous local tectonic features surround the Delhi region, including
the Aravalli–Delhi fold, Mahendragarh–Dehradun Fault, Moradabad Fault, Great
Boundary Fault, Sohna Fault, Delhi–Hardwar Ridge, Mathura Fault, Delhi–Lahore
Ridge, Sohna Fault, Mahendragarh–Dehradun Fault, and Moradabad Fault, as well
as other less significant lineaments and distinct tectonic features such as the Main
Boundary Thrust (MBT) and Main Central Thrust (MCT) of the Himalaya, as well
as various minor lineaments and far-off tectonic structures.
Several significant earthquakes have occurred close to Delhi in the past: Delhi
earthquake (1720) M: 6.5, Mathura earthquake (1803), M: 6.8, Bulandshahar earth-
quake (1956) M: 5.8, Delhi NCR earthquake (1960) M: 4.8, and small magnitude
earthquakes frequently occur in the area. Many small (1.0 ≤ Mw ≤ 3.0) and moderate
earthquakes occur on the Delhi NE border adjoining Uttar Pradesh. The Hazard
in Delhi originated due to the Himalayan thrust fault system which was further
associated with the collision between the Indian and Eurasian plates [1–4] (Fig. 1).
From seismological studies, the estimated average Moho depth is 43 km [5] and
46 km [6] from receiver function studies beneath the Delhi region,the joint inver-
sion of receiver function and Rayleigh wave group velocity shows crustal thickness
variation from 40 to 44 km beneath Delhi Fold Belt (DFB). Mandal et al. (2013)
used the common reflection stack method variation in crustal thickness across the
Aravalli Delhi fold Belt and found a variation from 38 to 50 km. Although lots of
models have been provided by different researchers for this area, those models have
a low resolution in the crust. The VELEST program provides a P and S wave high-
resolution velocity structure. Various seismological researches have been conducted
by considering the reference velocity model as the initial starting velocity model, and
this initial model is modified until there is a strong and positive correlation between
the majority of observed and predicted values. The impact of the initial reference
model on tomographic inversion results has not been properly grasped in many earlier
investigations. An incorrect initial reference model can affect the quality of the solu-
tion, and it can also introduce large uncertainties in the final results. To overcome this
kind of problem, we evaluated the 1D model for the Delhi region by using the joint
inversion of travel time data, revised hypocenter coordinates, and station corrections.
The evaluated minimum 1D velocity model can use as a reference model for the 3D
topography of the Delhi region.
1D Velocity Model for NW India in and Around Delhi 219
Fig. 1 Map of Delhi with the major regional tectonic features of Himalaya: Himalaya frontal
thrust (HFT), main boundary thrust (MBT), South Tibetan Detachment (STD), Delhi–Hardwar
ridge (DHR), Moradabad fault (MF), Sohna fault (SF), Mathura fault (MTF), and great boundary
fault (GBF). The distribution of earthquakes (red star) and NCS seismic observatories are shown in
black triangles
2 Data
The hypocenter parameters, P- and S-wave arrivals for the study region bounded by
between latitude 22ºN–33ºN and longitude 73ºE–80ºE, are collected from the seismo-
logical bulletins of the International Seismological Center (ISC) for the region (http://
www.isc.ac.uk/cgi-bin/collect?Reporter=NDI) and the website of NCS (https://sei
smo.gov.in/bulletins) for the period July 2000 to June 2020, i.e., 20 years of data.
We selected 413 earthquakes based on the criteria, (a) record of a minimum of
eight stations, (b) having an azimuthal gap less than 180º, and (c) root mean square
(RMS) travel time residuals less than 2.0s. There are 3191 P-arrivals and 2986 S-
arrivals on 51 recorded stations in the dataset that covers the whole study region
(Fig. 2).
220 D. Kumar et al.
Fig. 2 Different ray path directivities of 413 earthquakes (circles according to their focal depth
distribution) used for velocity inversion a P-wave ray paths and b S-wave ray paths
3 Methodology
The evaluation of travel time from seismic velocity models along their ray paths
and the location of earthquakes is an inverse nonlinear problem, and the problem is
linearized to solve it in the least-square sense [7]. The arrival time of a seismic wave
from the source to recoding station is a nonlinear function of station coordinates (s),
hypocentral location parameters (h), and velocity model (m) as given by [7, 8].
tobs = f (s, h, m) (1)
To solve Eq. (1) we start with the initial guess for these unknown parameters and
we calculated theoretical arrival times (tcal ). The difference between observed and
calculated arrival time is the residual travel time (tr es ) and can be expressed as
∂f ∂f
tr es = tobs − tcal = h k + m i + e (2)
k=1,4 ∂h k i=1,n ∂m i
where in matrix notation
[t] = [H ][h] + [M][m] + [e] = [A][d] + [e] (3)
[t] is a vector of travel time residuals, [H] is a matrix of partial derivatives of travel
time with respect to hypocenter parameters, [h] is a vector of hypocenter parameter
adjustment, [M] is a matrix of partial derivatives of travel times with respect to model
parameters, [m] is a vector of model parameters adjustment, [e] is a vector of travel
time error, [A] is a matrix of all partial derivatives, and [d] is a vector of hypocentral
and model parameter adjustments.
The mean residual time and RMS error are used to drive the velocity model where
the final model is defined in the least square sense. The initial models are generally
adopted from previous similar works for the regions, and refined models are then
recommended for their further use in 3D tomography.
3.1 Inversion of the Velocity Model
Bhattacharya [9], has evaluated Indo-Gangetic (IG11) velocity model from surface
wave dispersion data. We used this IG11 velocity model (Table 1) for our present
study area to evaluate the minimum 1D velocity model. We started with two different
velocity models with and without the sedimentary layer and consisting of six (the first
layer is the sedimentary layer) and five layers up to 100 km of depth, respectively.
The depth of the top of the 4th layer is Moho which is being tested for different
depths.
We started with two different models, one is the model (without sedimentary
layer) in which the top layer is above 0, the depth of the 2nd layer is 3.5 km, the
depth of the 3rd layer is at 20 km, the depth of the top of the 4th layer is Moho, and
5th layer is fixed at 100 km from IG11 model. In the second model (with sedimentary
layer), one sedimentary layer at depth of 0.5 km has been introduced in addition to
the layers considered in the first model.
We inverted for seismic velocities (P and S waves) but not thicknesses of layers,
and we divided the crust into 5 km layers. As suggested by Kissling et al. [10], we
also tested Moho depth by ± 5 km to the average depth of Moho, i.e., 38, 43, and
48 km3 . We relocated the hypocenter in the first iteration and inverted for the velocity
model in the second iteration; in this way we evaluated the hypocenter location and
velocity model simultaneously for up to 20 iterations and the layers with almost the
same velocities are combined. We finally determined with this process the average
Table 1 Velocity model of Indo-Gangetic plain

Depth of top of layer Thickness of layer (km) Vp (km/s) Vs (km/s) Density (gm/cm3 )
(km)
Above 0.0 3.5 3.40 2.00 2.00
3.5 16.5 6.15 3.55 2.60
20.0 23.0 6.58 3.80 3.00
43.0 57.0 8.19 4.60 3.30
100.0 20.0 8.30 4.60 3.30
222 D. Kumar et al.
Fig. 3 a Initial 1D velocity model for Vp and Vs from IG11 model. b Generated random initial
models from inversion. c The best models and random models are depicted by black and gray lines,
respectively
crustal thickness is 43 km in the Delhi region. Inversion steps’ example has been
shown for Moho 43 km using the IG11 model in Fig. 3.
We tested several randomly generated models to get the best velocity model. For
each model, we create 500 perturbations and further modify VP and VS for each layer
in the ± 10 range. We use VP/VS ratio range of 1.6–1.9, and then for each model,
we inverted using VELEST. The inversion process has been followed by Kissling
et al. [7], in which origin, hypocenter, and depth damping are fixed at 0.01, but the
velocity and station correction damping is fixed at 1.0. For each initial model, we
have accepted the models with the lowest (10%) travel time. We repeat this process
for 20 iterations. The best models are averaged to get the final velocity model. The
inversion result distributions for all the three tested Moho depths 38, 43, and 48 km3
with and without sedimentary layer are shown in Fig. 4.
We used New Delhi Station (NDI) as a reference station for station correction
which is located at the center of our study area. The station corrections are refined
further by choosing a higher damping value (=10.0) for the final velocity model. This
further results in there being no change in the velocity model, but we will get the
updated station corrections and hypocenters.
To check the quality of solutions, we used different velocity models with station
corrections and located events with the HYPOCENTER program [11]. Then for each
velocity model, we compared travel time residuals and found the relocated models
with sedimentary layers have higher means of travel time than that of initial locations
(Table 2).
Fig. 4 All accepted results from the random test are shown by gray color lines, with the final
velocity models shown by a red line, and the best models are represented by black lines with 38,
43, and 48 km3 Moho depth beneath the Delhi region from left to right: a outcomes of the model
having no sedimentary layer and b outcomes of the model having sedimentary layer
Table 2 Comparison of mean residuals from the initial and final models
S. No. Velocity model Initial mean residual (s) Final mean residual (s)
(km) Without With Without With
sedimentary sedimentary sedimentary sedimentary
layer layer layer layer
1 38 2.620509 1.166025 0.950155 0.988897
2 43 2.446367 1.165017 0.890986 0.967577
3 48 2.448025 1.370489 0.944641 1.017219
224 D. Kumar et al.
We used the earthquake catalog from July 2000 to June 2020 with their updated loca-
tion obtained by simultaneous inversion through the VELEST program of SEISAN
software [12]. The shallower (focal depth less than 50 km) earthquakes are used for
the processing of travel times. A total of 313 earthquakes are used in the present
study which is recorded at least by eight stations. We have selected the velocity
model having Moho depth at 43 km with P and S wave velocities of 8.18 km/s and
4.71 km/s as our final 1D velocity model (Table 3 Fig. 5a) based on the minimum
travel time residual (0.890986s and 0.967577s) travel time. As the error difference
due to the presence of a sedimentary layer is negligible in comparison to the resolu-
tion of the processing so no definite conclusions may be drawn about the presence
of a sedimentary layer in this region.
Several studies show the Moho depth in the Delhi region varies from 38 to 45 km.
Rai et al. [13] computed Moho depth variation around 40 to 75 km for Delhi and
Taksha at the Karakorum fault by using joint inversion of Rayleigh wave dispersion
data and receiver function analysis. The P and S wave velocities found at Moho
were 7.6 km/s and 4.4 km/s respectively which is comparable to the present study
Table 3 Final velocity model

Top of the layer (km) Vp (km/s) Vs (km/s)
with Moho 43 km
Above 0 1.50 0.83
0.5 (sedimentary layer) 3.77 1.96
3.5 6.06 3.54
20 6.46 3.78
43 8.18 4.71
100 8.56 5.01
Fig. 5 a Initial and final models’ seismic velocities plots. b The regional comparison of seismic
velocities with the present study
Table 4 Regional
Models V P (km/s) V S (km/s) Moho (km)
comparison of Moho with
different models Kumar et al. [14] 7.79 4.50 43
Rai et al. [15] 7.79 4.50 36
Rai et al. [13] 7.61 4.40 40
Julià et al. [16] 7.61 4.40 40
Present study 8.19 4.73 43
Vp √
Note Vp has been calculated by using the relation Vs = 3
(Fig. 5b). Different methods have been used to obtain the 1D velocities models for
the Delhi region, and we compared our results with these studies (Table 4, Fig. 5b).
The crust beneath the south shield has a similar Moho of ~36 km and similar
seismic velocities [15], and the Indian shield has 33–39 km of Moho depth [14].
Reference [16] has also evaluated the seismic velocity structure with crustal thickness
40 km beneath Delhi for the Indian shield. The P and S wave velocities of the present
study at Moho are 8.19 km/s and 4.73, respectively, which is a little bit higher than
all other comparison models (Table 4, Fig. 5b).
Variation in the crustal velocity and thickness can cause significant arrival time
delays [17]. Singh et al. [18] showed that the crustal velocity in the Indian shield
is higher than in the Himalayan region. The stations located in the northern part of
our study area show delayed P and S arrivals (positive delays gray color triangles),
and this may be due to lower crustal velocity around the Himalayas and the stations
located in the southern part of the area which show early P and S arrivals (negative
delays red color triangles) and have higher crustal velocity (Fig. 6).
Fig. 6 P and S wave’s travel-time station corrections were obtained from VELEST inversion
226 D. Kumar et al.
5 Conclusion
In the present study, a 1D velocity model of the Delhi region has been estimated using
413 earthquakes recorded at 51 stations. We observe the negative delays or early
arrivals observed in the center and the southern part of the study area are probably
related to this higher crustal velocity, and the lower velocity around the Himalayas
can cause positive delays or late arrivals. We confirm the new velocity model, which
consists of 6 layers up to 100 km of depth with P and S wave velocities varying from
1.5 km/s to 8.56 km/s. The sedimentary layer seems to be at a depth of top 0.5 km with
P and S wave velocities of 1.50 km/s and 0.83 km/s, respectively; however, the error
difference due to the presence of the sedimentary layer is negligible in comparison
to the resolution of the processing so no definite conclusions may be drawn about the
exact thickness of the sedimentary layer in this region. Moho depth appears to be at
43 km having minimum residual time and maximum precision in earthquake location
from the tested models with the P and S velocities of 8.18 km/s and 4.71 km/s,
respectively. The final estimated velocity model can be used to determine source
parameters and as an input for the studies of 3D tomography of the Delhi region.
Data and Resources
The phase data of earthquakes have been taken from the monthly seismological
bulletins published on the NCS website (https://seismo.gov.in/bulletins) and Inter-
national Seismological Center (ISC) (http://www.isc.ac.uk/cgi-bin/collect?Report
er=NDI) for the region. The figures were made using Generic Mapping Tools and
ArcGIS software.
Acknowledgements The authors are thankful to the Director of National Center for Seismology,
New Delhi, India for making the data available to carry out the study. Acknowledgments are due
to THDCIL, Rishikesh, for help and data.
References
1. Chandra, U.: Seismotectonics of the Himalayas. Curr. Sci. 62, 40–71 (1992)
2. Joshi, G.C., Sharma, M.L.: Strong ground-motion prediction and uncertainties estimation for
Delhi, India. Nat. Hazards 59(2), 617–637 (2011). https://doi.org/10.1007/s11069-011-9783-y
3. Sharma, M.L., Wason, H.R., Dimri, R.: Seismic zonation of the Delhi region for bedrock
ground motion. Pure Appl. Geophys. 160(12), 2381–2398 (2003). https://doi.org/10.1007/s00
024-003-2400-6
4. Sharma, M.L.: Seismic hazard in the Northern India region. Seismol. Res. Lett. 74(2), 141–147
5. Mitra, S., Kainkaryam, S.M., Padhi, A., Rai, S.S., Bhattacharya, S.N.: The Himalayan foreland
basin crust and upper mantle. Phys. Earth Planet. Inter. 184(1–2), 34–40 (2011). https://doi.
org/10.1016/j.pepi.2010.10.009
6. Borah, K., Kanna, N., Rai, S.S., Prakasam, K.S.: Sediment thickness beneath the Indo-Gangetic
Plain and Siwalik Himalaya inferred from receiver function modelling. J. Asian Earth Sci. 99,
41–56 (2015). https://doi.org/10.1016/j.jseaes.2014.12.010
7. Kissling, E., Ellsworth, W.L.L., Eberhart-Phillips, D., Kradolfer, U.: Initial reference models
in local earthquake tomography. J. Geophys. Res. 99(B10), 19635–19646 (1994)
8. Kissling, E.: Geotomography with local earthquake data. Rev. Geophy. 26, 659–698 (1988)
9. Bhattacharya, S.N.: Crustal and upper mantle structure of India from surface wave dispersion.
Curr. Sci. 62, 94–100 (1992)
10. Kissling, E., Kradolfer, U., Maurer, H.: VELEST User’s Guide–Short Introduction. (Institute
of Geophysics and Swiss Seismological Service, ETH Zurich, 1995), p. 25
11. Lienert, B.R., Havskov, J.: A computer program for locating earthquakes both locally and
globally. Seismol. Res. Lett. 66(5), 25–36 (1995). https://doi.org/10.1785/gssrl.66.6.26
12. Havskov, J., Voss, P.H. , Ottem¨ oller, L.: Seismological observatory software: 30 Yr of SEISAN.
Seismol. Res. Lett. 91(3), 1846–1852 (2020)
13. Rai, S.S., Priestley, K., Gaur, V.K., Mitra, S., Singh, M.P., Searle, M.: Configuration of the
Indian Moho beneath the NW Himalaya and Ladakh. Geophys. Res. Lett. 33(15), 3–7 (2006).
https://doi.org/10.1029/2006GL026076
14. Ravi Kumar, M., Saul, J., Sarkar, D., Kind, R., Shukla, A.K.: Crustal structure of the Indian
shield: New constraints from teleseismic receiver functions. Geophys. Res. Lett. 28, 1339–1342
(2001). https://doi.org/10.1029/2000GL012310
15. Rai, S.S., Priestley, K., Suryaprakasam, K., Srinagesh, D., Gaur, V.K., Du, Z.: Crustal shear
velocity structure of the south Indian shield. J. Geophys. Res.: Solid Earth 108(B2) (2003).
https://doi.org/10.1029/2002jb001776
16. Julià, J., Jagadeesh, S., Rai, S.S., Owens, T.J.: Deep crustal structure of the Indian shield from
joint inversion of P wave receiver functions and Rayleigh wave group velocities implications
for Precambrian crustal evolution. J. Geophys. Res.: Solid Earth 114(10), 1–25 (2009). https://
doi.org/10.1029/2008JB006261
17. Shiddiqi, H.A., Ottemöller, L., Rondenay, S., Halpaap, F., Gradmann, S., Michálek, J.: Crustal
structure and intraplate seismicity in Nordland, Northern Norway: insight from seismic
tomography. Geophys. J. Int. 230(2), 813–830 (2022). https://doi.org/10.1093/gji/ggac086
18. Singh, A., Ravi Kumar, M., Mohanty, D.D., Singh, C., Biswas, R., Srinagesh, D.: Crustal struc-
ture Beneath India and Tibet: new constraints from inversion of receiver functions. J. Geophys.
Res.: Solid Earth 122(10), 7839–7859 (2017). https://doi.org/10.1002/2017JB013946
Anomalous Deviations in Atmospheric
Parameters as Pre-earthquake
Signals-A Case Study on Sumatra Region
Earthquakes (M ≥ 6.0)
Ramya Jeyaraman and N. Venkatanathan
Abstract One of the most demanding disciplines in geoscience is earthquake fore-

casting, which is widely debated due to its heterogenetic nature. Several strategies for
forecasting the destructive earthquake events have been developed in recent decades.
This work focuses on exploring and understanding of the precursory signals which
is associated with the occurrence of earthquakes. The earthquake precursors are
identified by anomalous deviation in outgoing longwave radiation (OLR) and rela-
tive humidity (RH) before strong magnitude earthquakes. This work concentrates
to study the correlation of anomalous variations in atmospheric parameters such as
RH and OLR flux index associated with the earthquakes through lithosphere–atmo-
sphere–ionosphere coupling (LAIC) model. The retrospective analysis of temporal
variations is done on the daily datasets of RH, and OLR is carried out for eight earth-
quakes with magnitude ≥ 6.0 occurred during 1991–2021 within 75 km radius of
December 26, 2004. The exploratory data analysis is carried out based on the investi-
gation of earthquake activity in the study region taken from United States Geological
Survey (USGS), anomalous OLR flux index, anomalous RH index for the period
of six months. According to our preliminary findings, the relationship among the
anomalous drop of RH and anomalous increase in OLR flux index anomalies with
the occurrence of earthquakes were observed near epicentral region as proposed by
the LAIC model. This anomalous behavior was observed 3 days to 3 months prior to
the occurrence of earthquakes. Hence, the authors inferred that there exists a possible
link between the abnormal deviations in atmospheric parameters and the occurrence
of earthquakes. Further incorporation of various precursors along with OLR and RH
studies can help us to develop an effective tool for forecasting the earthquake on a
short-term basis.
Keywords Earthquake precursors · Outgoing longwave radiation · Relative

humidity · Earthquake forecasting
R. Jeyaraman · N. Venkatanathan (B)

School of Electrical and Electronics Engineering, SASTRA Deemed to Be University, Thanjavur,
Tamil Nadu, India
e-mail: venkatanathan@eee.sastra.edu
https://doi.org/10.1007/978-981-99-1459-3_19
230 R. Jeyaraman and N. Venkatanathan
1 Introduction
Identification of potential earthquake precursors poses challenging research in short-

term earthquake forecasting studies due to its heterogeneous nature. Numerous inves-
tigations revealed anomalous changes in space–time transients related to OLR and
RH during the earthquake preparation process.
Recent advances in remote sensing technologies have improved our understanding
of the earth’s surface seismic signals. The change in OLR radiation can reflect the
evolution of the critical state of ground stress [1]. Satellite thermal imaging data can
detect anomalies in OLR [2].
The energy emitted by the earth is electromagnetic radiation that travels through
the atmosphere and into space as OLR [3]. The release of energy into the atmosphere
causes a rise in air temperature. Even if the radon quantity in the air is negligible, the
energy efficiency of the ionization process is exceptionally high [4, 5] which leads
to anomalous variations in atmospheric parameters such as RH, OLR (more than
108 ) [6]. OLR is affected by the temperature near the surface and in the atmosphere,
air humidity, and cloud presence [7]. Analyzing longwave radiation data using time
series, comparison, and eddy field methods have been performed. The association
between abnormality and earthquake occurrence has shown that seismic anomalies
can be detected using OLR singularities [8]. The parameters of the atmosphere and
ionosphere are jointly analyzed with variations in both temporal and spatial aspects.
It has been demonstrated that abnormalities over the atmosphere and ionosphere
frequently develop close to an earthquake’s epicenter [9]. A potential coupling mech-
anism between the lithosphere and the ionosphere of the earth is proposed by the
LAIC model. According to the model, radon gas emissions from the earth’s crust
are caused by pre-earthquake seismic activity triggered by the movement of tectonic
plates in the lithosphere. Radon gas emissions cause the atmosphere’s air to ionize,
which renders possible anomalies [10–12].
Using an interdisciplinary approach, pre-earthquake signals such as proton (H+
ion) permeability, RH, radon gas emission, and OLR can be employed as a precursor.
The movement of H+ ions from the earth’s core to the surface causes an increase in
tectonic activity, leading to the emission of radon gas. It causes the air molecules in
the atmosphere to ionize, which lowers the RH on the surface of the earth. Latent
heat release caused by the upward migration of ionized air molecules from the earth’s
surface causes the anomaly in OLR flux [13–15].
The possible relationship between TEC, air temperature, and RH and Mexico
M7.4 earthquake is studied and shows that RH displayed a negative anomaly above
the earthquake’s epicenter [16]. Sumatra is vulnerable to earthquakes due to its
location in an active subduction zone between the Indo-Australian plate and the
Eurasian plate and its location on a fault along the Sumatra fault [17]. In this work,
the possible link with the atmospheric parameters such as RH and OLR with respect
to the occurrence of the earthquakes in the Sumatra-Indonesia region is explored.
Anomalous Deviations in Atmospheric Parameters as Pre-earthquake … 231
1.1 Study Area
The study area taken for this work is Sumatra-Indonesia. Indonesia is located at
the intersection of three massive tectonic plates: the Indo-Australian Plate on the
southern side, the Eurasian Plate on the Northern side, and the Pacific Plate from
northeast to southwest [18]. Due to its location on the plate boundary between the
Indo-Australian plate to the north and the Eurasian plate to the south, the island of
Sumatra is situated in an active seismic subduction zone.
Initially, the Indo-Australian plate was moving at 86 mm per year, but due to
the collision event, this fell to 40 mm per year. At the beginning of a new tectonic
configuration process, this drop continues to occur up to 30 mm per year. In addition,
the rate of return has dramatically risen to 76 mm per year [19]. Sumatra resides in an
active subduction zone between the Indo-Australian and Eurasian plates, making it
prone to earthquakes [17]. The data from the 75 km radius of the Sumatra-Indonesia
earthquake period from 1991 to 2021 is taken from USGS. The earthquake magnitude
employed in this study is magnitude > 6 Mw, with a configuration depth of 0–35 km
[20]. As shown in Fig. 1, the following map denotes the eight earthquakes in the
Sumatra-Indonesia region taken for study as mentioned in Table 1.
2 Methodology
OLR represents energy emitted to space by the earth’s surface and atmosphere. The
energy flux of OLR from the earth’s surface is measured in Wm − 2. The raw data
are converted into a standard anomaly index. The data collected is in a grid resolution
of 1° × 1° (latitude × longitude). OLR data are time-series data that cover the entire
earth.
Fig. 1 Map shows eight

earthquakes within a 75 km
radius of the December 26,
2004 earthquake from 1991
to 2021. Here the red star
denotes the epicenter
Table 1 List of earthquakes taken for study

Date Latitude Longitude Depth Magnitude
01–09–1993 2.986 96.122 34 6.3
02–11–2002 2.824 96.085 30 7.4
26–12–2004 3.295 95.982 30 9.1
07–04–2007 2.916 95.7 30 6.1
20–02–2008 2.768 95.964 26 7.4
09–12–2009 2.759 95.91 21 6
09–05–2010 3.748 96.018 38 7.2
25–07–2012 2.707 96.045 22 6.4
Exploratory data analysis is carried out in the OLR raw data, and Z-factor normal-
ization is applied to the raw OLR data to find the anomalous OLR flux index (E Flmτ ).
From the mean OLR flux of the previous “n” years, anomalous variations in OLR
flux have been computed by,

n
E Flmτ = E Flmτ (1)
τ =1
where
For a given location (l, m) and time (τ ), “n” is the number of predefined ten years
for which mean OLR flux is determined
E Flmτ − E Flmτ
dE_index(Flux index)(E Flmτ ) = (2)
σlmτ
where
[Elmτ ] Flux index for a given latitude (l), longitude (m), and time of data
acquisition (τ ).
Elmτ Current OLR flux for a given location (l, m) and time (τ ).
E Flmτ Mean OLR flux for a given location (l, m) and time (τ ).
Anomalous flux index energy “[Elmτ ]∗ ” can be determined by removing the

energy flux index value below the +2σ level of mean OLR flux, which helps maintain
the duration of anomalous change observed.
If Elmτ ≥ Elmτ + 2σ
Then,
Elmτ = [Elmτ ]∗ ELSE Elmτ = 0 (3)

where
[Elmτ ]∗ Anomalous energy flux index observed for given location and time.
A similar approach can be used to measure RH. Keeping the flux index energy
“[Elmτ ]∗ ”, the anomalous RH index can be determined by removing the energy
index value below −2σ level of mean RH.
3.1 01–09–1993: M6.3 Simeulue, Indonesia Earthquake
On September 1, 1993, a magnitude 6.3 earthquake struck Simeulue, Indonesia, at

2.986°N 96.122°E occurred due to a thrust fault with 34.0 km depth. The anomalous
variations in RH were observed on 29–07–1993, which is 34 days before the occur-
rence of the earthquake, followed by 02–08–1993 (30 days prior) with an anomalous
index of −2.378 and −2.253, respectively (Fig. 2). The anomalous variations in OLR
were observed on 01–08–1993 which is 31 days before the occurrence of earthquake
and anomaly occurred for three days consecutively on 02–08–1993, 03–08–1993, and
04–08–1993 with an anomalous index of 2.468, 3.167, 2.071, respectively (Fig. 2).
5
Date: 01-09-1993
4 Lat. Long: 2.986 N 96.122E
RH
Magnitude: 6.3Mw OLR
3
1
dE_Index
-1
-2
-3
-4
-5
01-03-1993 01-04-1993 01-05-1993 01-06-1993 01-07-1993 01-08-1993 01-09-1993
DATE
Fig. 2 Anomalous variations of RH and OLR before the 01–09–1993, Simeulue, Indonesia earth-
quake (6.3 Mw). The other anomalies are suppressed to understand better the pattern of anomalous
variations corresponding to this particular earthquake
5
Date: 02-11-2002 RH
4 Lat. Long: 2.824 N 96.085E
Magnitude: 7.4 Mw OLR
3
1
dE_Index
-1
-2
-3
-4
-5
02-05-2002 02-06-2002 02-07-2002 02-08-2002 02-09-2002 02-10-2002 02-11-2002
DATE
Fig. 3 Anomalous variations of RH and OLR before 02–11–2002, Simeulue, Indonesia earth-
A magnitude 7.4 Mw earthquake struck Simeulue, Indonesia, at 2.824°N latitude,

96.085°E longitude on 02–11–2002, caused by a shallow depth of 30 km. This
earthquake is considered a foreshock of the great 26–12–2004 Sumatra–Andaman
Islands earthquake [21]. The anomalous variations in RH were started 42 days before
the main shock on 21–09–2002 with an anomalous index of −3.4385, followed by a
consecutive anomaly on 25–09–2002 with an anomalous index of −2.688. On 20–10–
2002, the anomalous OLR flux was found to be 2.6667, followed by a further increase
in tectonic activity has shown consecutive abnormal emissions on 21–10–2002 with
an anomalous index of 2.7181, followed by 24–10–2002 with an anomalous OLR
index of 2.4553 (Fig. 3).
3.3 26–12–2004: M9.1 Sumatra–Andaman Islands

Earthquake
A devastating subduction zone earthquake occurred on December 26, 2004, with a

magnitude of 9.1 at a depth of around 30 km and is located at 3.295N latitude and
95.982E longitude. The anomalous variations in RH started on 05–10–2004, which
occurred 82 days before the mainshock with an anomalous index of −2.21802,
followed by 11–10–2004 with an anomalous index of −3.04569. The anomalous
5 Date: 26-12-2004
Lat. Long: 3.295 N 95.982 E
RH
4 Magnitude: 9.1Mw OLR
3
1
dE_Index
-1
-2
-3
-4
-5
26-06-2004 26-07-2004 26-08-2004 26-09-2004 26-10-2004 26-11-2004 26-12-2004
DATE
Fig. 4 Anomalous variations of RH and OLR before the 26–12–2004, Sumatra–Andaman Islands
earthquake (9.1 Mw). The other anomalies are suppressed to understand better the pattern of
anomalous variations corresponding to this particular earthquake
variations in OLR are observed on the same day on 05–10–2004, where anomalous
RH is observed with an anomalous index of 5.1214. The anomalous variations appear
continuously on 06–10–2004 with an anomalous index of 2.77629, followed by
16–10–2004, 20–10–2004, 21–10–2004, 22–10–2004 with an anomalous index of
2.78658, 2.46929, 4.6173, 3.70549, respectively, from 82 to 67 days before the
mainshock. Again, the anomalous pattern started 50 days before the mainshock
RH anomaly was observed on 06–11–2004 with an anomalous index of −2.49715,
followed by an OLR anomaly on 19–11–2004 with an anomalous index of 2.33085
which is 37 days before the main shock (Fig. 4).
On April 7, 2007, a shallow thrust fault earthquake hit Simeulue, Indonesia, with a
magnitude 6.1 Mw. At a focal depth of 30.0 km, the epicenter was at 2.916°N and
95.700°E. The anomalous variations in RH and OLR were observed on the same day
on 29–03–2007, nine days before the main shock, with an anomalous index of −
2.37585. It is followed by a continuous anomalous variation in RH on 30–03–2007,
31–03–2007, 01–04–2007, and 04–04–2007 with an anomalous RH index on −3.14,
−3.4205, −2.7417, −3.98304, respectively, followed by OLR anomalous variations
on the day of the main event with an anomalous index of 3.02711 (Fig. 5).
5
Date: 07-04-2007 RH
4 Lat. Long: 2.916 N 95.7E
Magnitude: 6.1
OLR
3
1
dE_Index
-1
-2
-3
-4
-5
07-10-2006 07-11-2006 07-12-2006 07-01-2007 07-02-200707-03-2007 07-04-2007
DATE
Fig. 5 Anomalous variations of RH and OLR before the occurrence of 07–04–2007, Simeulue,
Indonesia earthquake (6.1 Mw). The other anomalies are suppressed to understand better the pattern
of anomalous variations corresponding to this particular earthquake
3.5 20–02–2008: M7.4 Simeulue, Indonesia
On February 20, 2008, a magnitude 7.4 earthquake occurred in Simeulue, Indonesia,

at 2.76°N 95.964°E. The earthquake was caused by shallow thrust faulting along
the boundary between the Australia and Sunda plates. The anomalous variation in
RH started to occur on 08–01–2008, which is 43 days before the mainshock with
the anomalous index of −2.22351. The anomalous variations appear continuously
on 09–01–2008, 13–01–2008 with an anomalous index of −2.27413, −2.39277
respectively, from 42 to 38 days before the mainshock. The anomalous variations
succeed in OLR on 18–01–2008, which is 33 days before the main shock with an
anomalous index of 2.3788 (Fig. 6).
3.6 09–12–2009: M6.0 Simeulue, Indonesia
A 6.0 Mw magnitude earthquake due to a shallow thrust fault occurred in Simeulue,

Indonesia, on December 9, 2009, located at 2.759°N, 95.910°E at a depth of around
21.0 km. The continuous anomalous variations in RH occurred on 14–10–2009 which
is 56 days before the main shock with an anomalous index of −2.0964 followed by
16–10–2009 with an anomalous RH index of −4.25346, respectively. The anomalous
OLR is observed on 01–11–2009, anomalous index of 2.06218, which is 38 days
before the main shock (Fig. 7).
5
Date: 20-02-2008 RH
4 Lat. Long: 2.768 N 95.964E
Magnitude: 7.4 OLR
3
1
dE_Index
-1
-2
-3
-4
-5
20-08-2007 20-09-2007 20-10-2007 20-11-2007 20-12-2007 20-01-2008 20-02-2008
DATE
Fig. 6 Anomalous variations of RH and OLR before the 20–02–2008, Simeulue, Indonesia earth-
5
Date: 09-12-2009
4 Lat. Long: 2.759 N 95.91E RH
Magnitude: 6 OLR
3
1
dE_Index
-1
-2
-3
-4
-5
09-06-2009 09-07-2009 09-08-2009 09-09-2009 09-10-2009 09-11-2009 09-12-2009
DATE
Fig. 7 Anomalous variations of RH and OLR prior to the occurrence of 09–12–2009, Simeulue,
Indonesia earthquake (6.0 Mw). The other anomalies are suppressed for better understanding of the
pattern of anomalous variations corresponding to this particular earthquake
5
Date: 09-05-2010 RH
4 Lat. Long: 3.748 N 96.018E OLR
Magnitude: 7.2
3
1
dE_Index
-1
-2
-3
-4
-5
09-11-2009 09-12-2009 09-01-2010 09-02-2010 09-03-2010 09-04-2010 09-05-2010
DATE
Fig. 8 Anomalous variations of RH and OLR before the M7.2 Northern Sumatra, Indonesia earth-
quake on 09–05–2010. The other anomalies are suppressed to understand better the pattern of
anomalous variations corresponding to this particular earthquake
3.7 09–05–2010: M7.2 Northern Sumatra, Indonesia
The earthquake that hit northern Sumatra on May 9, 2010 was occurred by shallow
thrust faulting on or near the subduction interface plate boundary between the Indo-
Australian and Sunda plates. The magnitude 7.2 earthquake occurred at 3.748°N
latitude and 96.018°E longitude, about 38 km depth. The anomalous variations in
RH were observed on 22–03–2010 with an anomalous index of −2.5504 which is
48 days prior to the occurrence of earthquake. The anomalous variations in OLR
appear on 12–04–2010 with an anomalous index of 4.12514 which is 27 days prior
to mainshock (Fig. 8).
3.8 25–07–2012:M6.4 Simeulue, Indonesia Earthquake
On July 25, 2012, an earthquake with a magnitude of 6.4 occurred in Simeulue,

Indonesia, which has a latitude of 2.707°N and a longitude of 96.045°E with a depth
of 21.0 km.
The anomalous variations in RH were observed on 04–07–2012 with an anoma-
lous index of −2.5504, which is 48 days before the occurrence of an earthquake.
The anomalous variations in OLR appear on 12–04–2010 with an anomalous index
of 4.12514, which is 27 days before the mainshock (Fig. 9).
5
Date:25-07-2012 RH
4 Lat. Long: 2.707 N 96.045E
OLR
Magnitude: 6.4
3
1
dE_Index
-1
-2
-3
-4
-5
25-01-2012 25-02-2012 25-03-2012 25-04-2012 25-05-2012 25-06-2012 25-07-2012
DATE
Fig. 9 Anomalous variations of RH and OLR before 25–07–2012, Simeulue, Indonesia Earth-
3.9 Discussion
From the above results, it is inferred that

1. The anomalous variations in RH and OLR appear three days to 3 months before
the occurrence of the earthquake.
2. The anomalous variations of OLR succeed the anomalous variations in RH, which
indicates the possible radon activity on atmospheric gases due to air ionization
and ion-formed condensation leading to latent heat exhalation from water vapor
condensation causes thermal anomalies. Hence, these anomalous variations could
be potential precursors for the impending earthquakes.
3. There is a strong correlation between the strike of the fault with the number of
days prior to the earthquake events and the anomalous drop in RH observed as
mentioned in Table 2.
4. Similar type of relation between the strike of the fault and the number of days
for which the OLR anomalies are observed.
5. Based on the clusters mentioned in the above two statements (3 and 4), the
number of days of OLR anomalies observed in the past six months has a positive
correlation with the magnitude of the earthquakes (Table 2).
6. Our results vindicate the proposal given by the LAIC model–LAI coupling).
The most important mechanism is the ionization of the air, which is caused by
an increase in the emission of gases such as radon from the earth’s crust in
the area of active tectonic faults. The enhanced gas emission sparked a series of
physical processes and chemical reactions from the ground surface, including the
ionization of air molecules, which led to the development of complex molecular
Table 2 Earthquakes grouped based on near similar strike value

Date Latitude Longitude Depth Magnitude Strike Rh-EQ event No. of OLR
anomalies in
6 months
01–09–1993 2.986 96.122 34 6.3 131 34 11
02–11–2002 2.824 96.085 30 7.4 133 42 17
26–12–2004 3.295 95.982 30 9.1 132 82 18
09–12–2009 2.759 95.91 21 6 135 56 10
09–05–2010 3.748 96.018 38 7.2 128 48 15
07–04–2007 2.916 95.7 30 6.1 138 9 4
25–07–2012 2.707 96.045 22 6.4 137 21 12
20–02–2008 2.768 95.964 26 7.4 137 43 18
ions, and the formation of massive ion clusters that can grow up to the size
of aerosols. The rise in surface temperature is causing a significant increase in
outgoing thermal radiation, particularly in the epicentral regions. Ionization of
the air also results in thermal effects, which cause the temperature and humidity
of the air, which might lead to an anomalous drop in RH and an anomalous
increase in OLR.
4 Conclusion
The time-series analysis is performed on the daily datasets of RH and OLR for eight
earthquakes of magnitude 6.0 that occurred between 1991 and 2021 within a 75-km
radius of the epicenter of the December 26, 2004 earthquake. This study investi-
gates and comprehends the precursory signals related to earthquake occurrence. The
observation of anomalous deviations in OLR and RH is analyzed before the event
of each earthquake. Through the LAIC model, the correlation between anomalous
variations in atmospheric parameters such as RH and OLR flux index concerning the
occurrence of earthquakes is investigated. The anomalous variations in RH and OLR
arise between 3 days to 3 months before an earthquake. Anomalous variations follow
the anomalous variations in RH in OLR, which suggests possible radon activity on
atmospheric gases due to air ionization and ion-formed condensation, which in turn
leads to a drop in humidity and increase in thermal anomalies, especially around the
epicentral region. Our findings are consistent with the LAIC model, which explains
the majority of these events as a synergy between various processes and anoma-
lous variations that are referred to as short-term earthquake precursors. The interval
between anomalous variations in RH and the occurrence of an earthquake increases
as the magnitude of earthquakes with the near similar strike value. Additionally, the
number of OLR anomalies in the preceding six months increases as the magnitude
increases. Consequently, these unusual variations could be considered precursors of

impending earthquakes to forecast on a short-term basis.
Acknowledgements The authors are thankful to the Ministry of Earth Sciences, India, for finan-
cial assistance (Project No: MoES/P. O(seismo)/1(343)/2018) to carry out research on earthquake
forecasting. We would like to acknowledge SASTRA Deemed to be University for facilitating and
encouraging to do this research work.
References
1. Zhang, X., Kang, C., Ma, W., Ren, J., Wang, Y.: Study on thermal anomalies of earthquake
process by using tidal-force and outgoing-longwave-radiation. 34(3), 234–235 (1997)
2. Carreno, E., Capote, R., Yague, A.: Observations of thermal anomaly associated to seismic
activity from remote sensing. In: General Assembly of European Seismology Commission
Portugal, pp. 265–269, (2001)
3. Fu, C.C., Lee, L.C., Ouzounov, D., Jan, J.C.: Earth’s outgoing longwave radiation variability
prior to M ≥ 6.0 earthquakes in the Taiwan area during 2009–2019. Front. Earth Sci. 8, (Sept
2020). https://doi.org/10.3389/feart.2020.00364
4. Pulinets, S.A., Boyarchuk, K.A., Lomonosov, A.M., Khegai, V., Liu, A.Y.: Ionospheric precur-
sors to earthquakes: a preliminary analysis of the JoF2 critical frequencies at Chung-Li
ground-based station for vertical sounding of the ionosphere (Taiwan Island) (2002)
5. Pulinets, S.A., Boyarchuk, K.A.: Ionospheric Precursors of Earthquakes. Springer, Berlin,
Heidelberg, New York (2004)
6. Pulinets, S.A., Romanov, A.A., Urlichich, Y.M., Romanov, A.A., Doda, L.N., Ouzounov, D.:
The first results of the pilot project on complex diagnosing earthquake precursors on Sakhalin.
Geomag. Aeron. 49(1), 115–123 (2009). https://doi.org/10.1134/S0016793209010162
7. Ouzounov, D., Liu, D., Chunli, K., Cervone, G., Kafatos, M., Taylor, P.: Outgoing long wave
radiation variability from IR satellite data prior to major earthquakes. Tectonophysics 431(1–4),
211–220 (2007). https://doi.org/10.1016/j.tecto.2006.05.042
8. Chen, Y., Shen, X., Jing, F., Xiong, P.: Application of outgoing longwave radiation data
for earthquake research. In: Proceedings–2010 IEEE International Conference on Intelligent
Computing and Intelligent Systems, ICIS 2010, vol. 2, pp. 46–48 (2010). https://doi.org/10.
1109/ICICISYS.2010.5658836
9. Ouzounov, D., Pulinets, S.A., Liu, J.Y., Hattori, K.: NASA ADS: validation of atmospheric
signals associated with major earthquake activities. A statistical study, 2009. [Online].
Available: https://www.researchgate.net/publication/253886298
10. Ouzounov, D., Pulinets, S., Hernandez-Pajares, M., Hattori, K., Garcia-Rigo, A.:
Ouzounov@chapman Edu, Geo-space observation of atmospheric environmental effects
associated with 2011 Fukushima nuclear accident !
11. Pulinets, S.A., et al.: Thermal, atmospheric and ionospheric anomalies around the time of the
Colima M7.8 earthquake of 21 January 2003. Ann. Geophys. 24(3), 835–849 (2006). https://
doi.org/10.5194/angeo-24-835-2006
12. Ouzounov, D., Pulinets, S., Kafatos, M.C., Taylor, P.: Thermal radiation anomalies associated
with major earthquakes, pp. 259–274, (2018). https://doi.org/10.1002/9781119156949.ch15
13. Venkatanathan, N.: Indispensable coordinated interdisciplinary studies for earthquake predic-
tion. Int. J. Earth Sci. Eng. 5(3), 365–372 (2012)
14. Venkatanathan, N., Natyaganov, V.: Outgoing longwave radiations as pre-earthquake signals:
preliminary results of 24 September 2013 (M 7.7) earthquake. Curr. Sci. 106(9), 1291–1297
(2014). https://doi.org/10.18520/cs/v106/i9/1291-1297
15. Prakash, R., Srivastava, H.N.: Diurnal variations of outgoing long wave radiation (OLR) vis a
vis 4 January, 2016 Manipur earthquake (Mw: 6.7): an earthquake precursor? (2017)
16. Salh, H., Muhammad, A., Ghafar, M.M., Külahcı, F.: Potential utilization of air temperature,
total electron content, and air relative humidity as possible earthquake precursors: a case study
of Mexico M7.4 earthquake. J. Atmos. Solar Terr. Phys. 237, 105927 (2022). https://doi.org/
10.1016/J.JASTP.2022.105927
17. Indah, F.P., Syafriani, S., Andiyansyah, Z.S.: Analysis of seismotektonic patterns in sumatra
region based on the focal mechanism of earthquake period 1976–2016. IOP Conf. Ser.: Mater.
Sci. Eng. 335(1) (2018). https://doi.org/10.1088/1757-899X/335/1/012011
18. Plummer, C.C., Carlson, D.H., Hammersley, L.: Physical Geology. McGraw-Hill Book
Company, Inc., New York (2003)
19. Natawidjaja, D.H.: Neotectonics of the Sumatran fault and paleogeodesy of the Sumatran
subduction zone (2002)
20. USGS, https://neic.usgs.gov/
21. Sieh, K.: The Sunda megathrust-past, present and future, (2007). [Online]. Available: www.
worldscientific.com
Spatial Distribution of Stress Orientation
by Inversion of Focal Mechanism
Solutions Using MSATSI: A Case Study
Across Japan Trench
Sucheta Das, Sandeep, Sonia Devi, Himanshu Mittal, Praveen Kumar,

and Monika
Abstract Estimation of stress field orientations is a necessary aspect for recognition

of crustal mechanics as well as the physics behind occurrence of earthquakes. A
case study employing the new MATLAB software package Spatial And Temporal
Stress Inversion (SATSI) for stress inversion utilizing the focal mechanism data is
presented here to produce stress orientation models in Northeast (NE) Japan. In this
work, the study region is divided into 49 small sub-regions so that the stress tensors
and focal mechanisms can independently fit in each sub-region. Determination of
any stress variation is strongly needed by the data while eliminating the artifacts due
to overfitting of noisy or nonuniquely fitting data. To resolve it, a damped inversion
procedure was applied which inverted the stresses in all sub-regions, while at the
same time reducing the difference in stress between adjacent sub-regions. Earthquake
focal mechanisms have been used to determine the stress patterns at depths capable
of generating earthquakes in NE Japan since 1960–2021. In this work, 0D, 1D, and
2D stress inversion using the MSATSI (MATLAB package for Spatial And Temporal
Stress Inversion) routine was performed and examined the spatial variation of stress
orientations over NE Japan along the Japan Trench and put forward recent knowledge
about the stress pattern. From the obtained 2D inversion results, a spatially varying
stress regime is observed in the crust which demonstrates normal faulting on the
subducting Pacific plate which changes to reverse faulting on the Okhotsk plate
through an intermediate state of oblique faulting.
S. Das · Sandeep (B) · S. Devi · Monika

Department of Geophysics, Banaras Hindu University, Varanasi, India
e-mail: sandeepfgp@bhu.ac.in
Monika
e-mail: dassucheta@iitgn.ac.in
H. Mittal
National Centre for Seismology, MoES, Noida, India
P. Kumar · Monika
Wadia Institute of Himalayan Geology, Dehradun, India
S. Das
Discipline of Earth Sciences, IIT Gandhinagar, Gandhinagar, India
https://doi.org/10.1007/978-981-99-1459-3_20
244 S. Das et al.
Keywords MSATSI · Stress inversion · Focal mechanism solution
1 Introduction
Occurrence of earthquake takes place when tectonic stresses in the crust release
fragmentarily due to abrupt brittle rupture. Earthquakes alter the stress field as they
reshuffle the stresses that have developed throughout the interseismic period. There-
fore, to understand the physical process of earthquake initiation and also the up-to-
date tectonic movement, particulars about the stress state are necessary. In the 1980s,
the usual procedures of inverting stress (e.g., Gephart and Forsyth [12]; Michael [14],
1987) replaced the classical methods due to the efficacy to determine the average
stress orientations of already separated blocks using the Focal Mechanism Solution
(FMS) data of occurred earthquakes by the means of least square procedure.
The study area extends from 37.2° to 40° north and 142–144.8° east, i.e., across
a part of the NE Japan islands governed by the Japan Trench. Japanese islands
have a very complex tectonic setting. The Pacific plate is going below the North
American plate (NAP) across the Kuril–Japan Trench, whereas the Philippine Sea
plate across the Izu-Bonin Trench, and the Philippine Sea plate underneath the
NAP across the Sagami trough and the Eurasian plate across the Suruga-Nankai
trough and its South-Western (SW) prolongation, the Ryukyu Trench. The system
of tectonic stress fields in Japan and its adjacent regions is caused by this type
of complex tectonic background. Various scholars (e.g., Yoshii [15]; Huzita [16];
Wesnousky et al. [17]; Nishimura et al. [18] and Townend [19]) have separately
evaluated the stress regime of the seismically active area in and around Japan over
the last three decades by analyzing the geophysical and geological data. Subduction
of the oceanic plate has disfigured the seaward edge of the continental plate near
the Japan Trench convergent plate margin. A multichannel seismic survey reported
that the northern Japan Trench is bounded by the strong continental substructure
by a reflector dipping landward noticeably. Due to the interaction of two planes by
subduction phenomena, there is a development of progressive distortion or demo-
lition of a horst structure across the upper part of the oceanic crust which is being
subducted and also deformation in the oceanic plate [1]. An even reflector indicating
a steady slip plane is seen across and over the oceanic crust at greater than 45 km
toward the land from the trench axis. From this location, seismicity of interplate
earthquakes swiftly increases landward. The data of Japan islands used here are
collected from Global CMT Catalog Search (https://www.globalcmt.org/CMTsea
rch.html/) and USGS Earthquake catalog 7 (https://earthquake.usgs.gov/) during the
temporal variation of 61 years (11th March 1960–11th March 2021). The evaluation
of the stress field across Japan Trench is prepared by compiling the data from these
networks. In this work, we have performed 0D, 1D, and 2D stress inversion using
the MSATSI (Matlab package for Spatial And Temporal Stress Inversion) by Patricia
Martínez-Garzón, Grzegorz Kwiatek, Michèle Ickrath, and Marco Bohnhoff routine
Spatial Distribution of Stress Orientation by Inversion of Focal … 245
and examined the spatial changes of stress patterns over NE Japan along the Japan
Trench and put forward recent knowledge about the stress pattern.
2 Stress Inversion
Stress inversion procedures are applied to assess the state of stress from numerous
focal mechanisms of earthquakes (established from the Bott hypothesis, which
presumes that the slip vector rests on the fault plane and the maximum resolved
shear stress in the same plane is parallel to it). It does not appraise non-double-couple
components of source. After being selected and discretized, the focal mechanisms
are feasible for inversion of the stress field orientation (Fig. 1).
In this method, a given area is divided into a number of sub-regions so that the
stress tensor and focal mechanism data can be independently fitted. This method
might supposedly give rise to spatial variability which is basically an artifact due
to overfitting of data embedded with noise or nonuniquely fitting data. Therefore,
in order to remove all these artifacts while determining any stress variation which
is strongly needed by the data, a damped inversion procedure is introduced to for
stress inversion in all sub-regions while minimizing the difference in stress between
neighboring sub-regions at the same time.
The nonlinear stress inversion problem is changed into a linear one by presuming
that the shear traction magnitude of every fault plane is nearly equal. This method
focuses on searching the stress regime which reduces the inconsistency between the
direction of resolved shear stress and the slip for the total number of seismic events
in the dataset. If the FMSs only provide the fault orientation and the slip direction
but not the relative displacement magnitude (Δu), then just a segment of the regional
stress tensor is approximated concerning the direction of the principal axes and also
their relative magnitudes [3]. To be more specific, the inversion gives us a ‘reduced’
deviatoric stress tensor (T ), which is specified by four parameters. Among them,
three parameters explain the principal stress axes orientations (S1, S2, S3), and the
fourth one is stress ratio R (known as shape factor or shape ratio, stress shape, aspect
ratio or stress magnitude parameter) [12]:
σ2 − σ1
R=
σ3 − σ1
Here σ 1 is the magnitude of maximum stress axis, σ 2 is the magnitude of the inter-
mediate stress axis, and σ 3 is the magnitude of the minimum stress axis derived from
the deviatoric stress tensor. Value of R states whether the magnitude of σ 2 is closer
to the magnitude of σ 1 or σ 3 .
If R is increased up to 1 from 0, σ 2 will decrease in the limit between σ 1 and σ 3 ,
where R having value near 0 denotes σ 2 ≈ σ 1 and R having value near 1 indicates
σ 2 ≈ σ 3 (Fig. 2). Both of the ranges stand for the conditions of uniaxial extension
(where R = 0) and uniaxial compression (where R = 1), respectively, while R = 0.5
246 S. Das et al.
Fig. 1 Flowchart abridging various steps made up of a the selection of data and b inversion of the
stress tensor from FMS (after Martínez-Garzón et al. [2])
Fig. 2 Stress ellipsoids and the corresponding Mohr diagram for various stress orientations spec-
ified by the stress ratio (R), for: a uniaxial extension (R = 0, oblate ellipsoid), b triaxial stress (R
= 0.5), and c uniaxial compression (R = 1, prolate ellipsoid). It is to be noted that here the stress
magnitudes seem to have all positive values. Figure after Lejri [20]
indicates that σ 2 = (σ 1 + σ 3 )/2. If the principal stresses combinedly give rise to

either R > 1 or R < 0, then they are inaccurate [13]. A comparable parameter Φ is put
forward as given below (Angelier et al. [11]; Angelier [9, 10]; Michael [14], 1987):
σ2 − σ3
R=
σ1 − σ3
The combination of the above two equations provides the relation Φ = 1 − R.

Both Lame’s stress ellipsoid and Mohr diagram represent a specific state of stress
and also provide us a geometrical portrayal of R which shows the connection with
stress magnitudes. The bootstrap resampling procedure implemented to the input
FMS causes uncertainties not only in the stress axes patterns, but also in the relative
stress magnitude.
3 Methodology
To estimate the spatial and temporal stress orientation from FMS data, several algo-
rithms are accessible. The algorithm of the method used here involves the collection
of FMS data (dip direction, dip angle, and rake) followed by division of study area
in grids and then running the MSATSI routine in MATLAB as per our need of 0D,
1D, and 2D inversion.
SATSI is an alteration of Michael’s code which inverts FMS for stress field varying
both in space and/or time. The stress orientations using SATSI are determined by
splitting an area in a number of sub-areas and then inverting the FMS at the same
time for each sub-areas, looking for the optimal stress through every cell during the
process of reducing the difference in stresses between neighboring areas [4]. Other
than the problem regarding noise, it also smoothens out disruption of abnormal
stresses which might be insignificant for the arrangement of regional stress states.
248 S. Das et al.
A damped least-squares inversion strategy was employed to invert a stress tensor

for every grid point at the same time to acquire a smooth solution. SATSI was executed
because of the robustness of the formal stress inversion (FSI) procedure along with
the damped least squares optimization and also due to its advantages over processing
of datasets.
The input FMS can be assembled earlier to the inversion into a number of ‘grid
points’ which are dispersed over a number of dimensions, from 0D (single FSI in
one grid point is conducted) and 1D (e.g., temporal changes of stress field) up to 4D
which gives the spatiotemporal distribution (Fig. 3), with time considered as an extra
coordinate.
The study area extends from 37.2°–40° north and 142–144.8° east, i.e., across a
part of the NE Japan islands. The selected area is divided into 7 × 7, i.e., 49 grids
having the dimension of 0.4° × 0.4°. The data of Japan islands used here are collected
from Global CMT Catalog Search (https://www.globalcmt.org/CMTsearch.html/)
and USGS Earthquake catalogue (https://earthquake.usgs.gov/) during the temporal
variation of 61 years (11th March 1960–11th March 2021). The input matrix for
Fig. 3 Demonstration of feasible dimensions of the FSIs conducted with SATSI and MSATSI.
(1) 0D gives a single static stress field. (2) 1D gives a set of stress tensors scattered over single
coordinate (e.g., temporal distribution). (3) 2D forms a map of stress-tensor outputs changing over
two coordinates (e.g., the surface distribution of stress). (4) 3D performs the spatial distribution of
the stress field. (5) 4D performs a spatiotemporal distribution of stress-tensor orientations. (From
Martinez-Garzon et al. [5])
stress tensor inversion changes depending on the type of inversion performed, i.e.,
whether 0D/1D/2D. In case of 0D, 1D, and 2D, we have used the n-by-5 size input
matrix where n denotes the number of input data (i.e., focal mechanisms). There are
no first-hand details on the location of certain earthquakes, but only the details about
the grid point the specified FMS are associated with and then the solutions from
each grid point are inverted for the stress tensor. The different types of the plot are
described in user manual of MSATSI-([5], Sect. 4.7).
For 0D inversion, a set of 106 focal mechanisms in a single grid point [0 0] was
selected. A damped inversion cannot be performed, and because it was only one grid
point, neighboring grids were absent. Hence, the ‘Damping’ was naturally selected
as ‘off’. Therefore, there was not any trade-off curve output. P and T axes for this
grid (‘PTPlots’) were also activated, and the uncertainty was calculated by 2000
bootstrap resamplings.
For 1D inversion, the FMSs are arranged into seven different grid points which
vary over Y coordinate; this makes the case 1D. Each grid has 20 focal mechanisms.
A damped inversion which was settled by default was performed, and uncertainty
was calculated with 2000 bootstrap resamplings. Here ‘ConfidenceIntervals’ was
turned off so that the pictures only include the trajectory marked out by the best
solution across the grid points.
For 2D inversion, the FMSs were divided into 49 grid points in total which changes
over both X and Y coordinates and thus makes the case 2D. Every grid has 20 focal
mechanisms. Here also a damped stress inversion was performed by default, and
uncertainty was settled by 2000 bootstrap resamplings. Furthermore, a minimum
limit of 30 events per node was set.
4 Estimated Stress Patterns and Their Comparison

with Tectonic Motions
The FMS data inversion method is applied to around 1000 earthquakes inside the
magnitude range of 0.3–9.0 in the NE Japan along the Japan Trench (11th March
1960–11th March 2021). 0D inversion of the grid point with X = 0, Y = 0 gave the
following stereonet for a set of 106 FMS, and also best R value = 0.59 is obtained
(Fig. 4).
Another stereonet shows the uncertainties mapped as colored dots straight from
the bootstrap resampling, where red, green, and blue, respectively, denotes σ 1, σ 2,
and σ 3. This type of stress orientation indicates reverse faulting regime (Fig. 5).
Results from the 1D inversion (Fig. 6) are in good agreement with the work of
Terakawa [6], Asano et al. [7] and Yoshida et al. [8] indicating reverse faulting mech-
anism caused by compressive shortening of the crust. In reverse faulting mechanism,
the minimum principal stress σ 3 (blue) is vertical and at the center of lower hemi-
sphere projection, whereas the maximum principal stress σ 1 (red) is horizontal. Trend
is the azimuth direction measured from north (−360 to 360 degrees), and plunge is
250 S. Das et al.
Fig. 4 Stereonet for [0 0].

Blue dots denote tension and
red dots denote pressure
Fig. 5 Stereonet for σ 1 , σ 2 ,

and σ3 . Best stress
orientation is indicated by a
+ sign
the downward angle measured from horizontal (−90 to 90 degrees). To further look
into the spatial development of stress field orientation, plunge and trend of σ 1−3 are
plotted with their 95% confidence interval (Fig. 7). The trend of σ 1 almost remains
constant at 90°, whereas the plunge drops from 45° to 30°. Trend and plunge of σ 2
show a variation of 180° and 60°, respectively. Trend of σ 3 starts with 112.5°, drops
to 40°, and then gradually increases and retains at 112°. Almost similar drift is seen
for the plunge of σ 3 which varies between 10° and 60°.
2D damped inversion is performed on 49 grids (0.4° × 0.4°) covering the total
area from 37.2–40° north to 142–144.8° east, i.e., approximately 96,600 km2 which
consists of a part along the Japan Trench. The uncertainty evaluation was carried out
Fig. 6 1D inversion results

of study area
by setting bootstrap resampling at 2000 and confidence interval at 95%. Figure 8

shows the orientation of the maximum horizontal stress for each of the FSIs carried
out by making use of the World Stress Map (WSM) plot. For most grid points, the
orientation of σ 1 is in well accordance with the regional stress field along Japan
Trench [6].
Figure 9 shows the location of principle stress axes using Stereomap in Universal
Transverse Mercator (UTM) coordinates. For the cases where bootstrap resampled
points are dispersed across huge regions (e.g., σ 2 axes with Easting = 589 km
and Northing = 4274 km), the portrayal of the confidence interval with bootstrap
resampling should come up with a better practical solution. In the southwestern (SW)
segment of the area (inversions with Easting = from 589 to 448 km, Northing = from
4118 to 4222 km), the bootstrap resampling for σ 2 axis is scattered across a large
number of azimuths, this suggests that in this area the magnitude of σ 2 axis is less
than σ 1 and σ 3 . Therefore, trend of σ 1 and σ 3 has higher degree of uncertainty. The
results are also in good agreement with the distribution of R values for every grid
where comparatively low R value (0.43, 0.29, 0.37, etc.) for the SW part is seen.
According to the stress pattern in Fig. 10, it can be seen that the Pacific plate
exactly before being subducted is experiencing tensile stress perpendicular to stress
axis which gives rise to normal faulting in the eastern part of study area. Also,
in Fig. 11, it can be seen that the stress patterns corresponding to oblique (grey)-
type faulting across plate boundaries indicate longstanding slip motion due to the
resistance caused by friction. These type of stress orientations specifying the zones
of plate boundary are appreciably prominent in Fig. 10. The stress orientation in
the Pacific plate going underneath the NE Japan arc alters from tensional forces
(red spheres denoting normal faulting regime) to compressional forces (blue spheres
with opposite mechanism, i.e., reverse faulting) in the Okhotsk plate through a neutral
252 S. Das et al.
Fig. 7 Stress tensor results from 1D inversion for selected grids in lower hemisphere projection.
Colored dots indicate sampled bootstrap points in the 95% confidence interval, and best stress
orientation is indicated by a + sign. Variations with trend and plunge in each grid are shown for
maximum compressive stress-σ 1 (red), intermediate compressive stress-σ 2 (green), and minimum
compressive stress-σ 3 (blue) along the grid points. Uncertainties in trend and plunge are represented
with error bars
state (gray spheres denoting oblique faulting pattern). The outcomes are also in good
agreement with those of WSM (Fig. 8).
5 Conclusion
With the MSATSI inversion routine based on Michael’s grid search algorithm [4],
we estimated the 2D tectonic stress pattern covering the total area from 37.2–40°
north to 142–144.8° east, i.e., approximately 96,600 km2 which consists a part along
the Japan Trench for around 1000 seismic events (0.3 ≤ M ≤ 9.0) from 11th March
Grid along y coordinate
Grid along x coordinate
Fig. 8 2D inversion results employing WSM. For every grid maximum horizontal stress is
illustrated. Red signifies normal faulting regime according to best solution
Northing (km)
Easting (km)
Fig. 9 2D inversion results using Stereomap in UTM coordinates. Best stress orientation is
indicated by a + sign
254 S. Das et al.
Fig. 10 Principle stress

orientation in Stereomap of
study area
Fig. 11 2D stress inversion

results indicating the type of
faults. Red denotes normal
faulting; blue denotes reverse
faulting, and gray denotes
oblique faulting
1960–11th March 2021. The goal of the method was to find the model with minimal
complex structure which is compatible with the selected data. The method also
includes the concept of damping which is used for minimizing the difference between
components of stress tensor in neighboring sub-regions, accompanied by the expected
reduction of the misfit of stress tensor to FMS of seismic events. A systematic stress
spatially varying stress regime is observed in the crust. The stress orientation denotes
normal faulting on the subducting Pacific plate which changes to reverse faulting
on the Okhotsk plate through an intermediate state of oblique faulting. Thus, it is
concluded that Pacific plate exactly before being subducted is experiencing tensile
stress. The calculated stress pattern has an acceptable correspondence with the latest
(Quaternary) complex tectonics of the islands in and around Japan from regional to
local scale which confirms efficacy of the method used. For most grids, the orientation
of σ 1 is in well accordance with the stress field along Japan Trench. Therefore, it is
concluded that the MSATSI routine can further be utilized for inverting the damped
models of stress by the means of temporal changes through performing 3D as well
as 4D inversion.
References
1. Tsuru, T., Park, J.-O., Takahashi, N., Kodaira, S., Kido, Y., Kaneda, Y., Kono, Y.: Tectonic
features of the Japan trench convergent margin off Sanriku, northeastern Japan, revealed by
multichannel seismic reflection data. J. Geophys. Res.: Solid Earth 105, 16403–16413 (2000)
2. Martínez-Garzón, P., Ben-Zion, Y., Abolfathian, N., Kwiatek, G., Bohnhoff, M.: A refined
methodology for stress inversions of earthquake focal mechanisms. J. Geophys. Res.: Solid
Earth 121, 8666–8687 (2016)
3. Kassaras, I.G., Kapetanidis, V.: Resolving the tectonic stress by the inversion of earthquake
focal mechanisms. Application in the Region of Greece. A Tutorial. Moment Tensor Solutions
(2018)
4. Hardebeck, J.L., Michael, A.J.: Damped regional-scale stress inversions: methodology and
examples for southern California and the Coalinga aftershock sequence. J. Geophys. Res.:
Solid Earth 111, n/a–n/a (2006)
5. Martinez-Garzon, P., Kwiatek, G., Ickrath, M., Bohnhoff, M.: MSATSI: a MATLAB package
for stress inversion combining solid classic methodology, a new simplified user-handling, and
a visualization tool. Seismol. Res. Lett. 85, 896–904 (2014)
6. Terakawa, T., Matsu’ura, M.: The 3-D tectonic stress fields in and around Japan inverted from
centroid moment tensor data of seismic events. Tectonics 29, n/a–n/a (2010)
7. Asano, Y., Saito, T., Ito, Y., Shiomi, K., Hirose, H., Matsumoto, T., Aoi, S., Hori, S., Sekiguchi,
S.: Spatial distribution and focal mechanisms of aftershocks of the 2011 off the Pacific coast
of Tohoku earthquake. Earth, Planet. Space 63(7), 669–673 (2011)
8. Yoshida, K., Hasegawa, A., Okada, T.: Spatial variation of stress orientations in NE Japan
revealed by dense seismic observations. Tectonophysics 647–648, 63–72 (2015)
9. Angelier, J.: From orientation to magnitudes in paleostress determinations using fault slip data.
J. struct. geol. 11, (1–2), 37–50 (1989). https://doi.org/10.1016/0191-8141(89)90034-5
10. Angelier, J.: Tectonic analysis of fault slip data sets. J. Geophys. Res.: Solid Earth 89(B7)
5835–5848 (1984). https://doi.org/10.1029/JB089iB07p05835
11. Angelier, J., Tarantola, A., Valette, B., Manoussis, S.: Inversion of field data in fault tectonics
to obtain the regional stress ? I. Single phase fault populations: a new method of computing
the stress tensor. Geophys. J. Int. 69(3) 607–621 (1982)
12. Gephart, J.W., Forsyth, D.W.: An improved method for determining the regional stress tensor
using earthquake focal mechanism data: Application to the San Fernando Earthquake Sequence.
J. Geophys. Res. 89(B11), 9305–9320 (1984)
13. Gephart, J.W.: Principal stress directions and the ambiguity in fault plane identification from
focal mechanisms. Bulletin of the Seismol. Society of America, 75(2), 621–625 (1985)
14. Michael, A.J.: Determination of stress from slip data: faults and folds. J. Geophys. Res.: Solid
Earth, 89(B13), 11517–11526 (1984)
256 S. Das et al.
15. Yoshii, T.: A detailed cross-section of the deep seismic zone beneath northeastern Honshu,
Japan. Tectonophysics 55(3–4), 349–360 (1979)
16. Huzita, K.: Role of the median tectonic line in the Quaternary tectonics of Japanese islands.
Memoirs Geol. Soc. Japan, 18, 129–153 (1980)
17. Wesnousky, S.G., Scholz, C.H., Shimazaki, K.: Deformation of an island arc: rates of moment
release and crustal shortening in intraplate Japan determined from seismicity and Quaternary
fault data. J. Geophys. Res.: Solid Earth 87(B8), 6829–6852 (1982)
18. Nishimura, T., Hirasawa, T., Miyazaki, S.i., Sagiya, T., Tada, T., Miura, S., Tanaka, K.:
Temporal change of interplate coupling in northeastern Japan during 1995–2002 estimated
from continuous GPS observations. Geophys. J. Int. 157(2), 901–916 (2004)
19. Townend, J., Zoback, M.D.: Stress, strain, and mountain building in central Japan. J. Geophys.
Res.: Solid Earth 111(B3) (2006)
20. Lejri, M.: Subsurface stress inversion modeling using linear elasticity: sensivity analysis and
applications, Université de Montpellier (2015)
Seismic Landslide Hazard Assessment
of Mandi Town, Himachal Pradesh
A. Kothiala , P. S. Nayek , Maheshreddy Gade , and U. V. Kala
Abstract This study performs slope displacement-based probabilistic seismic land-

slide hazard assessment for Mandi town. Here, the slope angles of the concerned
region are obtained from the digital elevation map (DEM), and the material proper-
ties are obtained from lithological information and literature. The critical acceleration
of the slopes is estimated by combining the obtained slope angle and material prop-
erties. In this work, hazard assessment is performed considering the arid and fully
saturated condition of the soil mass of the slopes. The PGA values are estimated for
100, 475, and 2475 years return periods by performing probabilistic seismic hazard
analysis of the study region. Further, the PGA values and the slope displacement
prediction equation are used to estimate Newmark’s sliding displacement. Finally,
the hazard map in terms of the probability of slope displacement (SD) value exceeding
the threshold values of 5 cm is presented. The developed seismic landslide hazard
map highlights the areas that may experience co-seismic landslides in the future. The
probability of occurrence of co-seismic landslides gets as high as 93.4% for saturated
soil and 86.3% for arid soil for a return period of 2475 years. This hazard map will
help local authorities and planners with tools for assessing the seismic landslide risk
associated with land use and taking necessary measures to minimize the damages.
Keywords Co-seismic landslide · Seismic landslide hazard · Newmark sliding

displacement · Mandi
A. Kothiala (B) · M. Gade · U. V. Kala

School of Civil and Environmental Engineering, IIT Mandi, Kamand, India
e-mail: pts1802@students.iitmandi.ac.in
M. Gade
e-mail: maheshreddy@iitmandi.ac.in
U. V. Kala
e-mail: uday@iitmandi.ac.in
P. S. Nayek
Department of Civil Engineering, Sikkim Manipal Institute of Technology, East Sikkim, India
e-mail: partha.sn@smit.smu.edu.in
https://doi.org/10.1007/978-981-99-1459-3_21
258 A. Kothiala et al.
1 Introduction
The Himalayan region is one of the most seismically active regions in the world
[3, 4, 24]. It has endured many destructive quakes in the past, including the 1897
Shillong earthquake (M w 8.1), 1905 Kangra earthquake (M w 7.8), 1934 Bihar–Nepal
Earthquake (M w 8.2), 1950 Assam Earthquake (M w 8.6), 1991 Uttarkashi Earthquake
(M w 6.8), 1999 Chamoli Earthquake (M w 6.4), 2005 Kashmir Earthquake (M w 7.6),
2011 Sikkim Earthquake (M w 6.9), and 2015 Gorkha Earthquake (M w 7.8), etc. All
these earthquakes are known to have caused massive societal damage, including loss
of life. The earthquakes in the hilly terrain are accompanied by co-seismic landslides
[15, 26], and these co-seismic landslides further aggravate the catastrophe caused by
the earthquakes.
Earthquake-induced landslides have been a cause of concern for a long time
and have been widely studied in various regions. Among many models available
for performing the seismic landslide hazard analysis, the Newmark sliding block
displacement [23] is popular in the scientific community because of its simplicity
and ability to provide slope displacement values. Several researchers used slope
displacement prediction equations to perform regional-level co-seismic landslide
hazard assessments for various regions. Refice and Capolongo [27] performed slope
displacement-based landslides susceptible studies for the Sele Valley of southern
Italy using a pseudo-probabilistic approach. Nayek and Gade [22] developed a prob-
abilistic seismic landslide hazard map for the central seismic gap region based on
the Newmark sliding block method. These studies highlight the importance of iden-
tifying the areas prone to co-seismic landslides to minimize societal loss in case of
any such event.
The 1905 Kangra earthquake was the major earthquake that occurred in the
Himachal Pradesh state, and this event claimed around 20,000 lives [33]. Further,
this earthquake triggered massive landslides in the near-field region and created two
landslide-dammed lakes in Tirthan and Sainj gorges [19]. The Himalayan region has
not experienced a major earthquake (> M w 8) since 1952 [11], hence, the region
is long overdue for such an event. Gupta and Sabnis [12] estimated that in case of
a hypothetical midnight M w 8 earthquake, with Sundernagar in District Mandi as
the epicenter, around 1 million people will lose their lives across Himachal Pradesh,
Punjab, Haryana, and UT of Chandigarh. Chand and Sharma [5] prepared a landslide
hazard vulnerability map of district Mandi, highlighting the prone areas. However,
within the authors’ knowledge, the co-seismic landslide zonation of Mandi town is
not performed. Mandi town lies in the vicinity of the Main boundary thrust [31]
and is densely populated, as shown in Fig. 1 (Google Earth, n.d.). The Manali–
Ropar strike-slip fault also passes through Sundernagar–Mandi [32]. These faults
in its vicinity indicate the town’s vulnerability to such an event. The vulnerability
of the town to both earthquakes and landslides makes it imperative for a study on
earthquake-induced landslides in the town.
The study region extends from 31° 41, 16.71,, N to 31° 42, 14.59,, N latitude
and 76° 54, 31.20,, E to 76° 58, 12.47,, E longitude. The Google Earth, n.d. image
Seismic Landslide Hazard Assessment of Mandi Town, Himachal Pradesh 259
Fig. 1 Google Earth image of Mandi town, highlighting the study area
(Fig. 1) shows the present-day town in red. It is evident from Fig. 1 that the present-
day town has been swamped with urban construction. The town is likely to expand
in the future into the surrounding area. Keeping the future expansion of the town, a
1 km wide buffer area, shown in blue (Fig. 1), is selected for the present study. The
town is the district headquarter of District Mandi and houses the office of the district
collectorate, district hospital, and office of the district disaster management authority.
It serves as a vital road link to the tourist towns of Kullu and Manali. Given that the
city is located in the seismically active Himalayan region and has the potential to
experience a major earthquake in the near future, it is important to perform the co-
seismic landslide hazard analysis to identify the vulnerable regions. Thus, an attempt
is made in this work to develop earthquake-induced landslide hazard maps for the
Mandi town of Himachal Pradesh. The present results will help the administration
identify the town’s at-risk locations and devise a plan to minimize the seismic risk
of any such event.
2 Methodology
During seismic loading, as the shear strength mobilizes, permanent displacement

builds up due to plastic strain on the sliding surface [29]. The Newmark sliding
block approach [23] has been widely used to estimate the sliding displacement in
seismic landslide hazard assessment over the pseudo-static approach and the stress
deformation analysis. This method assumes a rigid block to slide down an inclined
plane when its base acceleration exceeds the critical acceleration (k y ). The block
continues to slide until its relative velocity with respect to the inclined plane becomes
zero. Now, to estimate the sliding displacement of the block, the relative velocity time
history is integrated. However, the Newmark sliding block model cannot be used in its
original form for co-seismic hazard mapping as we are predicting displacements due
to ground motions that might occur in the future. Therefore, various predictive models
based on the Newmark sliding block approach have been developed to estimate the
sliding displacement [2, 8, 13, 14]. Several researchers have used these predictive
models and prepared landslide-susceptible maps, in terms of the probability of slope
displacement exceeding a threshold value (mostly 5 cm), for different regions of
the world [27, 28]. The present study uses the slope prediction equation proposed
by Jibson, [13], which was developed using 2270 ground motions data from 30
earthquakes.

k y 2.341 k y −1.438
log D N = 0.215 + log 1 − ± 0.510 (1)
PGA PGA
DN = Newmark sliding displacement (cm); k y = critical acceleration of the slope

(g); PGA = Peak ground acceleration (g).
Further, the above equation is used to estimate the slope displacement, and the
probability of the displacement exceeding a threshold value of 5 cm is calculated
at each grid point. The estimation of k y and PGA values at each grid point of the
selected region is discussed below.
2.1 Estimation of Critical Acceleration (ky )
The critical acceleration depends on slope angle (β), saturation level of the ground
(m), and material properties such as cohesion (c), unit weight of soil (γ ), angle of
internal friction (ϕ), sliding mass thickness (t) enumerated in Eqs. 2 and 3.
c tan ∅ mγw tan ∅

FS = + − (2)
γ t sin β tan β γ tan β
k y = (FS − 1)g sin β (3)

First, we have estimated the slope angle at each grid point using a digital elevation
map (DEM) of 12 m resolution obtained from ALOS PALSAR—ASF, n.d. The slope
angle distribution (β) of the study is estimated from the DEM map using the ArcMap
software, and the corresponding results are shown in Fig. 2. It is evident from Fig. 2
that the present-day town has extended up to the area covered by gentler slopes. Future
expansion toward the steep slopes is inevitable. Next, we require the material prop-
erties (c, φ, t, and γ ) of the study region. The area around Mandi town consists of the
Shali and Sundernagar formations of weathered rocks [6], granite and mandi granite,
phyllite/schists, and basalt formations [16]. Mali et al. [17] have listed the presence of
the Shiwalik group, Dharamshala/Kasauli formation, and Shali/Sundernagar/Kullu
formation Jutugh group consisting of siltstones, sandstones, shale, phyllites, slate,
quartzite, gneisses, etc. in the Mandi area. The above-mentioned is the geology of
the neighboring areas of Mandi town. The Geological Survey of India (GSI) digital
lithological map for the study area is used in the present study. The area can be
classified into two groups. In Group I, we club together medium rocks (Calcareous
sandstone, Argillaceous limestones, and Marls), and in Group II, we club hard rocks
(Slate, Quartzites, and Gneiss), as shown in Fig. 3. These classifications are broad
and may not include the spatial variation in the study area.
Mali et al. [17] have carried out laboratory experiments in the vicinity of Mandi
town toward IIT Mandi and found the dry unit weight to range from 15.5 to 17.30
Fig. 2 Slope angle distribution of the study area

Fig. 3 Lithological groups of the study area (Group I represents medium rocks such as calcareous
sandstone, argillaceous limestones, and marls, and Group II represents hard rocks such as slate,
quartzites, and gneiss)
kN/m3 , cohesion value to range from 5 to 45 kPa, and angle of internal friction to
range from 9.5 to 24°. The geotechnical investigation carried out by Ghosh et al.
[9] along the Rishikesh–Uttarkashi–Gangotri National Highway in the vicinity of
Agrakhal village produced a proctor density of 1.74 g/cc, cohesion varied from 0.02
to 0.11 kg/cm2 and friction angle as 40°–46°. Sharma et al. [30] carried out a geotech-
nical investigation of Kotropi soil (near Mandi) and found the maximum dry density
as 1.69 g/cc, average cohesion 26.66 kN/m2 , and average friction angle as 32.66°.
Nayek and Gade [22] utilized the most likely values of strength parameters of the
lithological groups obtained from the literature for the landslide hazard assessment
of the central seismic gap region of the Himalayas, given in Table 1. For this study,
we considered the strength parameters given in Table 1 for our calculations.
Martha et al. [18] show that the sliding mass thickness (t) ranges from 3 to 8 m in
the Himalayan region of Nepal. Nayek and Gade [22] considered the sliding mass
thickness as 5 m for the central seismic gap region in the Himalayas. Considering
Table 1 Shear strength

Lithological Unit weight Cohesion (c) Angle of
parameters for the study area
group (γ) (kN/m3 ) (kN/m2 ) friction (φ) (°)
Group I 24 36 28
Group II 27 46 29
Fig. 4 Critical acceleration (k y ) values for the study region for a dry soil (m = 0), b saturated soil
(m = 1)
that our study region also falls in the lesser Himalayas, we considered the depth of
the sliding mass to be equal to 5 m. The estimation of critical acceleration has been
carried out for two saturation levels, arid soil (m = 0) and completely saturated soil
(m = 1), and the corresponding results are shown in Fig. 4.
In the present study, the PGA values are estimated by performing the PSHA.
Here, the PGA values are estimated for 100, 475, and 2475 return periods using
the recurrence parameters reported by Muthuganeisan and Raghukanth [20] and
the GMPE proposed by Dhanya and Raghukanth [7]. Further, the site condition is
assumed as C-type [21]. From PSHA studies, we have observed that the PGA values
estimated from PSHA are almost constant at each grid point. It can be attributed to
the fact that these points are located very close to each other. The estimated PGA
values for three return periods are 0.112 g, 0.272 g, and 0.564 g, respectively.
Here, the estimated critical acceleration, PGA, along with Eq. 1, are used to prepare
the landslide susceptibility maps of the study region. Initially, the mean prediction
of slope displacement at each grid point is estimated for two saturation conditions
and PGA values corresponding to three return periods. Further, the probability of
slope displacement exceeding a threshold value is estimated by assuming lognormal
distribution with a standard deviation value of 1.175 (given in Eq. 1). In this study, the
threshold displacement causes a landslide during the earthquake which is considered
as 5 cm. The seismic landslide hazard maps are shown in Fig. 5.
It can be observed from the hazard maps that the area prone to co-seismic land-
slides falls in the buffer zone, attributing to higher slope angles in that region. The
probability gets as high as 93.4% for a return period of 2475 years for saturated soil. It
Fig. 5 Seismic landslide hazard map of Mandi town for a return period 100 years for dry soil (m
= 0), b return period 475 years for dry soil (m = 0), c return period 2475 years for dry soil (m =
0), d return period 100 years for saturated soil (m = 1), e return period 475 years for saturated soil
(m = 1), and f return period 2475 years for saturated soil (m = 1)
is a major concern as new tourism projects like Shiv Dham are coming up in the buffer
zone, and the population settlement here is likely to increase rapidly. The study also
highlights the high probability of co-seismic landslides along the existing national
highways towards Kullu, Jogindernagar, and Bilaspur. The landslides on these roads
might obstruct rescue and relief operations, evacuation, and medical aid in case of an
earthquake. The co-seismic landslides along the riverbanks of Beas and Suketi Khud,
flowing through the study region, might lead to the formation of landslide-dammed
lakes. The formation of such lakes on the upstream end of the town will increase
the possibility of flash flooding on the downstream end. In addition, the helipad lies
within the zone of high probability for co-seismic landslides, reducing the chances
of aerial assistance. The study presented in this paper can help the administration
plan the town’s expansion wisely, keeping in mind the population’s best interest and
minimizing any societal losses in case of such an event.
4 Conclusions
In this study, the co-seismic landslide hazard maps have been prepared for Mandi
town of Himachal Pradesh. The study region extends from 31° 41, 16.71,, N to
31° 42, 14.59,, N and 76° 54, 31.20,, E to 76° 58, 12.47,, E. The Newmark sliding
displacement-based hazard analysis is adopted in the present study. At each grid
point of the study region, the probability of slope displacement exceeding 5 cm is
estimated for two saturation (dry and fully saturated) and PGA values corresponding
to three return periods. The probability of co-seismic landslides is low in the existing
city region. However, the probability gets as high as 93.4% for saturated soil and
86.3% for arid soil for a return period of 2475 years in the buffer region. These
observations can be attributed to the steep slopes in the buffer region. The present
study intends to help the administration in the detailed planning of the development
projects. One can explore effective mitigation techniques to minimize the vulner-
ability of the infrastructure located in the buffer zones. Further, we would like to
add that the buffer zone microzonation studies using the finer resolution DEM and
geotechnical material characterization will be the future scope of the present work.
Funding This research is supported by Indian Institute of Remote Sensing (IIRS), ISRO,
Government of India, under the grant no. IIRS/DO/DMSP-ASCB/2022/14.
References
1. ALOS PALSAR—ASF: Retrieved July 28, 2022, from https://asf.alaska.edu/data-sets/sar-data-

sets/alos-palsar/ (n.d.)
2. Ambraseys, N.N., Menu, J.M.: Earthquake-induced ground displacements. Earthquake Eng.
Struct. Dynam. 16(7), 985–1006 (1988). https://doi.org/10.1002/eqe.4290160704
3. Ambraseys, N., Bilham, R.: A note on the Kangra Ms = 7.8 earthquake of 4 April 1905. Curr.
Sci 79(1), 45–50 (2000)
4. Bilham, R.: Himalayan earthquakes: a review of historical seismicity and early 21st century
slip potential. Geol. Soc. 483(1), 423–482 (2019). https://doi.org/10.1144/SP483.16
5. Chand, D.K., Sharma, D.D.: Spatial pattern of landslide vulnerability at block level in district
Mandi of Himachal Pradesh: a GIS based approach. J. Clim. Change Water 2(1), 17 (2017)
6. Choubey, V.M., Mukherjee, P.K., Bajwa, B.S., Walia, V.: Geological and tectonic influence
on water–soil–radon relationship in Mandi-Manali area, Himachal Himalaya. Environ. Geol.
52(6), 1163–1171 (2007). https://doi.org/10.1007/s00254-006-0553-1
7. Dhanya, J., Raghukanth, S.T.G.: Ground motion prediction model using artificial neural
network. Pure Appl. Geophys. 175(3), 1035–1064 (2018). https://doi.org/10.1007/s00024-017-
1751-3
8. Gade, M., Nayek, P.S., Dhanya, J.: A new neural network–based prediction model for
Newmark’s sliding displacements. Bull. Eng. Geol. Env. 80(1), 385–397 (2021). https://doi.
org/10.1007/s10064-020-01923-7
9. Ghosh, A., Sarkar, S., Kanungo, D., Jain, S., Kumar, D., Kalura, A., Kumar, S.: Slope instability
and risk assessment of an unstable slope at Agrakhal, Uttarakhand. In: Proceedings of the India
Geotechnical Conference, Guntur, India (2009)
10. Google Earth: Retrieved July 28, 2022, from https://earth.google.com/web/ (n.d.)
11. Gupta, H.K., Gahalaut, V.K.: Can an earthquake of Mw ∼9 occur in the Himalayan region?
Geol. Soc. 412(1), 43–53 (2015). https://doi.org/10.1144/SP412.10
12. Gupta, H.K., Sabnis, K.A.: Developing an earthquake resilient society in the vicinity of
Himalaya. J. Geol. Soc. India 97(12), 1593–1602 (2021). https://doi.org/10.1007/s12594-021-
1918-5
13. Jibson, R.W.: Regression models for estimating coseismic landslide displacement. Eng. Geol.
91(2–4), 209–218 (2007). https://doi.org/10.1016/j.enggeo.2007.01.013
14. Jibson, R.W.: Predicting earthquake-induced landslide displacements using Newmark’s sliding
block analysis. Transp. Res. Record 1411 (1993). https://trid.trb.org/view/384547
15. Keefer, D.K.: Investigating landslides caused by earthquakes—a historical review 38 (2002).
https://doi.org/10.1023/A:1021274710840
16. Kumar, A., Sharma, R.K., Bansal, V.K.: GIS-based landslide hazard mapping along NH-3 in
mountainous terrain of Himachal Pradesh, India using weighted overlay analysis. In: Singh,
H., Garg, P., Kaur, I. (eds.) Proceedings of the 1st International Conference on Sustainable
Waste Management Through Design, vol. 21 (pp. 59–67). Springer International Publishing
(2019). https://doi.org/10.1007/978-3-030-02707-0_9
17. Mali, N., Shukla, D.P., Kala, V.U.: Identifying geotechnical characteristics for landslide hazard
indication: a case study in Mandi, Himachal Pradesh, India. Arab. J. Geosci. 15(2), 144 (2022).
https://doi.org/10.1007/s12517-022-09475-8
18. Martha, T.R., Roy, P., Mazumdar, R., Govindharaj, K.B., Kumar, K.V.: Spatial characteristics
of landslides triggered by the 2015 Mw 7.8 (Gorkha) and Mw 7.3 (Dolakha) earthquakes in
Nepal. Landslides 14(2), 697–704 (2017). https://doi.org/10.1007/s10346-016-0763-x
19. Middlemiss, C.S.: The Kangra Earthquake of 4th April, 1905, vol. 38. Geological survey of
India (1910)
20. Muthuganeisan, P., Raghukanth, S.T.G.: Site-specific probabilistic seismic hazard map of
Himachal Pradesh, India Part II. Hazard estimation. Acta Geophys. 64(4), 853–884 (2016)
21. Muthuganeisan, P., Raghukanth, S.T.G.: Site-specific probabilistic seismic hazard map of
Himachal Pradesh, India. Part I. Site-specific ground motion relations. Acta Geophysica 64(2),
336–361 (2016a)
22. Nayek, P.S., Gade, M.: Seismic landslide hazard assessment of central seismic gap region of
Himalaya for a Mw 8.5 scenario event. Acta Geophysica 69(3), 747–759 (2021). https://doi.
org/10.1007/s11600-021-00572-y
23. Newmark, N.M.: Effects of earthquakes on dams and embankments. Géotechnique 15(2),
139–160 (1965). https://doi.org/10.1680/geot.1965.15.2.139
24. Oldham, R.D.: Report of the Great Earthquake of 12th June, 1897. Office of the Geological
Survey (1899)
25. PH and HP State Unit, Chandigarh: Report on preliminary assessment of the landslide, team
of GSI office. https://employee.gsi.gov.in/cs/groups/public/documents/document/b3zp/mtyx/
*edisp/dcport1gsigovi161798.pdf. Accessed 1 Nov 2017
26. Parker, R.N., Densmore, A.L., Rosser, N.J., de Michele, M., Li, Y., Huang, R., Whadcoat, S.,
Petley, D.N.: Mass wasting triggered by the 2008 Wenchuan earthquake is greater than orogenic
growth. Nat. Geosci. 4(7), 449–452 (2011). https://doi.org/10.1038/ngeo1154
27. Refice, A., Capolongo, D.: Probabilistic modeling of uncertainties in earthquake-induced land-
slide hazard assessment. Comput. Geosci. 28(6), 735–749 (2002). https://doi.org/10.1016/
S0098-3004(01)00104-2
28. Rodriguez-Peces, M.J., Garcia-Mayordomo, J., Azanon, J.M., Jabaloy, A.: Regional hazard
assessment of earthquake-triggered slope instabilities considering site effects and seismic
scenarios in Lorca Basin (Spain). Environ. Eng. Geosci. 17(2), 183–196 (2011). https://doi.
org/10.2113/gseegeosci.17.2.183
29. Rollo, F., Rampello, S.: Probabilistic assessment of seismic-induced slope displacements: an
application in Italy. Bull. Earthq. Eng. 19(11), 4261–4288 (2021). https://doi.org/10.1007/s10
518-021-01138-5
30. Sharma, P., Rawat, S., Gupta, A.K.: Study and remedy of Kotropi landslide in Himachal
Pradesh, India. Ind. Geotechn. J. 49(6), 603–619 (2019). https://doi.org/10.1007/s40098-018-
0343-1
31. Singh, P., Ao, A., Thakur, S., Rana, S., Sharma, R., Krishnakanta Singh, A., Singhal, S.:
Geology, Structural, Metamorphic and Mineralization Studies Along the Mandi-Kullu-Manali-
Rohtang Section of Himachal Pradesh, NW-India (pp. 437–460) (2021).
32. Thakur, V.C., Jayangondaperumal, R., Joevivek, V.: Seismotectonics of central and NW
Himalaya: plate boundary–wedge thrust earthquakes in thin- and thick-skinned tectonic
framework. Geol. Soc. 481(1), 41–63 (2019). https://doi.org/10.1144/SP481.8
33. Thakur, V.C., Sriram, V., Mundepi, A.K.: Seismotectonics of the great 1905 Kangra earthquake
meizoseismal region in Kangra-Chamba, NW Himalaya. Tectonophysics 326(3–4), 289–298
(2000). https://doi.org/10.1016/S0040-1951(00)00126-8
Infill Wall Effect on Seismic Analysis
of Reinforced Concrete Buildings
C. H. Sirajudheen and Behera Dibyadarshi
Abstract Infill walls are used to fill the gap between the structural elements in a
framed structure. It protects the structure from external environment and also acts
as partition wall to create separate rooms according to the requirements. Different
types of infills are used to create masonry wall like red clay brick, fly ash brick,
and AAC blocks. In seismic analysis, the infill wall contribution is ignored as it is
considered as non-structural element. But in recent years, much research has been
conducted by many researchers to study the behavior of the unreinforced masonry
during seismic event. Here, the study is to find the infill wall effect on seismic
analysis and design. In our study, bare frame model (BFM), hybrid frame model
(HFM) and infilled frame model (IFM) are taken to perform the seismic analysis.
Equivalent diagonal strut method is used to include the unreinforced masonry infill
wall effect in seismic analysis using codal provisions of IS1893:2016. In this method,
the wall is converted as a diagonal strut which only takes compressive force when
the structure is subjected to some lateral forces such as seismic load. We performed
few experiments on red clay brick masonry to obtain the data required for designing
the diagonal strut dimension. Both static and dynamic analyses (response spectrum
method) are performed using ETABS software. G + 4, G + 9, and G + 14 models
are designed to study the infill wall contribution in earthquake analysis by varying
the elevation of the structure. Storey drift, storey displacement, member forces, and
time period are taken as the parameters. After performing both the analysis, this is
observed that the infill wall significantly affect the member forces and stiffness of
the structure during seismic event. So, the practicing engineers must include infill
wall effect while designing real structures.
Keywords Diagonal strut method · Hybrid frame · Infilled frame · Bare frame ·
Storey drift · Storey displacement · Response spectrum method
C. H. Sirajudheen
Assistant Professor, Department of Civil Engineering, College of Engineering and Technology,
Bhubaneswar, India
B. Dibyadarshi (B)
College of Engineering and Technology, Bhubaneswar, India
e-mail: dibyadarshibehera1996@gmail.com
https://doi.org/10.1007/978-981-99-1459-3_22
270 C. H. Sirajudheen and B. Dibyadarshi
1 Introduction
The infill wall is the unreinforced masonry (URM) panel used to fill the reinforced
concrete frames which accomplish the necessity of making separate rooms and
protect the buildings from the environment. There are different types of infills avail-
able to create wall, viz. AAC blocks, red clay brick, fly ash brick. Red clay bricks
are the oldest and most extensively used as building materials worldwide. Still, it is
the most common and ruling construction material because these are cheap, durable,
and easy to handle and to work with. Generally, infill walls are constructed without
leaving any gap between structural elements. But nowadays, gaps are given in infill
walls to build large windows for the aesthetical view. These soft storey shows rela-
tively less stiffness than other floors. Also, open-ground storey buildings are built
to provide parking area in the ground floor. From the past earthquake instances,
it is found that the irregularities in elevation due to the infill wall caused severe
damage to the building. Like in Bhuj earthquake (2001), the collapse of more than
a hundred RC frame buildings with open ground stories at Ahmedabad has empha-
sized that such buildings are extremely vulnerable under earthquake shaking [1]. So,
the contribution of infills on the overall capacity of structures is strongly dependent
on regularity of their distribution in plan and over the height [2]. Due to the lack
of knowledge of the complex behavior of the infilled frame, it is neglected in the
design. However, under seismic loading infill walls tend to interact with the frame
when the structure is subjected to lateral loads and also exhibit energy-dissipation
characteristics. Masonry walls contribute to the stiffness of the infill under the action
of lateral load [3]. The mass distribution and height of the building are the parameters
which contributes in fundamental time period. Therefore, dynamic forces like earth-
quake depend on dynamic characteristics like mass and stiffness of the structure. As
per research, regular distribution of infill wall reduces the fundamental time period
[4]. Displacement profiles for both equivalent static method (ESM) and response
spectrum method (RSM) have a sudden change of slope at first storey level. The
inter-storey drift demand is largest in the ground storey for all the models for both
ESM and RSM. The mode shape changes significantly when infill is present in the
building. Thus, the infill wall significantly increases the total base shear in both static
and dynamic analyses [5]. Infilling panels are found to increase stiffness of the struc-
ture. The increase in initial stiffness, obtained for small strains, can reach seven times
that of bare frame. Experimental data show that after the first shear cracking, appre-
ciable stiffness degradation occurs. This decrease continues when the displacement
increases. However, the stiffness of the infill frame remains higher than that of the
bare frame even after the collapse of the masonry panel [6]. Presently in India, the
URM wall is considered as non-structural element even though masonry infills give
strength and stiffness to the reinforced frame. But in higher seismic zone, the infill
wall design should be taken into consideration because the presence of infill wall
significantly affects the lateral displacement of the building, bending moments, shear
force, and axial forces in the beams and columns. In revised IS Code 1893:2016, the
codal provisions are also there for infill wall [7].
Infill Wall Effect on Seismic Analysis of Reinforced Concrete Buildings 271
Infill wall behaves like compressive strut in between beam and column. From
the modeling approaches of infill walls, there are two main different methods have
been used. Micromodel based or finite element techniques while macromodel is the
equivalent strut method. The genuine reason of using the equivalent strut method
which is a part of macromodeling is very simple while the computation based on the
physical understanding of the behavior of infills [8]. The infill walls are analytically
replaced by equivalent diagonal struts which only carry compressive force. The
endpoints of the strut connected to the frame are pin-jointed to avoid the moment
from frame to infill. In this method, the infill wall is idealized as diagonal strut, and
the frame is modeled as a truss element [9].
2 Methodology
2.1 Theoretical Background
In this study, three models are taken like bare frame model, infill frame model,
and hybrid frame model. Hybrid frame model signifies the open ground storey. To
compare the infill wall effect in seismic event, G + 4, G + 9, G + 14 buildings are
modeled with and without infill walls. The equivalent model with a single strut is not
able to describe the “local” interaction between the infill panel and the surrounding
structural elements, but it provides a rational basis for estimating the lateral strength
and stiffness of the infilled frames as well as the infill diagonal-cracking load. The
width of the strut has been calculated by equivalent strut method referring IS code
1893:2016.
For URM infill walls without any opening, width wds of equivalent strut (Fig. 1)
shall be taken as
wds = 0.175αh−0.4 L ds
where
/
E m t sin 2θ
αh = h 4
4E f Ic h
The modulus of elasticity E m (in MPa) of masonry infill wall shall be taken as:
E m = 550 f m
where f m is the compressive strength of masonry prism (in MPa) obtained as per IS
1905 or given by expression:
f m = 0.433 f b0.64 f mo
0.36
Fig. 1 Equivalent diagonal strut of URM infill wall
where f b = the compressive strength of brick in MPa; and f mo = the compressive

strength of mortar in MPa.
The experiments are conducted to in the laboratory (Fig. 2) to get the compressive
strength of red clay brick and mortar as per IS 3495-1992 and 2250-1981, respec-
tively. The linear static and dynamic analyses were performed using ETABS. This
is an engineering software product that serves to multi-storey building analysis and
design, modeling tools and templates, analysis methods, solution techniques, and
code-based load prescriptions. In dynamic analysis, the response spectrum method
has been adopted in which n modes are taken into consideration to get the maximum
response. In ETABS, 12 modes have been taken to get the results. The modes used in
the analysis for earthquake vibration along a specific direction has maintained that
sum total of modal masses of these modes should be at least 90% of the total seismic
mass. The peak responses of different modes have been combined by complete
quadratic combination method (CQC) in ETABS. To get the contribution of the infill
wall in seismic action storey displacement, storey drift, natural period, base shear,
and storey stiffness are taken into consideration. The storey displacement is the total
displacement of ith storey with respect to ground, and there is a maximum permis-
sible limit prescribed in IS codes for buildings. Maximum storey displacement limit
= H/250, where H is the total height of the building according to codal provisions.
Storey drift is the relative displacement of one level above or below between the floors
above and below under the storey consideration. Codal provision says that the storey
drift in any storey due to the minimum specified design lateral force, with a partial
load factor of 0.1 shall not exceed 0.004 times the storey height. While designing
according to code in RC and masonry structures, for beams 35% is considered as it is
stiff and the rest 65% is vulnerable to cracks. Similarly, for columns 70% is stiff and
Fig. 2 Compression test on brick and mortar
the rest 30% is vulnerable to cracks. The value for columns is more because they can
undergo compression and cracking is not easy. These factors need to be considered
as this results in more flexure value and more stiffness. The moment of inertia needs
to be reduced in order to prevent the concrete outside the stirrups from coming out.
2.2 Details of the Building Models
See Table 1.
2.3 Modelling of Infill Wall
In bare frame model (BFM), infill walls were not modeled as it is considered as
a non-structural elements although the masses of infill walls are included in the
model. In such a case, the mechanism to resist the lateral loads known as frame
action mechanism in which bending moments and shear forces are developed in
beams and column by means of rigid joint action. The infill walls are modelled as
a diagonal strut in infilled frame model (IFM) and hybrid frame model (HFM). The
width of the strut was calculated using IS 1893 provisions, and then it is modeled
as a strut member in ETABS. The strut only assigned to compressive load and the
ends of the strut made pin jointed not to take any moments. The only difference
between IFM and HFM is infill walls are distributed throughout the structure in case
of IFM, whereas in HFM, infill walls are present all stories except ground storey. The
properties of the building which is mentioned in Table 1 are based on IS 1893:2016
provisions [7] (Figs. 3, 4 and 5).
274
Table 1 Properties of building

Details of building Material properties Member properties Type of loads and their Seismic properties
intensities
SMRF (special moment Grade of concrete = M 25 Grade and thickness of slab Live load on floor = 3 Zone factor (Z) = 0.36, Zone
resisting frame) = M 25 and 165 mm kN/m2 =V
Plinth level = 2 m Grade of steel = Fe 500 and Fe 415 300 mm * 500 mm (primary Live load on roof = 1.5 Importance factor (I) = 1.2
beam), 300 mm * 400 mm kN/m2
(plinth beam)
Floor to floor height = Density of concrete = 25 kN/m3 Column size 400 mm * SDL on floor = 2.5 Response reduction factor (R)
3.35 m 400 mm (G + 4), 600 mm * kN/m2 =5
600 mm (G + 9, G + 14)
Type of building = Density of masonry wall (brick) and Internal wall thickness = SDL on roof = 1.5 Soil type = I
residential building plaster = 20 kN/m3 145 mm kN/m2 Damping ratio = 5%
External wall thickness = Parapet = 6.48 kN/m
270 mm
C. H. Sirajudheen and B. Dibyadarshi
Fig. 3 Bare frame model of G + 14 building
Fig. 4 Infill frame model of G + 14 building
3.1 Experimental Results
Compressive strength of the red clay brick (24 × 11.5 × 7 cm) = 6.9 N/mm2 .
Fig. 5 Hybrid frame model of G + 14 building
28 days compressive strength of mortar cube of side 7.06 cm = 14.76 N/mm2 .

Using these values, we calculated the compressive strength of masonry prism
(f m ).
As per IS 1893,
f m = 0.433 f b0.64 f mo
0.36
= 0.433 × 6.90.64 × 14.760.36
= 3.9 ≈ 4 N/mm2
3.2 Results from Numerical Analysis
3.2.1 Natural Period
In the static analysis for BFM, natural period of the building is obtained corre-
sponding to the fundamental translational mode as per IS 1893:2016. So here we got
a higher value in BFM and same value for both HFM and IFM. Static analysis is an
approximate method, and it gives the results independent of the structural properties
of the building. However, it is mainly depending on an empirical formula which is
influenced by the total height as well as the type of structure and its plan. From the
dynamic analysis in which we consider the multiple modes and its combined effect,
we found that modeling of infill with equivalent diagonal strut significantly affects
natural period of the building. In BFM, we found a higher natural period as compared
to HFM and IFM for all three models. The captured results indicate that the time
period of open storey building model, which is considered as hybrid frame structure
is in between time period of bare frame model and infilled frame model. In all three
models, the natural period of infilled frame is reducing almost 50% compared to bare
frame models. So, from the analysis of results, it is concluded that the presence of
infill wall in the structure increases its lateral stiffness and hence reduces the time
period of the structure (Figs. 6, 7, 8). From all the above results, it is also found that if
we increase the height of the structure then structure will become more flexible and
hence the time period also increases. Hence, the height also affects the time period
in dynamic analysis.
Fig. 6 G + 4 buildings
3.2.2 Maximum Storey Displacement
From Figs. 9, 10, and 11, we observed that maximum storey displacement in all
the models is below the limit value of (H/250) where H is the total height of the
building. For G + 4, G + 9, and G + 14 models, the limit values are 75, 142,
and 209 mm, respectively. Bare frame storey displacements are much higher than
frame with infill effect in all the cases. From this, we can understand that infill wall
contributes in lateral load resisting action. If we compare the values of HFM and
IFM, the displacement of hybrid frame is bit higher than infilled frame, and it is
due to the soft storey or open ground storey effect. Displacement of soft storey is
much higher compared to frame with infill effect. From this, we can understand that
soft storey effect is much higher in small buildings compare to medium or high rise
buildings. Displacement of Storey-2 of HFM is higher than that of IFM & BFM in
all the models due to presence of soft storey in all the cases. So, by comparing that
relative storey displacement (storey drift) of storey-2 to storey-1 for G + 4 models
(Fig. 9), we found that for IFM value of storey drift is 4.144 mm which is within limit
of 0.002 times of h, i.e., 6.7 mm mentioned in Table 6 of IS 1893:2016. In the case of
HFM, value of drift is 21.851 mm which is significantly higher than the limit value
6.7 mm. For BFM, we found a drift of 14.437 mm which is greater than limiting
value 13.4 mm, i.e., 0.004 times of h. In G + 9 storey drift for IFM is 4.792 mm
which is within the limit. For HFM 10.749 greater than limiting value and for BFM
9.755 mm which is within limiting value. In G + 14, IFM’s storey drift is 4.773
within the limiting value. HFM’s value is 10.2 mm greater than the limiting value
(Fig. 11). BFM’s value is 10.41 within the limiting value. Hence, if we see storey
drift, then storey drift of HFM is higher than limit value mentioned in the code. But
Storey Displacement in mm MAX. STOREY DISPLACEMENT

100
50
0
storey 6 storey 5 storey 4 storey 3 storey 2 storey 1 Base
BFM 51.806 47.995 40.945 31.008 18.501 4.064 0
HFM 38.354 37.509 35.764 33.241 28.912 7.061 0
IFM 18.285 17.31 15.34 12.481 8.945 4.801 0
LIMIT VALUE 75 75 75 75 75 75 75
BFM HFM IFM LIMIT VALUE
maximum storey displacement is coming within the limit. But, comparing all the
models by Storey drift and storey displacement IFM shows better result. To avoid
failure of building due to drift in HFM, shear wall should be designed or the column
dimension should be increased.
3.2.3 Base Shear
From Fig. 12, we can understand that presence of infill walls in the structure leads
to an increase in base shear force in all buildings. If we compare all the models, the
base shear also increases as the increase of number of stories. Among all models, G
+ 14 building has highest value of base shear. If we observe, all the three models,
there is not much difference between HFM and IFM in terms of base shear values.
The small difference due to the infill wall is not present in ground storey of HFM.
Fig. 12 Base shear for G + 4, G + 9, and G + 14 buildings
Hence, it is proved that the presence of infill wall increases the lateral stiffness of
the building which attracts more seismic forces to the structure.
3.2.4 Storey Stiffness
From Fig. 13, we found that ground storey stiffness of IFM is much higher than
BFM and HFM. Also stiffness of HFM is higher than BFM. In G + 4 model, storey
Fig. 13 Storey stiffness of G + 4, G + 9, and G + 14 buildings
stiffness of HFM is 1.2 times that of BFM, and similarly, storey stiffness of IFM
is 5.82 times that of BFM. In G + 9 model, storey stiffness of HFM is 1.56 times
that of BFM and similarly storey stiffness of IFM is 3.57 times that of BFM. In G
+ 14 model, storey stiffness of HFM is 1.6 times that of BFM and similarly storey
stiffness of IFM is 3.5 times that of BFM. In both G + 9 and G + 14 models, soft
storey stiffness is about 55% less compared to with infill effect and for G + 4 model
soft storey stiffness is about 80% less compared to IFM. Hence, storey stiffness is
significantly increases from G + 4 to G + 9 building, but the increase from G + 9 to
G + 14 building is very small. So, stiffness not only affected by infill but also height
of the building. High-rise building will be more flexible than medium rise building.
3.2.5 Comparison of Bending Moment Diagrams
The variation of bending moment is shown in Fig. 14. The value of bending moment
in BFM is higher than HFM and IFM for all stories. From the analysis results of G +
4, G + 9, and G + 14 buildings, we found that in G + 4 models, GF column bending
moment demand in infilled frame model (IFM) is about 72–76% less when compared
to BFM and HFM. In G + 9 models, GF column bending moment demand in infilled
frame model (IFM) is about 78.3–88% less when compared to BFM and HFM. In G
+ 14 models, GF column bending moment demand in infilled frame model (IFM) is
about 73–80% less when compared to BFM–HFM. It is also found that the interior
BFM (G+14) IFM (G+14) HFM (G+14)
Fig. 14 Variation of bending moment in BFM, IFM, and HFM
column has the higher value of bending moment than corner and exterior columns
due to seismic loads.
3.2.6 Comparison of Axial Force Diagrams
The columns are experiencing axial force in all the models shown in Fig. 15. But the
axial force in columns of IFM is higher than other models due to truss action. From
the analysis results of G + 4, G + 9, and G + 14 buildings, we found that, in G +
4 building, the axial force of IFM is 173.83% higher than BFM and axial force of
HFM is 110.7328% higher than BFM. In G + 9, IFM is 152% and HFM is 128%
higher than BFM while comparing to axial force of corner column. Similarly in G
+ 14, IFM is 104% higher than BFM, and HFM is 93.81% higher than BFM.
4 Conclusion
By comparing bending moment and axial force diagram, we can conclude that in bare
frame models (BFMs), the bending moment and shear force demand are much higher
in all the stories compared to infilled frame and hybrid frame models. Axial forces
are also present but less than that of infilled frames. Hence, it is a clear indication
of predominant frame action in BFM. In infilled frame models (IFM), the bending
moment and shear force demand are very less in comparison with bare frame and
soft storey HFM. At the same time, the axial forces are much higher than BFM and
HFM. Hence, it is a clear indication of predominant truss action in IFM. In hybrid
frame models (HFM), the bending moment and shear force demand are significant
BFM (G+14) IFM (G+14) HFM (G+14)
Fig. 15 Variation of axial force in BFM, IFM, and HFM
only in ground storey, and all other upper stories are having much lesser moments
and shear forces compared to BFM. But the axial forces are present in all columns
which is significantly higher than BFM. Hence, it is a clear indication of predominant
frame action in ground storey and truss action in all other upper stories. So, we can
call it as a hybrid action in HFM.
It is also analyzed that the presence of infills in building decreases the time period
by which the building will be less affect by the earthquake. From maximum storey
displacement values, it is found that the bare frame model has a higher value of
displacement than the infilled frame and hybrid frame models. But in the hybrid
frame model, maximum displacement is found in the soft storey as compared to
other stories. Hence, if any irregularities of infill wall present in any storey then on
that level damage will occur. As we observed, the storey displacement varied from
62 to 22% if we consider the infill wall. In all the cases, the relative displacement is
coming within limit but by considering infill wall the drift value decreases. But at
the soft storey drift value is higher all than other floors, as it attracts more earthquake
forces. Seismic analysis of the bare frame model leads to underestimations of base
shear. This underestimation of base shear may lead to the collapse of the structure
during earthquake shaking. As buildings are more capable of taking compressive
load than shear force and bending moment, so infilled frame model is effective in
seismic events. The stiffness of the infilled frame is higher than the others. The
higher the stiffness lesser will be the deflection. The presence of infill reduces the
lateral deflection of the building, displacement, bending moments in the frame, and
increasing axial forces in columns. Hence, there is a need to include the infill wall
effect in the seismic analysis of reinforced concrete buildings to predict the correct
behavior of buildings during an earthquake. So, it is concluded that we should analyse
at least three models—BFM, HFM, and IFM for the design of a reinforced concrete
buildings.
References
1. Murty, C.V.R.: Earthquake Tips. National Information Center of Earthquake Engineering. Indian
Institute of Technology Kanpur (2005)
2. Trapani, F.D., Macaluso, G.: Masonry infills and RC frames interaction: literature overview and
state of the art of macro modelling approach. Euro. J. Environ. Civ. Eng. (2015)
3. Davis, R., Krishnani, P.: Effect of infill stiffness on seismic performance of multi-storey RC
framed buildings in India. In: 13th World Conference on Earthquake Engineering Vancouver,
B.C., Canada (2004)
4. Kose, M.M., et al.: Parameters affecting the fundamental period of RC buildings with infill walls.
Eng. Struct. (2009)
5. Alam, T., Amanat, K.M.: Seismic response of randomly infilled reinforced concrete frames with
soft ground storey. Austr. J. Civ. Eng. (2020)
6. Elouali, T.: Effect of infill masonry panels on the seismic response of frame buildings. In: 14th
World Conference on Earthquake Engineering (14WCEE) (2008)
7. IS 1893:2016: Criteria for Earthquake resistant design of structure
8. Amalia, A.R., Iranat, D.: Comparative study on diagonal equivalent methods of masonry infill
panel. AIP Conf. Proc. (2017)
9. Abd-Elhamed, A., Mahmoud, S.: Linear and nonlinear dynamic analysis of masonry infill RC
framed buildings. Civ. Eng. J. (2017)
Geotechnical Seismic Base Isolation
Using Rubber Sand Mixtures—Review
S. L. Divyasree , K. M. Jithin , and Renjitha Mary Varghese
Abstract Automobile industry has achieved an exponential increase in the sales

of vehicles during the past decades. Consequently, the problem with the disposal of
vehicle tires has become a potential threat especially among the developing countries.
Geotechnical research community has resolved the problem by employing the tire
shreds for vibration isolation applications, retaining wall construction, as a sound
barrier, cushioning foams, etc. Current work discussed the previous studies related
to the historical background, engineering properties, mechanical behavior, dynamic
characteristics, and numerical modeling of rubber–sand mixtures. This review paper
will provide the researchers a general insight toward the factors influencing the
static and dynamic properties of rubber–sand mixtures, advantages, disadvantages,
contradictory results, and the areas which require future studies.
Keywords Automobile industry · Tire shreds · Vibration isolation · Dynamic

properties · Rubber sand mixtures · Contradictory
1 Introduction
Waste product disposal is a serious threat faced by the modern world due to industri-
alization. According to the reports of World Health Organization during 2015, esti-
mated percentage of motorized vehicles in developing countries is 53% and that in
developed countries is 46% [7]. Tire industry currently manufactures almost close to
1.6 billion tires globally. The outbreak of COVID-19 has led to tremendous increase
in the sales of vehicles [7]. After the end of life of vehicles, the disposal of the
waste tire becomes a serious concern especially for the middle-income countries.
Possible disposal methods of vehicle tires practiced by the developing countries are
stockpiling and dumping in landfill. Severe environmental as well as health risks are
associated with these disposal methods. Sometimes, the stock piles of rubber might
S. L. Divyasree (B) · K. M. Jithin · R. M. Varghese

NIT Calicut, Calicut, India
e-mail: divyasreesldh@gmail.com
https://doi.org/10.1007/978-981-99-1459-3_23
286 S. L. Divyasree et al.
cause fire outbreak and lead to the emission of toxic gases. Water accumulated in the
scrap tires become breeding places for mosquitos. Potential hazards due to improper
disposal of end-of-life tires have resulted in the exploration of alternate means to
utilize waste tires.
Scrap tires have the advantage of greater sorption capacity for volatile organic
compounds, and it is found to be useful in solid waste landfill as leachate drainage
layer [9]. End-of-life tires can be used for the production of tire derived fuel (TDF)
and tire pyrolysis oil (TPO). Every year, 100 million waste tires are being recycled.
Retreading and regrooving are the current methods exercised in India for the recycling
and reuse of waste tires. Tread, bead, and side walls are the secondary materials
obtained due to the shredding of whole tires. Further grinding of the shredded parts
may lead to a more valuable product known as rubber crumb. Crumb rubber or rubber
crumb finds applications as a fine aggregate for concrete manufacturing, raw material
for the production of floor mats, rubber sheets, hose pipes [7], surfacing of play
grounds and for making sports field. End-of-life tires have also found to be suitable
for applications such as lightweight backfill for embankments, retaining walls and
slopes in order to counteract bearing capacity failures, slope-stability problems, and
excessive settlement and to reduce the lateral earth pressure [1].
Seismic isolation using rubber–sand mixtures has been a topic of interest for the
researchers in Geotechnical Earthquake Engineering for the past few years. Large
numbers of numerical as well as experimental investigations have been done in order
to identify the seismic isolation characteristics of rubber–sand mixtures. Rubber–
sand mixtures are characterized by their high compressibility, large void ratio, high
damping, low shear modulus, low bulk unit weight, high friction angle, low liquefac-
tion potential, etc. Studies conducted by [4, 30] have also proved that scrap rubber
tire is a non- hazardous material which do not produce any kind of contamination
to ground water. The problem with self-heating can be avoided by mixing it with
soil. Current work discusses the historical background, mechanical, and engineering
properties of RSM, static and dynamic properties of RSM, and the numerical studies
related to the modeling of RSM.
2 Historical Background
Yegian and Catan [34] performed experiments on several geotextiles and suggested
that a polyethylene geotextile layer placed immediately below the footing can act
as an ideal geotechnical seismic isolation interface. These geotextile liners dissipate
the energy of the incoming seismic waves by slip deformations. Xu and Fatahi [33]
further investigated the efficacy of using geotextile layers with in the soil below the
foundation and were found to be quite satisfactory for seismic isolation because of
its high slip resistance nature. Tsang et al. [31] reported that compacted sand and
pebble layers can also serve the purpose of energy dissipation below the foundation.
Increasing demand for the application of vibration isolation technology to common
Geotechnical Seismic Base Isolation Using Rubber Sand … 287
residential buildings in developing countries led to the development of many alterna-

tive methods. Among them seismic isolation using rubber–sand mixtures proposed
by Tsang H. H gained much popularity due to the enormous advantages offered by the
end-of-life tires. Shivaprakash and Dinesh [29] introduced the concept of vibration
isolation using scrap rubber soil mixtures. Effectiveness of the suggested method
was checked by means of finite element analysis. Use of RSM can substantially
reduce the horizontal ground accelerations by 60–70% and vertical accelerations by
80–90%. Proposed method can be categorized under distributed seismic isolation
system since the rubber–sand mixtures isolate the entire surface of the foundation
structure. Advantages of rubber include its fire resistance, termite-proofing, dura-
bility, no significant outgas emission on burial, etc. Stockpiles of scrap tires can be
avoided by effectively using these scrap tires for engineering applications.
3 Mechanical and Engineering Properties of Sand-Tire

Chip Mixtures
Anbazhagan et al. [3] studied about the influence of size of rubber, rubber content,
gradation of sand on the shear, and volume behavior of sand–rubber mixtures (SRM).
Major factors which influence the shear strength of SRM are size of rubber and rubber
content. SRM with 30% Group VI granulated rubber (passing through 12.5 mm and
retained on 9.5 mm) possessed maximum shear strength with greater cohesion and
angle of friction values. With increase in rubber content, shear strength of SRM
increased for all sizes of rubber and their percentage content. Senthen Amuthan et al.
[28] determined the influence of fines content, plasticity, and relative density on shear
modulus and damping ratio of sand-fine mixtures at 0.005 to greater than 1% strain
levels. For low shear strains, shear modulus increased with increase in confining pres-
sure. At constant relative density, shear modulus increased with increase in confining
pressure. For larger shear strain amplitudes, shear modulus was found to be indepen-
dent of relative density. At low shear strains, shear modulus decreased with increase
in fines due to the increase in plasticity. But damping ratio increased with increase
in the percentage of fines. Effect of plasticity index on shear modulus was insignifi-
cant at larger strains. Also damping properties of samples increased with increase in
plasticity index. Granulated rubber is shown in Fig. 1.
Boominathan and Banerjee [5] have carried out a work regarding the compress-
ibility and permeability characteristics of sand–rubber tire shred mixtures. Sand–
rubber tire shred mixtures exhibit ductile behavior under shearing. Up to 30% rubber
content, peak angle of internal friction reduced. Beyond which its effect was insignif-
icant. Compressibility characteristics increased with increase in rubber tire shreds,
whereas inverse effect was observed for brittleness index. Permeability reduced with
increase in rubber content. Similar results were obtained for [13]. But contradic-
tory results were obtained for [22]. Rubber is a hydrophobic material. Resistance to
the flow of water along the paths with in the rubber matrix was high as compared
Fig. 1 Granulated rubber

[19]
to the flow resistance offered in sand matrix. That may contribute to the reduced
permeability.
Ghazavi [12] have determined the efficiency of using sand-tire crumb mix as infill
materials in geocell pockets of coir and HDPE geotextiles. They observed that soil
reinforced with HDPE geocell increased the bearing capacity of soil up to three
times. When the soil was reinforced with coir geotextiles, failure was gradual unlike
HDPE geocell reinforced soil. For HDPE geocell reinforcement, tire-crumb was
found to be a better infill material. For coir geocell, sand was found to be effective
for initial loads. But for higher loads and high strain values, sand-tire crumb infill
materials performed better. Also surface heave was considerably reduced when sand-
tire crumb infill materials were used with coir geotextiles. Foose et al. [10] explored
the possibility of using shredded waste tires to improve the shear strength of soil.
Their findings suggest that factors like normal stress, sand matrix unit weight, and
the amount of shreds in the mixture affected the shear strength of soil. Non-linear
strength envelope was obtained for sand having higher unit weight, whereas linear
strength envelope was obtained for loose and medium dense sand. This result was
in contrast to the result obtained by [11] but it is in agreement with the observations
reported by [18]. That may be due to the longer size of the reinforcing elements used
in the study conducted by [11].
4 Static and Dynamic Properties of Rubber–Sand Mixtures
Pincus et al. [23] have determined the dynamic characteristics of various combina-
tions of sand-granulated rubber mixtures and gravel-granulated rubber mixtures. For
constant rubber content and D 50,r
D50,s
greater than 1, as the confining pressure increased,
more rubber like behavior was observed. But less rubber like behavior was observed
for DD50,r
50,s
less than 1 and for same rubber content. Influence of void ratio on the
damping behavior of sand-granulated rubber mixtures or gravel-granulated rubber
mixtures was ascertained to be negligible. For low shear strain values less than
10–2 %, damping was mainly due to the friction between the sand grains. So damping
ratio was observed to be higher than the parent soil. For intermediate strain 10–2 –
0.5%, damping ratio was less than that of unmixed soil. For high shear strain, i.e.,
greater than 0.5%, damping ratio of the mixtures was higher than the parent soil.
At higher shear strain of 0.8%, greater damping ratio was obtained for mixtures
having DD50,r
50,s
greater than 1 and with rubber content less than 20%. For mixtures
having DD50,r
50,s
greater than 1, increase in damping was found to be less prominent. Li
et al. [14] conducted a series of consolidated undrained cyclic triaxial tests to under-
stand the dynamic behavior of rubber–sand mixtures subjected to large shear strain.
At constant rubber volume content and relative density, as the rubber particle size
reduced, shear modulus decreased. Reason can be attributed to the increased number
of contacts between the rubber particles due to the addition of small sized rubber
particles. Thus, it may increase the number of sand–rubber–sand and rubber–rubber
force transmission chains. At small shear strain amplitude, high deformation ability
of the rubber particles will predominate in the process of damping and subsequent
energy dissipation. Hence, sand–rubber mixtures under shear strain levels less than
critical strain exhibit higher damping ratio. Whereas at high shear strain levels, inter
granular friction between the particles causes dissipation of energy. Thus, damping
ratio reduces. Also, the damping ratio is inversely proportional to rubber particle size
and confining pressure. Under large shear strain, sand–rubber mixtures exhibit strain
hardening behavior.
Anastasiadis [2] have carried out torsional resonant column tests on pure sand,
gravel, sand–rubber mixtures, and gravel rubber mixtures under confining pressures
in the range of 25–400 kPa. As the percentage of soft rubber inclusions increased,
void ratio and dry unit weight of the mixture decreased. Dynamic response of the soil-
rubber mixtures is largely influenced by three factors, mean grain size of rubber to
soil grains, rubber fraction, and uniformity co-efficient. As the rubber fraction in the
mixtures increased, there will be more rubber-to-rubber contacts, hence the behavior
of the mixture changed from sand like to rubber like. Liu et al. [16] have examined the
behavior of tire shred-sand mixtures under dry conditions using direct shear, triaxial,
and dynamic triaxial tests. Different gravimetric proportions of rubber such as 0, 10,
30, 50, and 100% were mixed with sand. With the increase in tire shred content, angle
of internal friction corresponding to both the peak and large strain deviatoric strength
reduced. Internal friction angles were found to be independent of the rate of shearing
and confining pressure. Dilative behavior of rubber–sand mixtures decreased with
increase in rubber content due to the flexible nature of rubber particles. Number of
loading cycles has less impact on the shear modulus and damping ratio of dry sand-
rubber tire shred mixtures. Neaz Sheikh et al. [21] inspected the influence of ground
rubber additions on the shear modulus and damping ratio of sand-clay mixtures.
They concluded that the seismic isolation efficiency of sand can be improved by
mixing with rubber particles. Shear stiffness and damping ratio of sand-clay-rubber
mixtures can be increased by mixing the granulated sand with granulated rubber
up to 20% volume fraction. Soil rubber inclusions showed a greater improvement
in the damping behavior of mixtures than soil-acetal inclusions. When materials
such as rubber and sand having different elastic and thermal properties are blended
together, thermo elastic damping occurs at the sand–rubber interfaces. Hence, the
low strain damping ratio increased by three times. Rate of improvement decreased
at higher volume fraction of rubber due to the replacement of sand–rubber contacts
by sand-to-sand or rubber-to-rubber contacts.
Rao and Dutta [27] have performed direct shear tests on particulate rubber–sand
mixtures (PRSM) and particulate rubber fly ash mixtures (PRFM) at various normal
stresses 50, 100, and 150 kPa. Both PRSM and PRFM can be used for base isola-
tion applications due to its ductile nature. Compressible behavior of PRSM can be
significantly reduced by replacing sand with fly ash. Reason may due to the filling
of pores present on the surface of the rubber by the fly ash particles. Shear strength
and angle of internal friction were higher for PRFM as compared to PRSM. This
may be due to the increased particle contacts between particulate rubber and fly ash.
With the progressive replacement of sand with fly ash in PRSM50, shear strength
improved. Also, with the increase in confining pressure and relative density, shear
properties eventually got improved. Madhusudhan et al. [17] have determined static
and dynamic properties of sand–rubber shred mixtures. Rubber shreds were varied in
percentage of 0, 10, 30, 50, and 100% by weight of sand. Shear modulus and damping
ratio decreased with increasing rubber content. Both were decreased with increase
in strain. The behavior of the mixture changed to ductile due to the introduction of
rubber shreds. Mixtures with 10% rubber content can be effectively used for base
isolation applications since it possessed satisfactory static and dynamic properties.
Moghaddas Tafreshi et al. [20] have evaluated the shear and compressibility behavior
of sand-tire crumb mixtures. Sand-tire crumb mixtures were prepared by varying the
volume of tire crumb by 0%, 10, 20, 30, and 40%. They have concluded that the peak
shear strength and the axial strain capacity of sand were largely influenced by the
tire crumb content and the confining pressure. Greater reduction in shear strength
was imparted due to the higher volume fraction of tire crumb in the mixture. With
respect to the increase in confining pressure, peak shear strength and axial capacity
of the mixture got improved. Enhancement in the axial strain capacity was due to
the ductile nature of tire particles.
5 Numerical Studies on Soil-Scrap Tire Mixtures
Yegian and Lahlaf [35] have analyzed the seismic response of high-rise buildings
isolated by a layer of RSM and geotextile layer. Numerical analysis was done using
Plaxis 2D. Peak ground acceleration reduced by 18–48% by the use of RSM layer.
As the amplitude of input motion increased, reduction in peak acceleration due
to the inclusion of RSM layer also increased, whereas lateral displacement of soil
increased due to the inclusion of RSM. Tsang [30] have investigated the dynamic
response of soil-foundation-structure system reinforced with RSM by means of finite
element model created in Strand 7. Raft, pile foundations, and soil were modeled
using quadrilateral plane-strain elements. They have found that by means of RSM
40–60% reduction in acceleration can be attained. As the thickness of the RSM
layer increased, percentage reduction in peak roof acceleration increased from 47 to
73%, and the percentage reduction in peak footing acceleration increased from 35 to
65%. The suitability of using granulated rubber–sand mixtures for the enhancement
of seismic performance of reinforced concrete moment resisting frames (RCMRF)
was also checked by [25]. Sand with 30% by rubber weight was considered for
the study. OpenSees finite element platform was employed for modeling the soil-
structure interaction. The use of RSM was beneficial in the case of medium rise
and high-rise buildings since it reduced the base shear force, bending moment, and
inter-story drift to a considerable percentage. The efficacy of RSM-based isolation
method in terms of overturning stability was determined by [24]. Three-dimensional
analysis of the soil-structure system has been enabled with the help of Abacus.
That model can be regarded as the most effective model for predicting the dynamic
behavior of rubber–sand mixtures since it incorporated the non-linear behavior of
the RSM through shear modulus and damping ratio degradation curves. RSM-based
geotechnical seismic isolation layer benefitted the seismic response of structure by
reducing the base shear and displacements when the rubber content was up to 15–
20% by weight. When the rubber content increased beyond 40%, rocking behavior
of the structure increased. Slenderness ratio of the structure and characteristics of
the input motion also affected the performance of RSM-based GSI.
Xiao et al. [32] have determined the effect of stiffness, length, amount, and the
spacing of geotextile layers on the dynamic response of 15 story medium rise building
under maximum considered earthquake (MCE) loading using FLAC3D. With the
increase in stiffness, length, and number of geotextile layers and with the decrease in
spacing of reinforcing layers, base shear forces increased due to the decrease in energy
dissipation. Most of the stresses induced by rocking of footing were sustained by the
reinforcing layers placed close to the edges of footing. Rocking induced damage and
the post-earthquake settlements can be considerably reduced by increasing the stiff-
ness, length, number, and spacing of geotextile layers. Dhanya et al. [8] carried out
numerical studies using Abacus to assess seismic isolation potential of geotechnical
seismic isolation (GSI) system composed of sand–rubber mixture (SRM) positioned
between the footing and soil. GSI system composed of SRM with 30% rubber content
and one-tenth of the foundation width times thick and reinforced with double layered
geogrid layer were found to be cost-effective solution for reducing the dynamic
response of low-rise buildings. Proposed GSI system can decrease the peak spectral
acceleration (PSA) by 40% for low and medium frequency earthquakes and 30%
for high frequency earthquakes. When the GSI system was reinforced with double
layered geogrids, settlement can be reduced by about 55%. Due to the high damping
nature of rubber particles, amplitude of input motions reduced when it passes through
the GSI layer. Inter-story drift of the building reduced by 20% due to the presence of
reinforcements in the GSI system. Created FEM model has been depicted in Fig. 2.
Fig. 2 Finite element model of building isolated with GSI layer modeled using Abacus [8]
6 Limitations of Rubber–Sand Mixtures
Even though, RSM mixtures are considered as environmentally friendly, some studies
pointed out contradictory remarks regarding the environmental impact of end-of-life
tires. Under acidic condition, leaching from rubber occurs, leading to the release of
harmful metals into the soil. Also, chemicals such as methyl ethyl ketone and toluene
are also found as part of leachate from rubber. Although chemicals and metals are
present in leachate, its effect on ground water quality and thereby on human health
is considered to be marginal. However, long-term studies are required to accurately
predict the environmental effect of rubber–sand mixture [15]. Another disadvantage
is related to the effectiveness of RSM. Seismic isolation of RSM was found to be
effective for peak ground acceleration (PGA) in excess of 0.2–0.25 g. For lower levels
of PGA, non-linear response was not induced resulting in lesser seismic isolation
properties [6].
One of the possible demerits of geotechnical seismic isolation (GSI) system using
scrap tire sand mixtures is the greater amount of settlement induced due to the high
compressibility of rubber fraction. Concept of soil reinforcement can be implemented
to enhance the load carrying capacity and to decrease the excessive deformation
characteristics of RSM. Physical, mechanical, and chemical methods are widely used
to improve the behavior of weak ground. Over the past few decades, reinforcing the
soil by using natural or artificial fibers in order to enhance its strength and rigidity is
gaining more popularity.
7 Conclusions
A thorough literature review has been carried out related to the historical background,
engineering and mechanical properties, numerical modeling, and static and seismic
isolation properties of sand tire shred mixtures. The following conclusions have been
derived from the current study.
• Shear properties of rubber–sand mixtures were influenced by the size and amount
of rubber in the mixture, normal stress, and bulk unit weight of sand. During
shearing, sand tire mixtures exhibit ductile behavior. Compressible behavior of
sand tire shred mixture increased with increase in percentage of rubber in the
mixture. Whereas, brittleness index reduced with the increase in amount of rubber
particles in the matrix. Contradictory results have been obtained with respect to
the influence of rubber content on permeability properties of sand–rubber mixture.
• Contradictory results have been obtained related to the peak shear strength of sand
containing different percentages of shredded tires or tire chip mixtures. Some of
the studies suggested that a peak shear stress was obtained with the increase in
normal stress or horizontal displacement [10, 11], whereas some other works
showed that no such peak can be obtained with respect to the gradual increase in
normal stress [26].
• More rubber like behavior was observed with the increase in confining pressure
at the same rubber content and DD50,r
50,s
greater than 1.
• For shear strain values less than critical strain, damping of sand–rubber shred
mixtures was mainly due to the deformability of the rubber particles. At high shear
strain levels, friction between rubber particles was responsible for the reduction
in damping ratio.
• Uniformity co-efficient, percentage of rubber, and mean grain size of rubber to
sand are the factors which affects the dynamic response of sand–rubber mixtures.
• Peak ground accelerations can be significantly reduced by introducing a layer of
sand–rubber mixture between the foundation and surrounding soil.
• Possible demerit of rubber-based seismic isolation system was the excessive
amount of settlement imposed due to the high compressibility of rubber particles.
This can be avoided by reinforced the geotechnical seismic isolation system using
single or double-layer geogrid. Only few numerical studies have been conducted
on this regard. Future studies should be carried out in this direction to explore the
feasibility of GSI systems for field applications in developing countries.
References
1. Ahmed, I., Lovell, C.W.: Rubber soils as lightweight geomaterials. Transp. Res. Rec. 1422,
61–70 (1993)
2. Anastasiadis, A., Senetakis, K., Pitilakis, K.: Small-strain shear modulus and damping ratio of
sand-rubber and gravel-rubber mixtures. Geotech. Geol. Eng. 30(2), 363–382 (2012)
3. Anbazhagan, P., Manohar, D.R., Rohit, D.: Influence of size of granulated rubber and tyre chips
on the shear strength characteristics of sand–rubber mix. Geomechan. Geoeng. 12(4), 266–278
(2017)
4. Bosscher, P.J., Edil, T.B., Eldin, N.N.: Construction and performance of a shredded waste tire
test embankment. Transp. Res. Rec. 1345, 44–52 (1992)
5. Boominathan, A., Banerjee, S.: Engineering properties of sand–rubber tire shred mixtures. Int.
J. Geotechn. Eng. 15(9), 1061–1077 (2021)
6. Brunet, S., de La Llera, J.C., Kausel, E.: Seismic Isolation Using Recycled Tire-Rubber (2017)
7. Chaturvedi, B., Handa, R.R.: Circulating Tyres in the Economy. 1–68 (2017)
8. Dhanya, J.S., Boominathan, A., Banerjee, S.: Response of low-rise building with geotechnical
seismic isolation system. Soil Dyn. Earthq. Eng. 136, 106187 (2020)
9. Edil, T.B., Park, J.K., Kim, J.Y.: Effectiveness of scrap tire chips as sorptive drainage material.
J. Environ. Eng. 130(7), 824–831 (2004)
10. Foose, G.J., Benson, C.H., Bosscher, P.J.: Sand reinforced with shredded waste tires. J.
Geotechn. Eng. 122(9), 760–767 (1996)
11. Ghazavi, M.: Shear strength characteristics of sand-mixed with granular rubber. Geotech. Geol.
Eng. 22(3), 401–416 (2004)
12. Kolathayar, S., Chitrachedu, R.K.: Model Footing Tests on Sand Bed to Evaluate Efficiency of
Tire Crumb as Infill Materials in Geocells. Springer, Singapore (2022)
13. Li, B., Huang, M., Zeng, X.: Dynamic behavior and liquefaction analysis of recycled-rubber
sand mixtures. J. Mater. Civ. Eng. 28(11), 04016122 (2016)
14. Li, J., Cui, J., Shan, Y., Yadong, L., Ju, B.: Dynamic shear modulus and damping ratio of
sand-rubber mixtures under large strain range. Materials 13(18), 12113–12120 (2020)
15. Liu, L., Cai, G., Zhang, J., Liu, X., Liu, K.: Evaluation of engineering properties and envi-
ronmental effect of recycled waste tire-sand/soil in geotechnical engineering: a compressive
review (2020)
16. Madhusudhan, B.R., Boominathan, A., Banerjee, S.: Factors affecting strength and stiffness of
dry sand-rubber tire shred mixtures. Geotech. Geol. Eng. 37(4), 2763–2780 (2019)
17. Madhusudhan, B.R., Boominathan, A., Banerjee, S.: Static and large-strain dynamic properties
of sand-rubber tire shred mixtures. J. Mater. Civ. Eng. 29(10), 04017165 (2017)
18. Maher, M.H., Gray, D.H.: Static response of sands reinforced with randomly distributed fibers.
J. Geotechn. Eng. 116(11), 55 (1990)
19. Moghaddas Tafreshi, S.N., Joz Darabi, N., Tavakoli Mehrjardi, G., Dawson, A.: Experimental
and numerical investigation of footing behaviour on multi-layered rubber-reinforced soil. Eur.
J. Environ. Civ. Eng. 23(1), 29–52 (2019)
20. Neaz Sheikh, M., Mashiri, M.S., Vinod, J.S., Tsang, H.-H.: Shear and compressibility behavior
of sand-tire crumb mixtures. J. Mater. Civ. Eng. 25(10), 1366–1374 (2013)
21. Pamukcu, S., Akbulut, S.: Thermoelastic enhancement of damping of sand using synthetic
ground rubber. J. Geotechn. Geoenviron. Eng. 132(4), 501–510 (2006)
22. Pincus, H., Edil, T., Bosscher, P.: Engineering properties of tire chips and soil mixtures. Geotech.
Test. J. 17(4), 453 (1994)
23. Pistolas, G.A., Anastasiadis, A., Pitilakis, K.: Dynamic behaviour of granular soil materials
mixed with granulated rubber: influence of rubber content and mean grain size ratio on shear
modulus and damping ratio for a wide strain range. Innov. Infrastruct. Sol. 3(1) (2018)
24. Pistolas, G.A., Pitilakis, K., Anastasiadis, A.: A numerical investigation on the seismic isolation
potential of rubber/soil mixtures. Earthq. Eng. Eng. Vib. 19(3), 683–704 (2020)
25. Pitilakis, K., Karapetrou, S., Tsagdi, K.: Numerical investigation of the seismic response of
RC buildings on soil replaced with rubber-sand mixtures. Soil Dyn. Earthq. Eng. 79, 237–252
(2015)
26. Rao, G.V., Dutta, R.K.: Compressibility and strength behaviour of sand-tyre chip mixtures.
Geotech. Geol. Eng. 24(3), 711–724 (2006)
27. Senthen Amuthan, M., Boominathan, A., Banerjee, S.: Density and shear strength of particulate
rubber mixed with sand and fly ash. J. Mater. Civ. Eng. 30(7), 04018136 (2018)
28. Shivaprakash, B.G., Dinesh, S.V.: Dynamic properties of sand-fines mixtures. Geotech. Geol.
Eng. 35(5), 2327–2337 (2017)
29. Tsang, H.: Seismic isolation by rubber—soil mixtures for developing countries. Earthq. Eng.
Struct. Dyn. 37(2), 283–303 (2008)
30. Tsang, H., Lo, S.H., Xu, X., Sheikh, M.N.: Seismic isolation for low-to-medium-rise buildings
using granulated rubber—soil mixtures: numerical study. Earthq. Eng. Struct. Dyn. 41(14),
2009–2024 (2012)
31. Xiao, H., Butterworth, J.W., Larkin, T.: Low-technology techniques for seismic isolation.
NZSEE Conf. 36, 36–43 (2004)
32. Xu, R., Fatahi, B.: Influence of geotextile arrangement on seismic performance of mid-rise
buildings subjected to MCE shaking. Geotext. Geomembr. 46(4), 511–528 (2018)
33. Yegian, M.K., Catan, M.: Soil isolation for seismic protection using a smooth synthetic liner.
J. Geotechn. Geoenviron. Eng. 130(11), 1131–1139 (2004)
34. Yegian, M.K., Lahlaf, A.M.: Dynamic interface shear strength properties of geomembranes
and geotextiles. J. Geotechn. Eng. 120(2), 462–465 (1994)
35. Yildiz, Ö.: Geotechnical seismic isolation method using rubber-soil mixtures. NATURENGS
MTU J. Eng. Nat. Sci. Malatya Turgut Ozal Univ. 1, 34–43 (2021)
Development of Soil Amplification
Factors Using 1D and 2D Ground
Response Analysis
A. Sharma and S. Adhikary
Abstract 1D ground response analysis is quite popularly used for determining

soil amplification numerically. 1D GRA works well for layered soil material with
infinitely extending boundaries. However, in case of irregular ground, complex site
stratigraphy, river basin, etc., 2D analysis or even 3D analysis is preferable where
soil properties in all the corresponding direction are incorporated and thus depicts the
real behaviour when it is subjected to an earthquake motion. In the present study, an
attempt has been made to determine soil amplification using the 1D and 2D GRA. The
analysis is carried out on numerous multilayer real soil profiles taken from author’s
past work. As a first step, 1D EL GRA is conducted for 8 real sites from India, using
the DEEPSOIL software. Spectrum compatible time histories matched with the Type-
I spectra of Indian code, corresponding to effective peak ground acceleration (EPGA)
of 0.36 g, are used. Deconvoluted motion of earthquakes is applied at the bottom of
strata/ preferably on bedrock, and the response throughout the thickness of the strata
is recorded in order to get the soil amplification profile. Later, using commercially
available FEM software, ABAQUS 1D and 2D analysis are done where the layered
soil medium is modelled using continuum approach. The results from 1 and 2D are
compared together in terms of response spectra and amplification factors. This study
brought forth the comparison of GRA results using discrete and continuum approach
in 1D and 2D analysis.
Keywords Soil amplification · Seismic response · Ground response analysis ·

1D · 2D
1 Introduction
Earthquake is a naturally occurring phenomenon, which releases massive amount

of energy when triggered and takes place due to tectonic movement of earth crust,
A. Sharma · S. Adhikary (B)

Department of Civil Engineering, VNIT Nagpur, Nagpur, Maharashtra 440010, India
e-mail: sadhikary@civ.vnit.ac.in
https://doi.org/10.1007/978-981-99-1459-3_24
298 A. Sharma and S. Adhikary
incurring tremendous loss of infrastructure and lives. The earthquake strong ground
motions vary from place to place across the surface of earth depending on the seismic
zone and type of soil bed underneath. The local soil conditions play a very important
role in changing the dynamic characteristics of an earthquake. The properties of
different soil layers may amplify or deamplify the earthquake ground motion at the
soil surface. Due to this nature of soil, several structures resting on soft soils suffered
tremendous damage [1]. The primary features that influence the local alterations to
the applied strong ground motions are the stratigraphy of deposited soil, location of
bedrock, etc. [2].
Many researchers showcased their expertise in this direction and contributed along
the way. Desai and Choudhury [3] performed the ground response analysis of a few
crucial sites and ports of Mumbai city, India. SHAKE2000 [4] was used to perform
1D equivalent linear GRA on typical soil profiles from the ports and crucial locations.
They concluded that a few sites were highly susceptible to earthquakes amongst the
selected sites. Amorosi et al. [5] conducted ground response analyzes of a cohesive
deposit using finite element computer software PLAXIS2D [6]. This FE method
adopted a linear cum visco-elastic and visco-elasto-plastic constitutive soil models.
The finite element results from PLAXIS2D were later compared with 1D equivalent
linear, visco-elastic, and frequency-domain analyzes obtained from SHAKE91 [7]
computer software. GovindaRaju et al. [2] presented a case study by conducting a
ground response analysis of a site in Ahmedabad city, India during the Bhuj earth-
quake. They enlightened the engineering importance and difficulties faced whilst
performing GRA and proposed the method to perform the same. A total stress anal-
ysis approach [8] of ground response analysis was performed on a major earthquake-
hit site Lotung, Taiwan (M6.5 and M7.0) in 1986, using widely used equivalent
linear analysis code SHAKE91 [7] and time domain finite element nonlinear code
SPECTRA. These two algorithms were compared by analyzing records of major
earthquakes on selected sites in their research work.
Bard and Bouchon [9] have given an empirical relation to compute the funda-
mental frequency of the soil deposits. Also, Kumar and Narayan [10] have worked
on the response of different shapes of the subsoil basins (namely rectangular and
elliptical). Admirable correlation has been obtained between their observed funda-
mental frequency and the computed frequency using empirical relation by [9]. Later,
with these results, a new empirical relation has been generated via regression analysis
in order to predict the fundamental frequency of elliptical soil basins. And the results
from this relation have shown an increase in the fundamental frequency of soil with
increase in depth to width ratio of the basin.
Ranjan [11] gathered soil profile data from an Indian city, Dehradun, by measuring
shear wave velocity of soil using multi-channel analysis of surface waves method
(MASW). He collected this data from a total of 31 number of sites located across
the city. SHAKE91 [7] was extensively used for the GRA. The study classified
the entire city into different zones depending on the measured shear wave velocity
and observed spectral acceleration for the purpose of deterministic seismic hazard
analysis and probabilistic seismic hazard analysis.
Development of Soil Amplification Factors Using 1D and 2D Ground … 299
Shukla and Choudhury [12] prepared consistent seismic ground motions for the
structures situated closer to the ports located in Gujarat. The aim of their study was to
evaluate the potential of seismic hazard for the few chosen ports, viz. Hazira, Kandla,
and Mundra ports in the state and to develop site-specific ground motions for a scale
of three-level of ground shaking, viz. operational level earthquake, contingency level
earthquake, and maximum considered earthquake.
Similar study has been conducted for a few other sites in Kolkata city, India [13]
and Bangalore city, India [14] where they have collected data from several typical
soil sites and conducted equivalent linear ground response analysis using computer
programme SHAKE2000 [7] with a few typical earthquake ground motions. These
profiles were elaborated using MASW test to get their shear wave velocity profile
along the depth. Recently, Deoda and Adhikary [15, 16] developed the site classi-
fication and amplification factors for Indian site conditions using equivalent linear
and nonlinear GRA using DEEPSOIL software [17].
The present study aims to investigate the equivalent linear GRA using discrete and
continuum approach. For this purpose at first, 1D equivalent linear GRA is carried
out in DEEPSOIL for 8 sites considered, and then, the same problem is considered in
the commercially available FEM software ABAQUS [18] in 1D and 2D using plane
strain elements and proper boundary conditions. The results are compared in terms
of response spectra and amplification curves. It is observed that there is significant
difference between the approaches considered.
2 Selected Sites and Earthquake Ground Motions
2.1 Sites Selected for the Study
The aim of this study is to carry out ground response analysis of a few selected
sites of India chosen from the literature. The sites were classified into different soil
classes (I, II, and III as per IS: 1893-2016 [19]). These sites are spread across India
and are a good representation of a variety of soil profiles. This classification of
soils in Indian Standards can be easily understood from Table 1. These sites were
taken from various literature [15, 16] listed in Table 2 along with all the required
information to conduct equivalent linear ground response analysis in DEEPSOIL
and ABAQUS. Information like the complete profile of the site with thickness of
individual layers, their unit weight, shear wave velocity, and the dynamic properties
(strain dependent properties), viz. modulus reduction curve and damping ratio curve
is mentioned. For any particular site, depth of profile, average shear wave velocity,
fundamental period of profile, and site class as per Indian seismic code IS: 1893
(part-1) 2016 is mentioned in Table 2. The Indian seismic code classifies the soil
only on the basis of one parameter, i.e. the standard penetration test (SPT) value.
However, Deoda and Adhikary [15] suggest that only SPT value does not provide
the complete understanding of the soil underneath, but average shear wave velocity
Table 1 Classification of soil type as per IS 1893:2016

Site class-I Site class-II Site class-III
Name Hard soil Medium soils Soft soils
N SPT > 30 10–30 < 10
Table 2 Soil profiles considered in the present study and classified as per IS 1893:2016 and Deoda
and Adhikary [15]
Author(s) Site Depth of Avg. shear Fundamental Site class as Site class
profile wave period (s) per IS as per
(m) velocity, 1893:2016 Deoda and
V s,avg (m/s) Adhikary
[15]
Ranjan [11] Site 1 31.7 406.97 0.31 I B
Site 2 29.96 334.65 0.36 I B
Site 4 31.94 318.69 0.4 I B
Shukla and Hazira 30 234.32 0.51 II C
Choudhury Kandla 32 236.45 0.45 II C
[12]
Mundra 30 237.26 0.42 II C
Chatterjee BH-1 Park 31 258.54 0.48 III D
and Street
Choudhury BH-3 40.5 228.00 0.71 III D
[13] Rajarhat
is a more reliable parameter for the classification of soil. They suggest a few more
parameters must be taken into account to observe the behaviour of the soil. In their
study, fundamental period of profile, T o and average shear wave velocity, V s,avg were
dominantly used to assess the soil class, and the same has been followed in this study.
All these sites were selected from the literature carried out before conducting
this computational study. These sites are distributed across Indian subcontinent, viz.
Hazira, Kandla, and Mundra are located on the very western part of the country, i.e.
Gujarat. Site 1, Site 2, and Site 4 are located in the northern part of the country, i.e.
Dehradun, Uttarakhand. Whereas BH-1 Park Street and BH-3 Rajarhat are situated
in the eastern side of the country, i.e. Kolkata, West Bengal.
2.2 Ground Motion Selection
Selection of suitable ground motions has a very prominent aspect whilst performing
ground response analysis for any site. The ground motion we use to apply as a
base excitation motion to any soil profile should have the same frequency as that
of the soil class the site belongs. Hence, a total of ten numbers of natural recorded
Table 3 Earthquake ground motions considered in this study

Earthquake Recording Year of Magnitude, Distance Duration of
parameters station occurrence Mw from earthquake
Earthquake source (s)
(km)
India-Burma Umsning 1988 6.1 343.8 70.52
Border
India-Burma Umsning, India 1997 5.6 106.8 27.28
Border
NE India Pynursla 1986 4.5 48.2 18.52
Loma Prieta Gilroy Array #1 1989 6.93 8.84 79.89
Landers Lucerne 1992 7.28 2.19 96.16
Kobe, Japan Kobe University 1995 6.9 0.9 31.96
India-Bangladesh Nongkhlaw 1988 5.8 117.3 45.2
Border
Uttarkashi Uttarkashi 1991 7 34 39.84
El Mayor-Cucapah, Blythe 2010 7.2 164.38 120.01
Mexico
San Simeon, CA Diablo Canyon 2003 6.52 37.92 29.43
Power Plant
ground motions were made compatible with Type-I spectra of IS: 1893 (part-1)
2016 for 0.36 g (i.e. Zone V) were used in this study. These earthquake motions
were taken from databases, viz. Pacific Earthquake Engineering Research Centre
and The Consortium for Strong-Motion Observation Systems-Virtual Data Centre.
The necessary information of these ground motions is specified in Table 3. Spectral
plot of these spectrum compatible earthquakes is plotted in Fig. 1 along with the
target spectra of Type-1 spectra of IS: 1893 (part-1) 2016 for 0.36 g. The procedure
has been explained in detail [15].
3 Methodology
3.1 Ground Response Analysis
Ground response analysis is a simple and common test which is performed before
any construction and before conducting any computational research work. Seeking
this as a need, a few renowned and easy-to-use softwares were formulated using
codes in different computer language to perform this study, viz. SHAKE91 [7] and
DEEPSOIL [17].
The 1D GRA is widely used since early days for design projects using the same
methodology but often gives conservative results [2]. GRA is performed for various
Fig. 1 5% damped 1.5

acceleration response spectra
of selected ground motions 1.25
with the target of Type-1
spectra of IS: 1893-2016 1
PSA (g)
0.75
0.5
0.25
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Period (s)
reasons, viz. to calculate natural period of site, to assess ground amplification, to

prepare spectra for the site, and to evaluate liquefaction potential [2]. 1D equivalent
linear, visco-elastic, and frequency-domain analyzes with DEEPSOIL include the
use of shear wave velocity profile along the depth, modulus reduction curve which
gives the variation of shear modulus with shear strain, and the damping ratio curve
which gives the variation of damping with shear strain.
Ground response analysis (GRA) may be addressed via linear, equivalent linear, or
nonlinear approach in one, two, or three dimensions (1D, 2D, or 3D). The nonlinear
technique makes use of cyclic stress–strain models (backbone curves) in order to
describe the irregularity in the material properties shear modulus (G) and damping
ratio (ξ), whereas the linear approach asks for constant values of G and ξ for an
induced level of tiny shear strain in each layer. In the equivalent linear method, the
nonlinearity of the shear modulus and damping with cyclic shear strain is taken into
account by employing the equivalent linear soil characteristics through an iterative
technique, to get values of the G and ξ that are consistent with the effective shear
strains in each layer [2]. In this study, one-dimensional equivalent linear analysis is
performed in DEEPSOIL and later compared with the FEM model of the 1D analysis
in ABAQUS.
3.2 GRA Using DEEPSOIL (Discrete Soil Modelling)
DEEPSOIL [17] is a one-dimensional ground response analysis tool that can perform
1D equivalent linear frequency dependent analysis and 1D nonlinear time dependent
analysis even with pore water pressure generation. The software works on Kelvin-
Voigt model which can be easily represented by a parallel assembly prepared with
a viscous spring and a viscous damper hereby having the property of elasticity and
viscosity. The source code utilizes the fundamental idea of successive iterations
(a) (b) (c)
Fig. 2 Soil profiles prepared in different softwares, a DEEPSOIL, b ABAQUS 1D model and c
ABAQUS 2D model
to solve the 1-dimensional wave equations to represent the vertical propagation of

S–H waves. Bedrock is considered elastic half-space, and the input motions are
deconvoluted before been applied. Figure 2a shows the representative model of 1D
equivalent linear GRA using DEEPSOIL.
3.3 GRA Using ABAQUS (Continuum Soil Modelling)
For the purpose of 1D continuum modelling, ABAQUS software has been used. The
dynamic soil properties, i.e. modulus reduction curve and damping ratio curve, of
each soil layer were determined using the strain observed whilst performing the 1D
DEEPSOIL analysis. A similar study was conducted by Nautiyal et al. [20]. They
chose ABAQUS to perform a 1D, 2D, and 3D comparison of 2 sites with their soil
properties and shear wave velocity profile along the depth and found to be in exact
correspondence with each other.
The model for two-dimensional (2D) study is prepared with 4-noded, plane strain
elements with the individual element size kept 1 m × 1 m. The element size and
width of the 2D analysis used in this model were chosen so as to get the correct
solution after performing sensitivity analysis.
The prepared model is 36 m wide with element size of 1 m × 1 m (refer Fig. 2c) and
its total depth and thickness of individual layers vary as per the chosen soil profile.
Several different such models were prepared, and dynamic analysis was conducted
for ten numbers of earthquake motions for all the selected sites constituting eighty
individual analysis. The chosen earthquake motions and the soil profile for this study
were specified in the earlier sections.
The earthquake motion is applied at the base of the profile where the horizontal
boundary is fixed for the movement of the model, and non-reflecting infinite boundary
is provided for smoothly passage of the earthquake waves in the vertically downward
direction. Similar non-reflecting infinite boundary is provided at both the sides of
the profile in horizontal direction giving the model an adequate length to dissipate
the energy of the earthquake and restricting the waves to reflect and enter again in
the model.
The model for one-dimensional (1D) study is only two metre wide plane strain
elements, slender, and multilayer soil column (refer Fig. 2b), that is why, the non-
reflecting boundaries could not be employed here rather 1D dashpots are attached to
the nodes at both the sides of the model to create an artificial boundary to dissipate
the earthquake energy. These dashpots attached at the edges of the model which is
commonly referred as viscous boundary condition which helps in smooth passage of
the earthquake waves. This dashpot boundary condition works well with two degree
of freedom at the edge of the model, i.e. ux and uy . The stiffness coefficient for these
dashpots is required to be provided which were determined using the procedure given
in [21]. These calculated values of dashpot-stiffness changes with the material, i.e.
calculations have to be repeated for each layer and for all the soil profiles.
Structural damping is considered in the numerical modelling (1D and 2D) via
Rayleigh damping in terms of mass proportional coefficient, α (unit = s−1 ) and
stiffness proportional coefficient, β (unit = s) correspond to the first and the second
modal natural frequency of the soil profile obtained after performing a modal anal-
ysis for a particular soil profile, and the eigenvalues are evaluated using Lanczos
iteration method in ABAQUS. The procedure to calculate the values of these coef-
ficients is given in Eq. (1), where ωi and ωj are the first and second modal natural
frequencies (Hz), respectively, and ξ is the damping ratio considered for this study.
The values of first and second modal natural frequencies and their corresponding α
and β coefficients for a few sites are shown in Table 4 and are compared with the
profile’s natural frequency obtained in DEEPSOIL.
Table 4 Natural modal frequencies and stiffness coefficients obtained for two typical sites using
different softwares
Software DEEPSOIL ABAQUS 1D ABAQUS 2D
Site ω ωi ωj α (× β (× ωi ωj α (× β (×
10–4 ) 10–4 ) 10–4 ) 10–4 )
Site- 1 3.21 2.86 3.60 2005 4.92 3.20 3.29 2041 4.9
Site- 2 2.49 2.46 3.05 1714 5.770 2.46 2.51 1563 6.4
It was observed that the frequencies obtained using DEEPSOIL, ABAQUS 1D,
and ABAQUS 2D are in very good agreement with one another, and it should be,
because the soil profile and material properties are the entities required to perform
this study, and they were kept same in all the softwares. Also it was noted that
the two modal frequencies obtained using ABAQUS 2D are closer to each other
compared to the two modal frequencies obtained using ABAQUS 1D. And natural
frequency of soil column obtained using DEEPSOIL also lies very much between
the two modal frequencies obtained using ABAQUS 2D. It suggests modelling in
2-dimension gives better result when performing analysis of soil profile and possibly
for combined soil-structure profiles too.
2ωi ω j 2
α=ξ ,β = ξ (1)
ωi + ω j ωi + ω j
The deconvoluted ground motion time histories are applied at the base of each soil
profile considered and the output time history, commonly known as response time
history, at the surface of the soil layer is observed. This analysis of observing the
behaviour of soil layer on the surface by applying earthquake motion at the base
is termed as ground response analysis. This analysis is performed several times by
applying the ten selected earthquake motions on the eight selected sites, thus consti-
tuting total eighty number of analyzes for 1D case model in ABAQUS. The results
from these many analyzes are compared with the similar number of eighty analyzes
results performed in DEEPSOIL software. Similar eighty number of analyzes is also
performed in ABAQUS with the 2D model. The response history observed at the
surface and at the surface of all the adjacent layers of the soil profile is recorded for
all the soil profiles, and the peak acceleration value for all these layers was deter-
mined and plotted against the depth of the soil profile. This value for every response
at the surface of all layers is known as peak ground acceleration PGA (g). The vari-
ation of the PGA (g) along the depth of the profile is plotted for all the selected
sites and for all the earthquake motions as shown in Fig. 3. These curves are plotted
considering the mean of the response of all earthquake motions for all eight sites for
DEEPSOIL, ABAQUS 1D, and ABAQUS 2D. As it is evident from these plots, the
curves representing ABAQUS 1D and 2D are in very good agreement with the 1D
analysis results obtained from DEEPSOIL.
The spectra for the response observed at the surface of soil profile are plotted
alongside the spectra of input earthquake motion for all the analyzes conducted in
different softwares which are shown in Fig. 4. It can be clearly seen from these figures
that the spectra obtained from DEEPSOIL depict larger period than ABAQUS 1D and
ABAQUS 2D in all cases. It can also be stated as 1D discrete model analysis procedure
gives larger period than 1D or 2D continuum model analysis. One good explanation
Site- 1 Site- 2 Site- 4

PGA (g) PGA (g) PGA (g)
0.25 0.3 0.35 0.4 0.45 0.5 0.25 0.3 0.35 0.4 0.45 0.5 0.15 0.2 0.25 0.3 0.35 0.4 0.45
0 0 0
4 4 4
8 8 8
12 12 12
Depth (m)
Depth (m)
Depth (m)
16 16 16
20 20 20
24 24 24
28 28 28
DeepSoil DeepSoil DeepSoil
32 Abaqus 1D 32 Abaqus 1D 32 Abaqus 1D
Abaqus 2D Abaqus 2D Abaqus 2D
36 36 36
Site- Hazira Site- Kandla Site- Mundra
PGA (g) PGA (g) PGA (g)
0.25 0.35 0.45 0.55 0.65 0.75 0.25 0.35 0.45 0.55 0.2 0.3 0.4 0.5 0.6 0.7
0 0 0
4 4 4
8 8 8
12 12 12
Depth (m)
Depth (m)
Depth (m)
16 16 16
20 20 20
24 24 24
28 28 28
DeepSoil DeepSoil DeepSoil
32 Abaqus 1D 32 Abaqus 1D 32 Abaqus 1D
Abaqus 2D Abaqus 2D Abaqus 2D
36 36 36
Site- BH-1 Park Street Site- BH-3 Rajarhat

PGA (g) PGA (g)
0.35 0.4 0.45 0.5 0.55 0.25 0.35 0.45 0.55 0.65
0 0
4 4
8
8
12
12
Depth (m)
Depth (m)
16
16 20
20 24
28
24
32
28
DeepSoil 36 DeepSoil
32 Abaqus 1D 40 Abaqus 1D
Abaqus 2D Abaqus 2D
36 44
Fig. 3 Peak ground acceleration, PGA (g) along the depth for all considered sites
can be given for this like, DEEPSOIL predicts the natural period of any strata by
simply considering the average shear wave velocity, V s,avg and depth of the profile,
H referring Eq. (2), whereas ABAQUS keeps into account the material properties
also whilst determining the natural frequency of the soil column. Hence, gives better
results for natural period and for all other types of analyzes. Also it is noted from these
figures that the peak spectral acceleration obtained from DEEPSOIL, ABAQUS 1D,
and ABAQUS 2D is equal for all the soil profiles.
VS,avg
ω= (2)
4H
Figure 5 depicts the curves known as the amplification factor curves which are
plotted for all the sites with respect to the applied input motion. These plots simplify
the way of understanding the spectral amplification of soil response with respect to
the input motion for all the considered sites. Each curve represents the amplifica-
tion of the corresponding ordinates of the response spectra at the surface of soil to
the response spectra of the applied input motion. It can be clearly observed that the
amplification in the amplification factor curve representing “DEEPSOIL 1D Ampli-
fication Factor” is highly amplified in all the cases compared to the amplification
factor curve representing “ABAQUS 1D Amplification Factor” and “ABAQUS 2D
Amplification Factor”. The simple reason behind this huge amplification is that the
peak of the corresponding DEEPSOIL spectra has large period of natural oscillation
compared to the ABAQUS 1D and ABAQUS 2D spectra. Also the peak attained by
ABAQUS 2D curve is higher in all the cases compared to ABAQUS 1D. Similar

1.5 1.25 1.25
DeepSoil 1D Mean Spectra Surface DeepSoil 1D Mean Spectra Surface DeepSoil 1D Mean Spectra Surface
AbaQus 1D Mean Spectra Surface AbaQus 1D Mean Spectra Surface AbaQus 1D Mean Spectra Surface
1.25 Mean Input Motion 1 Mean Input Motion 1 Mean Input Motion
Spectral Acceleration (g)

1
0.75 0.75
0.75
0.5 0.5
0.5
0.25 0.25
0.25
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4
Period (sec) Period (sec) Period (sec)

2.5 2 1.75
DeepSoil 1D Mean Spectra Surface DeepSoil 1D Mean Spectra Surface DeepSoil 1D Mean Spectra Surface
AbaQus 2D Mean Spectra Surface AbaQus 2D Mean Spectra Surface 1.5 AbaQus 2D Mean Spectra Surface
2 Mean Input Motion Mean Input Motion Mean Input Motion
1.5
1.25
1.5
1
1
0.75
1
0.5
0.5
0.5
0.25
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4

2 1.5
DeepSoil 1D Mean Spectra Surface DeepSoil 1D Mean Spectra Surface
AbaQus 1D Mean Spectra Surface AbaQus 1D Mean Spectra Surface
1.75 AbaQus 2D Mean Spectra Surface AbaQus 2D Mean Spectra Surface
1.25
Mean Input Motion Mean Input Motion
1.5
1
1.25
1 0.75
0.75
0.5
0.5
0.25
0.25
0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4
Period (sec) Period (sec)
Fig. 4 Response spectra for all the considered sites


2.5 2.5 2.5
DeepSoil 1D Amplification Factor DeepSoil 1D Amplification Factor DeepSoil 1D Amplification Factor
AbaQus 1D Amplification Factor AbaQus 1D Amplification Factor AbaQus 1D Amplification Factor
2 2 2
Amplification Factor
1.5 1.5 1.5
1 1 1
0.5 0.5 0.5
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4

3.5 3 3.5
DeepSoil 1D Amplification Factor DeepSoil 1D Amplification Factor DeepSoil 1D Amplification Factor
3 3 AbaQus 2D Amplification Factor
AbaQus 2D Amplification Factor 2.5 AbaQus 2D Amplification Factor
2.5 2.5
2
2 2
1.5
1.5 1.5
1
1 1
0.5 0.5
0.5
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4

3 4.5
DeepSoil 1D Amplification Factor DeepSoil 1D Amplification Factor
AbaQus 1D Amplification Factor 4 AbaQus 1D Amplification Factor
2.5 AbaQus 2D Amplification Factor AbaQus 2D Amplification Factor
3.5
2 3
2.5
1.5
2
1 1.5
1
0.5
0.5
0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4
Period (sec) Period (sec)
Fig. 5 Amplification factor for all the considered sites
conclusion can be observed here as well that the peak of ABAQUS 2D is earlier
than the peak of ABAQUS 1D when comparing ABAQUS 1D and 2D modelling.
Also the ABAQUS 2D curve is highly deamplified compared to ABAQUS 1D and
DEEPSOIL, and the reason for this could be the nonlinearity of the soil layers which
significantly damped out the seismic waves.
5 Conclusion
In this study, several numerical analyzes were performed on various real sites consid-
ered from India with various earthquake motions, to get the PGA of the response for
all those sites along the depth. The selected sites and chosen earthquake ground
motions with the criteria of their selection are specified in the earlier sections.
These analyzes were performed in different softwares, i.e. DEEPSOIL as a discrete
approach and ABAQUS as a continuum approach and are summarized as a compar-
ative study. This comparison was shown in the results and discussion section and
that is in very good correspondence with each other, and the appropriate reasoning
is provided for all the deviations. The study reflects the difference in the consid-
ered approaches for modelling ground response analysis. The results obtained from
the discrete approach are on the conservative side. Further, it is observed that for
2D continuum approach, there is significant deamplification, and the GRA is not
captured efficiently for the considered site profiles. Nevertheless, for spatial variation
and heterogeneity in soil profile, 2D continuum approach is recommended.
References
1. Kramer, S.L.: Geotechnical Earthquake Engineering. Pearson Education, Inc. (2013)

2. GovindaRaju, L., Ramana, G.V.: Site-specific ground response analysis. JSTOR 87(10), 1354–
1362 (2004)
3. Desai, S.S., Choudhury, D.: Site-specific seismic ground response study for nuclear power
plants and ports in Mumbai. Nat. Hazards Rev. 16(4), 04015002 (2015). https://doi.org/10.
1061/(ASCE)NH.1527-6996.0000177
4. Ordóñez, G.A.: SHAKE2000—A Computer Program for the 1-D Analysis of Geotechnical
Earthquake Engineering Problems. GeoMotions, LLC, Lacey, Washington, USA (2012)
5. Amorosi, A., Boldini, D., Elia, G.: Parametric study on seismic ground response by finite
element modelling. Comput. Geotech. 37(4), 515–528 (2010). https://doi.org/10.1016/j.com
pgeo.2010.02.005
6. Bentley Systems, “PLAXIS”
7. Schnabel, P.B., Lysmer, J.: SHAKE91: a computer program for earthquake response analysis
of horizontally layered sites
8. Borja, R.I., Duvernay, B.G., Lin, C.-H.: Ground response in Lotung: total stress analyses and
parametric studies. J. Geotech. Geoenviron. Eng. 128(1), 54–63 (2002). https://doi.org/10.
1061/(ASCE)1090-0241(2002)128:1(54)
9. Bard, P.-Y., Bouchon, M.: The two-dimensional resonance of sediment-filled valleys. Bull.
Seismol. Soc. Am. 75(2), 519–541 (1985). https://doi.org/10.1785/bssa0750020519
10. Kumar, N., Narayan, J.P.: Study of 2D basins and site-city interaction effects on ground motion
characteristics. J. Indian Geophys. Union 22(1), 16–23 (2018) [online]. Available https://www.
researchgate.net/publication/320345540
11. Ranjan, R.: Seismic Response Analysis of Dehradun City, India. M.Sc. thesis, International
Institute for Geo-Information Science and Earth Observations, Enschede (2004)
12. Shukla, J., Choudhury, D.: Seismic hazard and site-specific ground motion for typical ports of
Gujarat. Nat. Hazards 60(2), 541–565 (2012). https://doi.org/10.1007/s11069-011-0042-z
13. Chatterjee, K., Choudhury, D.: Influences of local soil conditions for ground response in Kolkata
city during earthquakes. Proc. Natl. Acad. Sci. India Sect. A Phys. Sci. 88(4), 515–528 (2018).
https://doi.org/10.1007/s40010-016-0265-1
14. Anbazhagan, P., Sitharam, T.G.: Site characterization and site response studies using shear
wave velocity. J. Seismol. Earthq. Eng. 10(1), 53–67 (2008)
15. Deoda, V.R., Adhikary, S.: A preliminary proposal towards the revision of Indian seismic code
considering site classification scheme, amplification factors and response spectra. Bull. Earthq.
Eng. 18(6), 2843–2889 (2020). https://doi.org/10.1007/s10518-020-00806-2
16. Deoda, V.R., Adhikary, S.: Revision of IS 1893: proposal for an alternative soil classification
scheme and associated intensity-dependent spectral amplification factors. Nat. Hazards Rev.
23(2) (2022).https://doi.org/10.1061/(asce)nh.1527-6996.0000554
17. Hashash, Y.M.A., Musgrove, M.I., Harmon, J.A., Ilhan, O., Xing, G., Numanoglu, O.,
Groholski, D.R., Phillips, C.A.: DEEPSOIL v7.0, User Manual. Urbana, IL, Board of Trustees
of University of Illinois at Urbana-Champaign (2020)
18. Dassault Systemes: Abaqus/CAE, Simulia v. 2017

19. IS 1893 (Part-1): Criteria for Earthquake Resistant Design of Structures. Part 1: General
Provisions and Buildings. Bureau of Indian Standards, New Delhi (2016)
20. Nautiyal, P., Raj, D., Bharathi, M., Dubey, R.: Ground response analysis: comparison of 1D,
2D and 3D approach. Proc. Indian Geotechn. Conf. 2021, 607–619 (2019). https://doi.org/10.
1007/978-981-33-6564-3_51
21. Tidke, A.R., Adhikary, S.: Seismic fragility analysis of the Koyna gravity dam with layered
rock foundation considering tensile crack failure. Eng. Fail. Anal. 125, 105361 (2021). https://
doi.org/10.1016/j.engfailanal.2021.105361
Probabilistic Seismic Hazard Assessment
for Assam, North-East India
M. Borah, M. L. Sharma, and R. N. Dubey
Abstract In this study, probabilistic seismic hazard assessment (PSHA) is

performed for Assam, North-East (NE) India. NE India being bounded by latitude
20°–30° N and longitude 87°–98° E is considered as one of the most earthquake-
prone areas in the world. As per seismic zoning map of India [12], most of the states
in NE region have been placed in seismic zone V, which has the highest zone factor
in the country. Amongst the eight north-eastern states, Assam serves as the gateway
to the other seven states. As this region lies on one of the most vigorous tectonic
plates in the world, it has experienced several devastating earthquakes in the past.
The most devastating earthquakes include 1869 Cachar Earthquake (M w = 7.5), 1897
Assam-Shillong Earthquake (M w = 8.1), and 1950 Assam Earthquake (M w = 8.7).
Considering the seismicity of this region, seismic hazard assessment plays a signif-
icant role to assess the seismic risk for the future. The NE India region is broadly
divided into four seismogenic sources and further sub-divided into eleven seismo-
genic sources based on the tectonic features and seismicity characteristics. For the
study of hazard assessment, a unified moment magnitude catalogue has been used,
where the events are assembled from various databases (ISC, IMD, USGS-NEIC).
The catalogue has been declustered, and the seismicity parameters are calculated
for each source zone. The hazard maps have been presented at the bedrock level, in
terms of peak ground acceleration (PGA) and spectral acceleration (S a ) values. The
PGA values vary in between 0.16 and 0.57 g, whilst the S a values are obtained in
the range of 0.12–0.77 g. These hazard maps are expected to give insight to the local
site-specific seismic hazard variation for the Assam region and would be useful for
the preparedness of risk and disaster mitigation measures.
M. Borah (B) · M. L. Sharma · R. N. Dubey

e-mail: mborah@eq.iitr.ac.in
M. L. Sharma
R. N. Dubey
e-mail: rn.dubey@eq.iitr.ac.in
https://doi.org/10.1007/978-981-99-1459-3_25
312 M. Borah et al.
Keywords PSHA · NE India · Assam · PGA · Hazard map
1 Introduction
Earthquake, considered as one of the most damaging natural phenomena, has been
destroying human life and human-made structures up to a great extent. Due to earth-
quakes, numbers of shakings ranging from small to great have occurred from time
to time, and to date, there are no such measures developed to predict earthquakes
accurately. Indian sub-continent is prone to natural hazards due to its predominant
geo-climatic conditions. It has experienced numerous destructing disasters leading to
loss of lives and properties. Developing countries have more vulnerability to hazards,
and the lack of proper disaster management leads to increase in risk in these coun-
tries. Most of the civil structures and infrastructures concentrate in these developing
countries which are directly affected by earthquakes. For this very reason, seismic
hazard assessment plays an important role in the urban areas to minimize the damage
due to earthquakes up to a possible extent.
The NE India is one of the most seismically active regions around the globe. This
region has been witnessing a number of devastating earthquakes, including Cachar
earthquake (1869, M w = 7.8), Assam EQ (1897, M w = 8.1), Meghalaya EQ(1923,
M w = 7.1), Dhubri EQ (1930, M w = 7.1), Assam EQ (1943, M w = 7.6), Arunachal
Pradesh EQ (1947, M s = 7.7), Assam EQ (1950, M w = 8.7), Manipur EQ (1988, M s
= 7.3), Assam EQ (2009, M w = 5.1), and Sikkim EQ (2011, M w = 6.9). Recently,
starting from 28th April, 2021, Assam has been shaken by more than 100 earthquakes
in only a month duration, ranging from small to large, highest (main shock) being
6.4 in Richter magnitude followed by a number of aftershocks up to magnitude 2.0
in Richter scale. These records of past earthquake data indicate the vulnerability
of this region to high magnitude earthquake; hence, seismic hazard assessment is
really essential for the future preparedness of the important structures and affected
community.
Till date, various researchers have conducted seismic hazard assessment for India,
using both the approaches deterministic seismic hazard assessment (DSHA) and
probabilistic seismic hazard assessment (PSHA), such as [1, 3–5, 8, 9, 11, 13, 16–
23]. These researchers have either studied one specific state amongst the eight north-
eastern states or the entire NE region all together. The various studies display the
hazard maps in the form of PGA and S a with an exceedance probability of 0.5, 2,
5, and 10% in 50 years for fixed return periods such as 50, 100, 475, 1000, 2475,
and 10,000 years. Assam, being the gateway for all the NE states, it must be given
special attention to study the seismic activities throughout the last decades. The
earthquake catalogue has been collected from 1897 to 2022, including instrumental
as well as pre-instrumental data, provided by various agencies such as ISC, IMD, and
USGS-NEIC. The catalogue is first unified to moment magnitude (M w ) and further
declustered before using it as the input for finding the seismicity parameters. The
region is sub-divided into eleven source zones, and the seismicity parameters for
each source zone have been estimated. Various ground motion prediction equations
(GMPE) have been used for performing PSHA. The seismic hazard curves in terms
Probabilistic Seismic Hazard Assessment for Assam, North-East India 313
of PGA values have been developed for 2% and 10% probability of exceedance in
50 years, corresponding to return periods of 475 and 2475 and years, respectively.
The spectral acceleration values (S a ) are also presented for 10% of probability of
exceedance in 50 years, for time 0.2 and 1.0 s.
2 Geological and Tectonic Setting
The NE India region has high seismicity due to its diverse geological and tectonic
setting. Assam is located between latitude 24–28° N and longitude of 89–96° E and
at an average altitude of around 170 ft above mean sea level. As per seismic zoning
code of India [12], the states in North-East (NE) region have been considered in
seismic zone V having the highest zone factor (0.36 g) in the country. Amongst the
eight north-eastern states, Assam serves as the gateway to the other seven states.
According to Geological Survey of India, Assam is broadly divided into three phys-
iographic domains: Brahmaputra Valley, Central Assam Hills, and Barak Valley
covering 78,523 km2 areas. In the NE region, mainly, three tectonic belts are present
with two convergent boundaries, viz. the Himalayan arc to the north and the Indo-
Burmese region to the east, interacting with the Assam syntaxis. The Himalayan belt
is associated with the Indian plate and at the same time under-thrusts beneath the
Eurasian plate. The seismic activity of the Indo-Burmese arc is related to the subduc-
tion of the Indian plate. Moreover, intra-plate activities also take place within this
region, which comprise Naga thrusts, the Arakan-Yoma Fold Belt in the Indo-Burma
ranges, and the Shillong plateau. These thrusts are tectonically and geologically inter-
esting, as the subducted front of Indian plate is under Tibetan plate to the north and
Burmese plate to the southeast and it is associated with thrusts and shear along the
plate boundaries. Figure 1 shows the tectonic features of NE India. For better under-
standing of the tectonic setting of the NE India region, it has been broadly divided into
four zones, such as Eastern Himalaya region, Assam valley, Indo-Burmese region,
and Shillong-Mikir Plateau [4].
2.1 Seismic Source Models
One of the basic pre-requisites to seismic hazard assessment is the seismic source
zonation of the study area. Several studies such as Sitharam and Sil [23] and Borah
et al. [6] have been performed to sub-divide the NE India region into various source
zones. The source zones are modelled based upon the parameters, such as the past
seismicity, earthquake location, and tectonic features of the region. In the present
study, the NE India region is sub-divided into eleven source zones as shown in
Fig. 2. The source zones are taken as similar to the source zonation performed by
Borah et al. [6]. The NE India region has been categorized into eleven different source
zones based on the past seismicity and spatial variation of the seismicity parameters.
314 M. Borah et al.
Fig. 1 Tectonic features of NE India
It is assumed that the source zones act as a single point source model for earthquake
occurrence, and earthquakes may arise anywhere arbitrarily near the source zone.
The source zones 1, 2, 3, and 4 constitute the Eastern Himalayan region. Some
important thrusts such as MCT, MBT, MFT, and IST are present in these four zones.
Moreover, some more significant faults, namely Mishmi thrust and Lohit thrust, are
covered by Zone 4. Zone 5 covers Dhubri and Chedrang fault. Zone 6 contains a part
of the Assam valley, Shillong Plateau, and the Mishmi hill; having some important
faults such as Dudhnoi shear, Oldham fault, Kopili fault, and Dauki fault. Zone 7
contains a part of Assam valley and the Indo-Burmese range, with a segment of Naga
thrust, Mishmi thrust, and Lohit thrust. Zone 8 comes under the Bengal basin region
with Eocene Hinge zone, and Debagram-Bogra fault. Zone 9 covers partially Bengal
basin and the Indo-Burmese range, with important faults such as Sylhet fault and a
segment of the Dauki fault. Zone 10 entails the Indo-Burmese range with a segment
of Naga thrust, a segment of Dauki Wanding fault, and Mat fault. Zone 11 is situated
to the east of Zone 10 with Shan-Sagaing fault.
3 Earthquake Catalogue
For the execution PSHA, preparation of a reliable and homogenized earthquake

catalogue is one of the basic inputs. The catalogue has been collected from the
seismological agencies such as ISC, IMD, and USGS-NEIC; from 1897 to 2022
Fig. 2 Source zonation of NE India
for the region of latitude 20–30° N and longitude 87–98° E. The magnitudes are
obtained in different scales, such as M b , M L , M s , M d , and M w . For further processing,
the catalogue is first homogenized to moment magnitude, M w using the regression
relationship proposed by Sitharam and Sil [23] for NE India as follows:
Mw = 0.862 m b + 1.034 (1)
Mw = 0.637 M L + 1.730 (2)
Mw = 0.625 Ms + 2.350 (3)
Mw = 0.742 Md + 1.565 (4)
3.1 Declustering of Catalogue
A total number of 6550 events of earthquake are obtained from where the foreshock
and the aftershock events (dependent events) are declustered from the main shocks,
316 M. Borah et al.
Fig. 3 Seismicity map of NE India
and finally, a total number of 6010 events are left in the catalogue. The declus-
tering is performed by the algorithm developed by Uhrhammer [25], in the ZMAP
programme. Figure 3 shows the seismicity map of NE India with the declustered
catalogue showing events with magnitude greater than or equal to 4.
3.2 Evaluation of Seismicity Parameters
The seismicity parameters considered for PSHA include ‘a’ and ‘b’ values, seismic
activity rate (λ), β, M max , and threshold magnitude (M C ). These parameters are
evaluated for each source zone. The two parameters (‘a’ and ‘b’ values) relate the
seismic activity rate (λ) with the magnitude of earthquake for a particular region.
Gutenberg and Richter [10] developed the regression relationship as follows:
log λ = a − bM (5)
where ‘a’ and ‘b’ are the constants of the regression relationship, λ is the mean annual
rate of exceedance of a particular magnitude, M. The values of ‘a’ and ‘b’ show the
seismicity of a region, acting as the main inputs for PSHA. The high values of these
two parameters indicate a high level of seismicity with a larger proportion of small
events. ZMAP software package developed by Wiemer [26] is used for calculating
the ‘a’, ‘b’ values and M c . The M c is the completeness magnitude, termed as the
minimum magnitude or the threshold magnitude. At this magnitude, 100% of the
earthquakes are perceived in the time scale. This value is considered for obtaining
the sub-catalogues for each source zone to evaluate the values of λ, β (2.303b), and
M max .
3.3 Maximum Magnitude (Mmax )
The information about the maximum possible earthquake magnitude (M max ) is essen-
tial for the further processing of PSHA, which is a key parameter for seismic design.
It is the maximum magnitude or upper limit that can be generated by a seismic source
zone. There are several methods available to calculate the maximum magnitude of
seismic sources. In the present study, the programme developed by Kijko [14] has
been used to calculate the maximum magnitude. The seismicity parameters obtained
from the procedures by ZMAP and Kijko [14] have been tabulated for each zone as
shown in Table 1. These parameters are further considered for performing PSHA.
4 Ground Motion Prediction Equation (GMPE)
Selecting the appropriate ground motion prediction equations (GMPE) is a crucial

step for performing PSHA. A noteworthy variation can occur between the predicted
and the actual values provided by the attenuation models for the area. These attenu-
ation models account for the seismic energies for the tectonic zones. Thus, to reduce
the uncertainties in the prediction models, as many as possible attenuation models are
taken into account for the study. In this study, three attenuation models are used for
performing PSHA. GMPE labels the seismic function (Y ) such as the peak ground
acceleration (PGA) or S a , in terms of earthquake magnitude (M) and distance (r) as:
Y = F(M, r ) (6)
where ‘M’ is the moment magnitude and ‘r’ is the distance, defined as epicentral
distance, Joyner Boor distance or the closest distance.
The various source zones in NE India have different tectonic characteristics, so
selecting the appropriate GMPE for each source zone is an important step to consider.
The NE India region is classified into three regions, namely active shallow crustal
region, active intra-plate region, and subduction zone region. The zones 1, 2, 3, and 4
are considered as active tectonic regions; whilst the zones 5, 6, 8, and 9 are considered
as active intra-plate regions; and zones 7, 10, and 11 are considered as the subduction
zones. As there is lack of availability of strong ground motions in this area, GMPEs
developed for different regions around the globe having same tectonic features are
used for this study. In this study, Chiou and Youngs [7] for the active crustal regions;
318
Table 1 Seismicity parameters

Zone M obs ‘b’ value Std. deviation of ‘b’ value Mc β COV of β λ Std. deviation of λ M max Std. deviaton of M max
Zone 1 6.8 0.68 0.03 4.1 1.57 0.06 2.745 0.531 6.90 0.27
Zone 2 6.2 1.28 0.20 4.7 2.94 0.46 0.277 0.080 6.99 0.83
Zone 3 6.5 0.82 0.04 4.4 1.89 0.09 2.726 0.567 6.57 0.26
Zone 4 7.7 0.68 0.03 4.3 1.56 0.07 3.692 0.750 7.88 0.31
Zone 5 6.8 0.70 0.05 4.0 1.62 0.11 1.235 0.293 7.08 0.38
Zone 6 8.7 1.18 0.06 4.7 2.71 0.13 3.342 0.708 9.20 0.00
Zone 7 7.1 0.98 0.05 4.7 2.26 0.11 2.996 0.606 7.26 0.30
Zone 8 6.3 1.50 0.00 4.7 3.45 0.00 0.192 0.073 6.80 0.00
Zone 9 6.8 0.73 0.03 4.0 1.68 0.08 2.733 0.583 6.94 0.29
Zone 10 7.3 1.22 0.02 4.7 2.80 0.05 15.215 2.356 7.41 0.27
Zone 11 7.6 1.25 0.09 4.7 2.87 0.21 1.354 0.329 8.10 0.00
M. Borah et al.
[2] for the active intra-plate regions; and [15] for the subduction zone regions; are
used.
5 Computation Model for PSHA
The probability of exceeding an IM intensity level, given the occurrence of a future

earthquake from a single source, can be computed as:
m max rmax
P(IM > x) = ∫ ∫ P(IM > x|m, r ) f M (m) f R (r )dr dm (7)
m min 0
where P(IM > x) is obtained from the ground motion prediction model; f M (m)
and f R (r) are the probability density functions for magnitude and source to site
distance, respectively. The integration operation adds up the conditional probabilities
of exceedance associated with all possible magnitudes and distances.
In this present study, the area source method is assumed where all the seismicity
is distributed to the area source as a single point source model, and integration is
performed all over the source area. The software ‘R-Crisis’ is used to perform the
PSHA, where the G-R distribution is considered, in terms of M min , M max , β, and λ
values, as shown in Table 1. The study area is divided into different grids of size
0.10 × 0.10, and hazard curves are generated at bedrock level where V s = 1100 m/s,
and distance of 300 km is taken from each grid. The curves are developed for ten
time periods such as 0.01, 0.1, 0.2, 0.3, 0.5, 0.7, 1.0, 2.0, 3.0, and 4.0 s. The hazard
maps are generated at bedrock level in terms of PGA, for 2% and 10% probability
of exceedance in 50 years corresponding to return periods of 475 and 2475 years,
respectively. Besides, for time period of 0.2 and 1.0 s, the hazard maps are presented
for 10% probability of exceedance in 50 years.
5.1 Results and Discussion
PSHA for Assam state of NE India is performed by dividing the entire area into
grids of size 0.1° × 0.1°, and the results are presented in the forms of seismic
hazard curves. The hazard curves are generated at bedrock level for ten different
time periods. These maps are generated for 2% and 10% exceedance probabilities
in 50 years, corresponding to return periods of 475 and 2475 years, respectively.
Figures 4 and 5 show the hazard maps in terms of PGA for 2 and 10% exceedance
probabilities. The hazard maps in terms of S a values are also presented for 0.2 and
1.0 s, for 10% probability of exceedance (Fig. 6). The PGA and S a values are shown
in Tables 2 and 3 for different return periods. Also, the PGA values for six major
cities of Assam have been presented in Table 4. A comparison is done amongst the
320 M. Borah et al.
Fig. 4 Hazard map of PGA values for 2% probability of exceedance in 50 years
obtained results and the existing PSHA studies by various researchers, as presented
in Table 5.
It is observed that the PGA values vary from 0.24 to 0.57 g, and 0.16–0.33 g; for
2% and 10% probability of exceedance, respectively. The PGA spectra developed
are generally considered for the design of the structures. The results show a signif-
icant variation with the other existing studies conducted for the Assam region. The
major cities show higher values, indicating high seismicity concentrated around the
populated areas of the state.
6 Conclusions
In this study, an attempt has been made for generating the updated seismic hazard
curves for Assam, NE India using PSHA approach. The hazard maps show a signifi-
cant variation with the existing studies available for the region. The PGA values have
been evaluated at bedrock level, for 2% and 10% exceedance probabilities in 50 years
corresponding to 475 and 2475 years of return periods, respectively. The results are
validated with the expected values as per the given codes for the region. These results
are expected to give insight for the design of the engineering structures and future
preparedness of disaster mitigation for the community. Further, site-specific studies
logic tree approach is recommended for the improvement of the seismic hazard
scenario of the region.
Fig. 5 Hazard map of PGA values for 10% probability of exceedance in 50 years
322 M. Borah et al.
Fig. 6 Hazard maps of S a values for 10% probability of exceedance in 50 years for: a t = 0.2 s
and b t = 1.0 s
Table 2 PGA values for

2% probability of exceedance in 50 years 0.24–0.57 g
obtained for Assam region
10% probability of exceedance in 50 years 0.16–0.33 g
Table 3 S a values obtained

10% probability of exceedance in t = 0.2 s 0.24–0.77 g
for Assam region
50 years t = 1.0 s 0.12–0.20 g
References
1. Ameer, A.S., Sharma, M.L., Wason, H.R., Alsinawi, S.A.: Probabilistic seismic hazard assess-
ment for Iraq using complete earthquake catalogue files. Pure Appl. Geophys. 162(5), 951–966
(2005). https://doi.org/10.1007/s00024-004-2650-y
2. Atkinson, G.M., Boore, D.M.: Earthquake ground-motion prediction equations for Eastern
North America. Bull. Seismol. Soc. Am. 96(6), 2181–2205 (2006). https://doi.org/10.1785/
0120050245
3. Bahuguna, A., Sil, A.: Comprehensive seismicity, seismic sources and seismic hazard assess-
ment of Assam, North East India. J. Earthq. Eng. 24(2), 254–297 (2020). https://doi.org/10.
1080/13632469.2018.1453405
4. Bandyopadhyay, S., Parulekar, Y.M., Sengupta, A.: Generation of seismic hazard maps for
Assam Region and incorporation of the site effects. Acta Geophys. (2022). https://doi.org/10.
1007/s11600-022-00846-z
5. Baro, O., Kumar, A., Ismail-Zadeh, A.: Seismic hazard assessment of the Shillong plateau
using a probabilistic approach. Geomat. Nat. Haz. Risk 11(1), 2210–2238 (2020). https://doi.
org/10.1080/19475705.2020.1833989
6. Borah, N., Kumar, A., Dhanotiya, R.: Seismic source zonation for NE India on the basis of past
EQs and spatial distribution of seismicity parameters. J. Seismolog. 25(6), 1483–1506 (2021).
https://doi.org/10.1007/s10950-021-10037-w
7. Chiou, B.-J., Youngs, R.R.: Update of the Chiou and Youngs NGA model for the average
horizontal component of peak ground motion and response spectra. Earthq. Spectra 30(3),
Table 4 PGA values for six major cities of Assam for 10% probability of exceedance in 50 years
Cities 2% probability of exceedance in 50 years 10% probability of exceedance in
(g) 50 years (g)
Guwahati 0.37 0.28
Tezpur 0.38 0.28
Jorhat 0.38 0.28
Dibrugarh 0.55 0.32
Silchar 0.28 0.21
Nagaon 0.38 0.28
Table 5 Comparison of the

Sitharam et al. Bahuguna and Present study
present study with the existing
[24] (g) Sil [3] (g) (g)
studies for Assam region
2% 0.4–0.65 0.44–0.77 0.24–0.57
probability of
exceedance
10% 0.2–0.36 0.16–0.33
probability of
exceedance
324 M. Borah et al.
1117–1153 (2014). https://doi.org/10.1193/072813EQS219M

8. Das, R., Sharma, M.L., Wason, H.R.: Probabilistic seismic hazard assessment for Northeast
India Region. Pure Appl. Geophys. 173(8), 2653–2670 (2016). https://doi.org/10.1007/s00024-
016-1333-9
9. Das, S., Gupta, I.D., Gupta, V.K.: A probabilistic seismic hazard analysis of Northeast India.
Earthq. Spectra 22(1), 1–27 (2006). https://doi.org/10.1193/1.2163914
34, 185–188 (1944)
11. Harbindu, A., Sharma, M.L., Kamal: Stochastic ground-motion simulation of two Himalayan
Earthquakes: seismic hazard assessment perspective. J. Seismol. 16(2), 345–69 (2012).https://
doi.org/10.1007/s10950-011-9247-6
12. IS-1893: Criteria for Earthquake Resistant Design of Structures, Part-1. Bureau of Indian
Standards, India (2016)
13. Jena, R., Pradhan, B., Naik, S.P., Alamri, A.M.: Earthquake risk assessment in NE India using
deep learning and geospatial analysis. Geosci. Front. 12(3), 101110 (2021). https://doi.org/10.
1016/j.gsf.2020.11.007
14. Kijko, A.: Estimation of the maximum earthquake magnitude, m Max. Pure Appl. Geophys.
161(8), 1655–1681 (2004). https://doi.org/10.1007/s00024-004-2531-4
15. Lin, P.-S., Lee, C.-T.: Ground-motion attenuation relationships for subduction-zone earth-
quakes in Northeastern Taiwan. Bull. Seismol. Soc. Am. 98(1), 220–240 (2008). https://doi.
org/10.1785/0120060002
16. Mahajan, A.K., Thakur, V.C., Sharma, M.L., Chauhan, M.: Probabilistic seismic hazard map
of NW Himalaya and its adjoining area, India. Nat. Hazards 53(3), 443–457 (2010). https://
doi.org/10.1007/s11069-009-9439-3
17. Nath, S.K., Raj, A., Thingbaijam, K.K.S., Kumar, A.: Ground motion synthesis and seismic
scenario in Guwahati city—a stochastic approach. Seismol. Res. Lett. 80(2), 233–242 (2009).
18. Sarmah, T., Das, S.: Earthquake vulnerability assessment for RCC buildings of Guwahati city
using rapid visual screening. Procedia Eng. 212, 214–221 (2018). https://doi.org/10.1016/j.pro
eng.2018.01.028
19. Shanker, D., Sharma, M.L.: Estimation of seismic hazard parameters for the Himalayas and
its vicinity from complete data files. Pure Appl. Geophys. 152(2), 267–279 (1998). https://doi.
org/10.1007/s000240050154
20. Sharma, M.L.: Seismic hazard in the Northern India Region. Seismol. Res. Lett. 74(2), 141–147
21. Sharma, M.L., Lindholm, C.: Earthquake hazard assessment for Dehradun, Uttarakhand, India,
including a characteristic earthquake recurrence model for the Himalaya Frontal Fault (HFF).
Pure Appl. Geophys. 169(9), 1601–1617 (2012). https://doi.org/10.1007/s00024-011-0427-7
22. Sharma, M.L., Malik, S.: Probabilistic seismic hazard analysis and estimation of spectral strong
ground motion on bed rock in North East India (2006)
23. Sitharam, T.G., Sil, A.: Comprehensive seismic hazard assessment of Tripura and Mizoram
States. J. Earth Syst. Sci. 123(4), 837–857 (2014). https://doi.org/10.1007/s12040-014-0438-8
24. Sitharam, T.G., Kolathayar, S., James, N.: Probabilistic assessment of surface level seismic
hazard in India using topographic gradient as a proxy for site condition. Geosci. Front. 6(6),
847–859 (2015). https://doi.org/10.1016/j.gsf.2014.06.002
25. Uhrhammer, R.J.E.N.: Characteristics of northern and central California seismicity. Earthq.
Notes 57, 21 (1986)
26. Wiemer, S.: A software package to analyze seismicity: ZMAP. Seismol. Res. Lett. 72, 373–382
(2001)
Effect of Randomness of Slip and Source
Time Function on Pseudo-Dynamically
Simulated Ground Motion
Characteristics
Vishal, J. P. Narayan, and L. Joshi
Abstract This paper presents the effects of randomization of slip and the param-
eters of source time function on the pseudo-dynamically simulated ground motion
characteristics. In the case of numerical simulations, the radiation of seismic energy
from the rupture plane as per Brune’s model as well as to avoid the coherency effects
is a challenging job for the simulators. The randomization of slip, rise time, and
peak time of the source time function and the rupture arrival time, as well as the
incorporation of fault roughness and damage zone, play important roles in seismic
energy release from the rupture plane as well as in the reduction of coherency effects
on the high-frequency seismic radiations. Inversion of earthquake data or statistical
analysis of dynamic rupture simulations is used to estimate the slip distribution.
The statistical approach assumes that the earthquake slip follows a random distribu-
tion on the fault plane. The simulation of pseudo-dynamic ground motion has been
carried out using a fourth-order accurate staggered-grid time-domain 3D finite sdif-
ference method. The ground motions are simulated taking ten different slip patterns
for a hypothetical strike-slip M w 6.0 earthquake. In addition, for each slip pattern, a
stochastic perturbation in the parameters of the source time function is introduced.
The simulated results have been analyzed based upon some important parameters
such as arias intensity, peak ground acceleration, peak ground velocity, and peak
ground displacement. Considerable variation in the computed values for the afore-
said parameters is obtained with the change of slip patterns and parameters of the
STF. A good match of the computed average pseudo-spectral acceleration (PSA)
using the simulated ground motion with that obtained using NGA West2 GMPEs is
obtained in the frequency range 0.1–5.0 Hz and at an epicentral distance of 11 km.
Keywords Pseudo-dynamic rupture · Finite difference method · 3D ground

motion simulation
Vishal (B) · J. P. Narayan · L. Joshi

e-mail: vishal@eq.iitr.ac.in
https://doi.org/10.1007/978-981-99-1459-3_26
326 Vishal et al.
1 Introduction
The physics-based numerical simulation of strong ground motion (SGM) character-

istics is very important for site-specific SGM prediction, cost-effective earthquake-
resistant designs, and assessment of vulnerability and risk. The numerical strong
ground motion can be simulated using either the kinematic approach or the dynamic
approach. The dynamic approach is not often used for the SGM simulations because
of the lack of the required rheological and physical parameters around the source
volume as well as the need for very large computational memory and speed [1–3].
However, it is possible to simulate ground motion using a kinematic approach, which
is easier to implement and requires less computing power. Different kinematic dislo-
cation models have been used in the past, such as the Haskell model [4], Brune’s
model [5], and asperity and barrier models. The main difference between dynamic
and kinematic simulation models of earthquake rupture is whether or not the physics
of the fault breaking is taken into account. The former takes into account the physics
of the fault rupturing, while the latter ties the earthquake to a certain slip distribution.
In kinematic models, the rupture is described by a fault slip defined as a function
of position and time. In dynamic models, the fault slip is determined dynamically,
taking into account the size of the earthquake, how its focal mechanism works, and
the geology of the area.
Guatteri et al. [6] proposed pseudo-dynamic rupture implementation using param-
eters like a slip, rise time, and rupture arrival times derived from a wide range
of dynamic rupture simulations [7, 8]. The pioneer research works [9, 10] explain
how to improve broadband SGM using a kinematic approach. This is a mix of a
deterministic approach for low-frequency (≤ 1.0 Hz) simulations and a stochastic
method for high-frequency (> 1.0 Hz) simulations. Incorporating a so-called “pseudo-
dynamic rupture” into the numerical grid is the key to getting rid of the need for a
stochastic method for high-frequency ground motion [6, 7, 11, 12]. The characteri-
zation and inter-correlations of kinematic parameters in the case of pseudo-dynamic
rupture are directed by the rules produced from the statistical analysis of suites
of dynamic rupture simulations. These rules were constructed in order to lead the
process of pseudo-dynamic rupture. The incorporation of stochastic crustal velocity
perturbations [12], near-fault damage zone [13, 14], and the effects of fault rough-
ness [15] is some of the significant recent developments that have been made in the
pursuit of the reduction of coherency in the high-frequency radiations. In most of
the studies incorporating the pseudo-dynamic rupture for simulation of earthquake
ground motions [12, 16], the spatial and temporal distribution of slip is considered
random. This randomness of slip affects simulated ground motion. Therefore, the
primary objective of our research work is estimation of effect of randomization of
slip on the different characteristics of earthquake ground motion. The important
characteristics like peak ground acceleration, peak ground velocity, peak ground
displacement, and arias intensity are considered.
Effect of Randomness of Slip and Source Time Function … 327
2 Methodology Adopted
We have used the fourth-order staggered-grid viscoelastic time-domain 3D finite

difference code written by [17]. The capability of this code to simulate the SGM as per
physics of the rupture propagation was enhanced by [18] by means of implementation
of pseudo-dynamic rupture, wherein there is stochastic perturbation of slip, rise time,
and rupture arrival times correlated with slip distribution, peak time of the STF, and
fault roughness [7, 9, 12, 15, 19]. First, we computed the seismic moment release
for the postulated earthquake (M w 6.0) using Eq. 1.
log10 Mo
Mw = − 10.71 (1)
1.5
where M w is the moment magnitude, M o is seismic moment is dyne-cm. In the next
step, the dimensions of faults length and width in the strike and down dip directions
are calculated using the scaling laws given by [20]. For the postulated magnitude M w
= 6.0, the seismic moment, length, and width of the rupture come out to be 1.122
* 1025 dyne-cm, 14 km, and 11 km, respectively. The average slip comes out to be
0.21 m.
Once the initial model parameters have been defined, the next step in the simu-
lation process is prescribing a random spatial and temporal distribution of slip and
rupture arrival times at all the point sources distributed over the rupture plane. The
randomization of spatial distribution of slip on rupture plane is achieved using the
methodology given by [21]. The temporal evolution of slip/slip velocity at each point
source in the fault plane is assumed to be known and is described using an STF having
rise time partially correlated with the slip at that particular point source. The STF
used in this study is a Kostrov-like pulse proposed by [19], as given below:
⎧
⎪
⎪ C N 0.7 − 0.7 cos πτ1t + 0.6 sin 0. 5π t
0 ≤ τ < τ1
⎪
⎨ τ
1
ṡ(t) = C N 1.0 − 0.7 cos πτ1t + 0.3 cos π (t−τ τ2
1)
τ1 ≤ τ < 2τ2 (2)
⎪
⎪
⎪
⎩ C N 0.3 + 0.3 cos π (t−τ )
τ2
1
2τ1 ≤ τ < τ
where ṡ(t) is the slip velocity, CN = π/(1.4π τ1 + 1.2τ1 + 0.3π τ2 ), τ is average

rise time, τ1 is peak time and τ2 = τ − τ1 .
The average rise time comes out to be 0.355 s. The rupture arrival times at each
point source are calculated using [22] methodology after assigning the hypocenter
location within the rupture plane and with a background rupture velocity equal to
80% of the local S-wave velocity to which perturbations are further added. The
random distributions of parameters such as slip, rupture arrival time, and rise time
are done using an amalgamation of the procedures devised by [7, 19]. The stochastic
perturbations to the rupture arrival time and rise time partially correlated with the
slip distribution reduce the coherency effects on the high-frequency radiations. Near
fault damage zone and the effects of fault roughness have been incorporated to reduce
328 Vishal et al.
the coherency effect in the radiation of high frequency (> 1.0 Hz). The two cases of
peak times have been considered. In first case, the peak time has been kept constant
at 0.13 times of average rise time, while perturbation in peak time has been added
for the second case.
The methodology developed by [23] has been used to represent the source in
a deterministic manner. According to this theory, the whole rupture plane may be
broken down into multiple sub-faults, and each of them can be thought of as an
independent point source. On the rupture plane, there are a total of 61,600 point
sources that have been taken into consideration, and the distance that separates any
two consecutive point sources in either the horizontal or vertical plane is 50 m.
By utilizing the STF and stress tensor components, a specific point source may be
introduced into the numerical grid that constitutes the velocity-stress scheme.
STF(t)∗Δt ∗ Mx x
(σx x )i+
n
1
, j+ 1 ,k+ 1
= (σx x )i+
n
1
, j+ 1 ,k+ 1
− (3)
2 2 2 2 2 2 V
n STF(t)∗Δt ∗ M yy
(σ yy )i+
n
1
, j+ 1 ,k+ 1
= σ yy i+ 21 , j+ 21 ,k+ 21
− (4)
2 2 2 V
STF(t)∗Δt ∗ Mzz
(σzz )i+
n
1
, j+ 1 ,k+ 1
= (σzz )i+
n
1
, j+ 1 ,k+ 1
− (5)
2 2 2 2 2 2 V
n STF(t)∗Δt ∗ Mx y
(σx y )i,n j,k+ 1 = σx y i, j,k+ 21
− (6)
2 V
n STF(t)∗Δt ∗ Mzy
(σzy )i+
n
1
, j,k
= σzy i+ 21 , j,k
− (7)
2 V
STF(t)∗Δt ∗ Mx z
(σx z )i,n j+ 1 ,k = (σx z )i,n j+ 1 ,k − (8)
2 2 V
where
Mx x = −m o (sin δ cos γ sin 2φs + sin 2δ sin γ sin 2φs ) (9)
Mx y = m o (sin δ cos γ cos 2φs + 0.5 sin 2δ sin γ sin 2φs ) = M yx (10)
Mx z = −m o (cos δ cos γ cos φs + cos 2δ sin γ sin φs ) = Mzx (11)
M yy = m o (sin δ cos γ sin 2φs − sin 2δ sin γ cos 2φs ) (12)
M yz = −m o (cos δ cos γ sin φs − cos 2δ sin γ cos φs ) = Mzy (13)
Mzz = m o (sin 2δ sin γ ) (14)

Table 1 Rheological
Thickness (km) V S (m/s) V P (m/s) Density (kg/m3 )
parameters of considered
layered earth model 0–2 1500 3123 2000
2–7 2000 3742 2200
7–11 2600 4500 2400
> 11 3200 5543 2600
where δ, γ , and φs represent dip, strike, and rake angle taken as 90, 90, and 160
degree, respectively. mo is the seismic moment for each sub-fault. σ ij is the respec-
tive stress components, STF(t) is the value of the STF at time instant “t”, Δt is
the time step adopted in the FD simulation. M ij is the moment tensor component as
described earlier, and V is the volume of the finite difference grid used in the FD simu-
lation. The essential wave propagation effects, which include viscoelastic damping
and a heterogeneous velocity structure, are simulated by a three-dimensional finite
difference algorithm, the specifics of which are outlined in detail in [17]. The model
is surrounded on three sides by absorbing boundary conditions [24], and the free
surface effect is simulated on the uppermost layer using a technique called VGR-
stress imaging [25]. Further in both the horizontal directions grid spacing was taken
as 50 m to get reliable results up to 5 Hz. Finally, a time step of 0.005 s was chosen to
keep in the mind the stability criteria for fourth-order finite difference simulations.
Table 1 provides the velocity model used in the simulation.
3 Ground Motion Simulations
In order to estimate the effect of randomization of the spatial distribution of slip,

ten different PBS1-PBS10 slip models, rise time, and rupture arrival times on the
rupture plane have been considered. Figure 1a–j shows the slip patterns, rise times,
and rupture arrival times for the PBS1-PBS10 slip models. A considerable change in
the rise time distribution can be inferred since it is correlated with the slip distribution.
The rupture for the considered strike-slip earthquake extend from a depth of 1.0–
12 km. The focus of the earthquake is considered at a depth of 6.5 km and at distance of
7.0 km from the left-edge of the rupture. The right panels of Fig. 1 depict the variation
of rupture arrival time accordingly. All the horizontal distances are measured with
respect to the epicenter of the earthquake.
All the three components of the ground motion are simulated on a circular array
with its radius as 11.0 km, and center of this array is collocated with the epicenter
(Fig. 2). There are 881 receiver points on the circular array at a horizontal distance
of 50 m. The R1 and R441 receiver points are located on the eastern and western
edge of a line passing along the diameter of the circular array, which is oriented in
the EW-direction. So, R2 to R440 receiver points are on the circular array and north
of this diameter oriented in EW-direction.
330 Vishal et al.
Fig. 1 a–j depicts the spatial variation of slip (left panel), rise time (middle panel), and rupture
arrival time (right panels) on the rupture plane of the considered PBS1-PBS10 slip models
Fig. 1 (continued)
332 Vishal et al.
Fig. 2 Sketch for a circular array around the epicenter at a distance of 11.0 km with 881 receiver
points at a horizontal distance of 50 m (Note: rupture plane is oriented in the EW-direction)
The left, middle, and right panels of Fig. 3 depict the NS-, EW-, and vertical
components of the simulated ground motion at 12 recording stations (R1–R441) at a
horizontal distance of 2 km (at an interval of 15° in an azimuthal range − 90° to 90°).
The effects of rupture directivity can be observed in the NS-component at the R1 and
R441 recording points. A decrease of directivity effects can be inferred on rest of the
receiver points. Also, there is more amplitude in the EW-component as compared
to NS-component for receiver’s perpendicular to the rupture plane. There is a lot
variation in ground motion amplitude and duration with the change of azimuth of the
receiver points. The vertical components have also relatively comparable amplitude
to those in the two horizontal components due to recordings made in the epicentral
zone. The duration of ground motion is of the order of 12–15 s at different receiver
points. Similarly, the strong ground motions are simulated on the same circular array
for all the considered PBS1-PBS10 slip models.
Fig. 3 NS-(left panel), EW-(middle panel), and vertical (right panel) components of the simulated
ground motion on the circular array corresponding to the PBS1 slip model
Fig. 4 Comparison of
simulated PSA with the
average of PSA obtained
using NGA West2 GMPEs.
Gray lines show the PSA
computed for each R1–R441
using ROTD50 of NS- and
EW-components
3.1 Validation of Pseudo-Dynamic Rupture Implementation
The NGA West2 GMPEs are well constrained for epicentral distances less than
20–30 km for the moderate earthquakes (M6.0–M6.5) due to an adequate number
of earthquake records [26–29]. We computed pseudo-spectral acceleration (PSA)
with 5% damping for both the horizontal components at each receiver points with
epicentral distances 11 km after band-pass filtering in a frequency bandwidth of 0.1–
5.0 Hz. Figure 4 shows the median of pseudo-spectral acceleration (PSA) obtained
using average of NGA West2 GMPEs (black color) with standard deviation (red and
blue) in a period range of 0.2–10 s. The gray lines indicate the PSA for all the R1–
R441, while green line indicates the average PSA of 441 receivers. The median of
PSA using NS- and EW-components of each receiver is obtained using ROTD50 as
described by [30]. The analysis of Fig. 4 shows that the average PSA for all receivers
matches very well with the results derived using NGA West2 GMPEs. It validates
the efficiency of code in simulating the strong ground motion. The good matching of
the obtained average PGA as 0.13 g with that computed (0.11 g) using NGA West2
GMPEs further validates the capability of the enhanced 3D FD code by [18] for the
broadband ground motion.
3.2 Effect of Randomization of Slip Pattern
Figure 5 shows the variation of peak ground acceleration (PGA) in the NS-, EW-, and
vertical components in the case of used slipping PBS1-PBS10 models. The obtained
range of PGA is 0.05–0.33 g, 0.07–0.44 g, 0.06–0.35 g, 0.06–0.35 g, 0.08–0.51 g,
0.06–0.36 g, 0.07–0.36 g, 0.07–0.45 g, 0.07–0.48 g, and 0.06–0.38 g in the case
of PBS1-PBS10 slip models, respectively. Similarly, the obtained average PGA is
0.13 g, 0.13 g, 0.11 g, 0.11 g, 0.14 g, 0.11 g, 0.12 g, 0.12 g, 0.15 g, and 0.11 g in
the case of PBS1-PBS10 slip models, respectively. Although, the computed PGA at
different recording stations is highly variable with the change in the slip pattern, but,
334 Vishal et al.
Fig. 5 Variation of PGA in the NS-component (left), EW-component (middle), and vertical
component (right) with receiver positions
in most of the considered slip models, there is good match of the average PGA with
that computed using NGA West2 GMPEs. However, the obtained up to 30% larger
average PGA in some of the slip models may be due to the rupture directivity effect.
Similarly, Figs. 6 and 7 show the variation of peak ground velocity (PGV), peak
ground displacement (PGD), Arias intensity for NS-, EW-, and vertical components
for PBS1–PBS10 slip models, respectively. The analysis of these results shows that
the spatial variation PGV and PGD also match with the variation of PGA. The
obtained very large PGV of the order of 0.26–0.40 m/s in the NS-component at
receivers R1 and R441 in the different slip models as compared to that computed as
0.04 m/s using NGA West2 GMPEs also validates the efficacy of numerical modeling
to simulate the rupture directivity effects as per the physics of the rupture propaga-
tion. Figure 8 depicts the variation of Arias intensity for the NS-, EW-, and vertical
components for the PBS1-PBS10 slip model models, respectively. The obtained Arias
intensity is very high at receivers R1 and R441 due to rupture directivity effects in
the NS-component.
Fig. 6 Variation of PGV in the NS-component (left), EW-component (middle), and vertical
Fig. 7 Variation of PGD in the NS-component (left), EW-component (middle), and vertical
Fig. 8 Diagram showing the variation of Arias intensity for NS-component (left), EW component
(middle), and vertical component (right) with receivers
4 Conclusions
A good match of the computed average PSA in a period range of 0.2–10 s. with
NGA West2 GMPEs for a hypothetical strike-slip earthquake of M w = 6.0 reveals
the efficiency of 3D FD code to appropriately radiate seismic energy in a broadband
and avoids the occurrence of coherency effect on the high-frequency radiation [12,
15]. The obtained very large PGV of the order of 0.26–0.40 m/s in the NS-component
at receivers R1 and R441 in different slip models as compared to that computed as
0.04 m/s using NGA West2 GMPEs also validates the efficacy of numerical modeling
to simulate the rupture directivity effects as per the physics of the rupture propagation.
It also validates the implementation of strike-slip rupture in this study. A lot of
variation in the ground motion characteristics as well as PGA, PGV, PGD, and
the Arias intensity was observed with the change of slip pattern. For example, the
obtained range of PGA was 0.05–0.33 g, 0.07–0.44 g, 0.06–0.35 g, 0.06–0.35 g,
0.08–0.51 g, 0.06–0.36 g, 0.07–0.36 g, 0.07–0.45 g, 0.07–0.48 g, and 0.06–0.38 g
for the PBS1-PBS10 slip models, respectively. Similarly, the obtained average PGA
was 0.13 g, 0.13 g, 0.11 g, 0.11 g, 0.14 g, 0.11 g, 0.12 g, 0.12 g, 0.15 g, and 0.11 g in the
case of PBS1-PBS10 slip models, respectively. Similarly, a range of variations in the
336 Vishal et al.
PGV, PGD, and the Arias intensity was obtained with the change of slip pattern. The
obtained average PGA in the different slip models is either comparable or somewhat
larger than that obtained using NGA West2 GMPEs (0.11 g). This finding reflects
that slip pattern plays an important role in the spatial variation of ground motion
characteristics and the engineering parameters.
References
1. Day, S.M.: Three dimensional simulation of spontaneous rupture: the effect of non-uniform
prestress. Bull. Seismol. Soc. Am. 72(6), 1881–1902 (1982)
2. Oglesby, D.D., Day, S.M.: Stochastic fault stress: implications for fault dynamics and ground
motion. Bull. Seismol. Soc. Am. 92(8), 3006–3021 (2002). https://doi.org/10.1785/012001
0249
3. Ripperger, S., Mai, P.M., Ampuero, J.P.: Variability of near-field ground motion from dynamic
earthquake rupture simulations. Bull. Seismol. Soc. Am. 98(3), 1207–1228 (2008). https://doi.
org/10.1785/0120070076
4. Haskell, A.: Total energy and energy spectral density of elastic wave radiation from propagating
faults. Bull. Seismol. Soc. Am. 54(6), 1811–1841 (1964)
5. Brune, J.N.: Tectonic stress and the spectra of seismic shear waves from earthquakes. J.
Geophys. Res. 75(26), 4997–5009 (1970)
6. Guatteri, M., Mai, P.M., Beroza, G.C.: A pseudo-dynamic approximation to dynamic rupture
models for strong ground motion prediction. Bull. Seismol. Soc. Am. 94(6), 2051–2063 (2004).
https://doi.org/10.1785/0120040037
7. Schmedes, J., Archuleta, R.J., Lavallée, D.: Correlation of earthquake source parameters
inferred from dynamic rupture simulations. J. Geophys. Res. Solid Earth 115(3), 1–12 (2010).
https://doi.org/10.1029/2009JB006689
8. Song, S.G., Somerville, P.: Physics-based earthquake source characterization and modeling
with geostatistics. Bull. Seismol. Soc. Am. 100(2), 482–496 (2010). https://doi.org/10.1785/
0120090134
9. Graves, R.W., Pitarka, A.: Broadband ground-motion simulation using a hybrid approach. Bull.
Seismol. Soc. Am. 100(5A), 2095–2123 (2010). https://doi.org/10.1785/0120100057
10. Graves, R., Pitarka, A.: Refinements to the Graves and Pitarka (2010) broadband ground-motion
simulation method. Seismol. Res. Lett. 86(1), 75–80 (2015). https://doi.org/10.1785/022014
0101
11. Mena, B., Dalguer, L.A., Mai, P.M.: Pseudodynamic source characterization for strike-slip
faulting including stress heterogeneity and super-shear ruptures. Bull. Seismol. Soc. Am.
102(4), 1654–1680 (2012). https://doi.org/10.1785/0120110111
12. Graves, R., Pitarka, A.: Kinematic ground-motion simulations on rough faults including effects
of 3D stochastic velocity perturbations. Bull. Seismol. Soc. Am. 106(5), 2136–2153 (2016).
https://doi.org/10.1785/0120160088
13. Cochran, E.S., Li, Y.G., Shearer, P.M., Barbot, S., Fialko, Y., Vidale, J.E.: Seismic and geodetic
evidence for extensive, long-lived fault damage zones. Geology 37(4), 315–318 (2009). https://
doi.org/10.1130/G25306A.1
14. Ben-Zion, Y., et al.: Basic data features and results from a spatially dense seismic array on the
San Jacinto fault zone. Geophys. J. Int. 202(1), 370–380 (2015). https://doi.org/10.1093/gji/
ggv142
15. Mai, P.M., Galis, M., Thingbaijam, K.K.S., Vyas, J.C., Dunham, E.M.: Accounting for fault
roughness in pseudo-dynamic ground-motion simulations. Pure Appl. Geophys. 174(9), 3419–
3450 (2017). https://doi.org/10.1007/s00024-017-1536-8
16. Pitarka, A., Akinci, A., De Gori, P., Buttinelli, M.: Deterministic 3D ground-motion simula-
tions (0–5 Hz) and surface topography effects of the 30 October 2016 Mw 6.5 Norcia, Italy,
Earthquake. Bull. Seismol. Soc. Am. 112(1), 262–286 (2022). https://doi.org/10.1785/012021
0133
(2014). https://doi.org/10.1007/s10596-014-9442-y
18. Joshi, L.: Quanti cation of Site City Interaction Effects on Responses of Buildings and Basin
Under Realistic Earthquake Loading for Development of Economic Smart City (2021)
19. Liu, P., Archuleta, R.J., Hartzell, S.H.: Prediction of broadband ground-motion time histories:
hybrid low/high-frequency method with correlated random source parameters. Bull. Seismol.
Soc. Am. 96(6), 2118–2130 (2006). https://doi.org/10.1785/0120060036
20. Thingbaijam, K.K.S., Mai, P.M., Goda, K.: New empirical earthquake source-scaling laws.
Bull. Seismol. Soc. Am. 107(5), 2225–2246 (2017). https://doi.org/10.1785/0120170017
21. Mai, P.M., Beroza, G.C.: A spatial random field model to characterize complexity in earthquake
slip. J. Geophys. Res. Solid Earth 107(B11), ESE 10-1–ESE 10-21 (2002). https://doi.org/10.
1029/2001jb000588
22. Afnimar, Koketsu, K.: Finite difference traveltime calculation for head waves travelling along
an irregular interface. Geophys. J. Int. 143(3), 729–734 (2000). https://doi.org/10.1046/j.1365-
246X.2000.00269.x
23. Hartzell, S.H., Heaton, T.H.: Waveform data for the fault rupture history of the 1979. Bull.
Seismol. Soc. Am. 73(6), 1553–1583 (1983)
24. Kumar, S., Narayan, J.P.: Absorbing boundary conditions in a fourth-order accurate SH-wave
staggered grid finite difference algorithm. Acta Geophys. 56(4), 1090–1108 (2008). https://
doi.org/10.2478/s11600-008-0043-9
25. Narayan, J.P., Kumar, S.: A fourth order accurate SH-wave staggered grid finite-difference
algorithm with variable grid size and VGR-stress imaging technique. Pure Appl. Geophys.
165(2), 271–294 (2008). https://doi.org/10.1007/s00024-008-0298-8
26. Boore, D.M., Stewart, J.P., Seyhan, E., Atkinson, G.M.: NGA-West2 equations for predicting
PGA, PGV, and 5% damped PSA for shallow crustal earthquakes. Earthq. Spectra 30(3),
27. Campbell, K.W., Bozorgnia, Y.: Campbell-bozorgnia NGA-West2 horizontal ground motion
model for active tectonic domains. Earthq. Spectra 30, 1087–1115 (2014). https://doi.org/10.
4231/D3MS3K235
28. Chiou, B.S.J., Youngs, R.R.: Update of the Chiou and Youngs NGA model for the average
horizontal component of peak ground motion and response spectra. Earthq. Spectra 30(3),
29. Idrissa, I.M.: An NGA-West2 empirical model for estimating the horizontal spectral values
generated by shallow crustal earthquakes. Earthq. Spectra 30(3), 1155–1177 (2014). https://
doi.org/10.1193/070613EQS195M
30. Boore, D.M.: Orientation-independent, nongeometric-mean measures of seismic intensity from
two horizontal components of motion. Bull. Seismol. Soc. Am. 100(4), 1830–1835 (2010).
https://doi.org/10.1785/0120090400
Quantification of Ridge-Weathering
Effects on the Simulated Ground Motion
Characteristics Across 2D and 3D
Topography Models
Vishal and J. P. Narayan
Abstract This paper quantifies the impact of ridge-weathering on the simulation

of ground motion across 2D and 3D topographical models. The models are excited
with plane wave-fronts of Gabor wavelet. The seismic responses of a 3D ellipsoidal
ridge topography having shape ratio 1.0 and one of its cross-Sect. (2D model) are
simulated for different thickness and velocity of weathered layer. The SH- and SV-
waves responses of a 2D cross-section of 3D topography are computed to under-
stand the effect of dimensionality on the amplification pattern. The analysis of simu-
lated responses reveals very large spectral amplifications and ASA in the case of
3D topography as compared to 2D. An increase of weathering effect on ground
motion is inferred with an increase of weathering-thickness and a decrease of weath-
ering velocity. A considerable variation of topography amplification with elevation is
obtained. Finally, it is concluded that the shape of ridge topography and rheological
parameters of the weathering layer should be taken in to account for the quantification
of topographical effects on the ground motion characteristics.
Keywords 3D topography effect · Finite-difference method · Azimuth effect ·

Weathering effect
1 Introduction
Increasing constructions on higher topographies have prompted the focus of seismol-

ogists and earthquake engineers on the measurement of impacts of surface topography
on the ground motion characteristics. In addition, essential infrastructure design must
explicitly account for the amount of variation in ground motion characteristics caused
by topography. Records of ground motion, studies of damage patterns from earlier
earthquakes, and numerical modeling have all shed light on the topography amplifica-
tions. For example, exceptionally high accelerations of 1.25 g at the Pacoima Dam [1]
Vishal (B) · J. P. Narayan

e-mail: vishal@eq.iitr.ac.in
https://doi.org/10.1007/978-981-99-1459-3_27
340 Vishal and J. P. Narayan
and 1.78 g at Tarzana hill were reported during the 1971 San Fernando earthquake and
the 1994 Northridge earthquake, respectively [2]. In addition during the 1985 Chile
Earthquake [3], the 1987 Superstition Hill Earthquake [4], the 1987 Whitter Narrows
Earthquake [5], 2015 Gorkha Nepal Earthquake [6], and 2016 Norcia Italy M w 6.5
Earthquake [7], considerable damages were reported on the elevated topographies.
It has been noticed that in the frequency range of earthquake engineering interest,
numerically anticipated topographic amplifications rarely exceeded 3.0 [8–10]. In
addition, earthquake data show that topographic amplifications of more than 10 have
occurred [8, 10–12]. Possible reasons for this difference include using 2D models
instead of 3D models, homogeneous ridges instead of heterogeneous ridges, ground
motion records near the base of the hill as a reference site for calculating spectral
amplifications, and the simulations not taking into account the topography of the area.
Also, it has been observed that there are no thick soil deposits in the hilly region. But,
there is a thin weathered layer. This layer of weathered rock acts as a low velocity
layer and induces variation in ground motion as of topography-induced amplification
and de-amplification and also the fundamental frequency of topography [11–14].
The primary objective of this study is to contribute to a better understanding of the
impact of thickness and velocity of the weathering layer on 2D and 3D topography
on the earthquake ground motion. The simulation of 2D (a cross-section of the same
3D topography) and 3D seismic responses of topography with various thickness and
velocity of weathered layer has been carried out. Attempts have been made to evaluate
the effect of model dimension by comparing 2D and 3D topographic amplification
patterns.
2 Methodology Adopted
The 2D and 3D seismic responses of the topographical models have been simu-
lated using fourth-order staggered-grid viscoelastic SH-wave, P-SV-wave, and 3D
finite-difference (FD) programs written by Narayan and his co-researchers [15–
17], respectively. These programs are effective enough to provide frequency and
phase-dependent damping in the time-domain simulations utilizing the GMB-EK
rheological model [18, 19]. Material independent anelastic function was employed
because it is more appropriate for the air-rock discontinuities in the FD grid [17,
19]. Table 1 provides the P- and S-wave velocities and quality factors at reference
frequency 1.0 Hz and density (ρ) for viscoelastic homogeneous rock and air. As a
free surface boundary condition, an enhanced vacuum formulation proposed by [20]
is adopted. In this improved vacuum formulation, the harmonic and arithmetic means
of unrelaxed moduli and density, respectively, are employed. The sponge absorbing
boundary condition was put on the model’s bottom, side edges, and faces to stop
edge reflections [21, 22].
Quantification of Ridge-Weathering Effects on the Simulated Ground … 341
Table 1 Rheological characteristics for the viscoelastic rock and air

Materials Velocities and quality factors at F R = 1.0 Hz Density
V P (m/s) V S (m/s) QP QS ρ (kg/m3 )
Air 320 0 32 0 1.29
Rock 2600 1500 260 150 2000
3 Geometry of Models
In order to study the impacts of weathering on seismic responses of 2D and 3D ridge

topography, 10 different models of 3D ellipsoidal topography have been considered.
The SH- and SV-wave responses of a 2D plane passing through the NS-axis of 3D
topography have also been computed and analyzed. Figure 1 depicts the sketches
for 3D ellipsoidal ridge models with base width 1000 m and height 500 m (shape
ratio = 1). The velocity of the weathered layer has been varied between 500 m/s to
1100 m/s, with an interval of 150 m/s. The thickness of the weathered layer is taken
as 5 m (WERM1-WERM5) and 10 m (WERM6-WERM10) for the 3D ellipsoidal
topography. The details of these models are given in Tables 2 and 3. The thickness
of the weathering layer at the base is assumed to be twice the thickness at the top
of the topography. The effect of 3D topography model without weathering is also
estimated and named as ERM model.
Fig. 1 Sketch for ellipsoidal ridge model along with NS- and EW-receiver arrays and point sources
at a depth of 1535 m to generate a plane S-wave front propagating vertically upward
Table 2 Velocity, density, and thickness of considered weathered ridge models WERM1-WERM5
Name of models WERM1 WERM2 WERM3 WERM4 WERM5
V s (weathering) (m/s) 500 650 800 950 1100
Density (kg/m3 ) 1600 1660 1720 1780 1840
Thickness (m) 5 5 5 5 5
Table 3 Velocity, density, and thickness of considered weathered ridge models WERM6-WERM10
Name of models WERM6 WERM7 WERM8 WERM9 WERM10
V s (weathering) (m/s) 500 650 800 950 1100
Density (kg/m3 ) 1600 1660 1720 1780 1840
Thickness (m) 10 10 10 10 10
The axis of the ridge is used as a reference point for measuring horizontal
distances. The vertical distances are measured with respect to the free surface level.
In both, the horizontal directions grid spacing was taken as 5 m. However, in the
vertical direction, grid spacing was taken 5 m from the top of model to a depth of
45 m below the free surface and 10 m thereafter. To prevent stability problems, a time
step of 0.0005 s was chosen. A plane wave front of S-wave was generated in the FD
grid employing 1000 point sources placed 5 m apart in a horizontal plane at a depth
of 1535 m. A particular point source for a particular polarization of the S-wave was
generated using shear stress in the form of Gabor wavelet. For example, to generate
a S-wave with NS-polarization, shear stress σ X Z was used to generate all the 1000
point sources. The dominant frequency of Gabor wavelet was taken as 4.0 Hz with a
frequency bandwidth of 0.1–10 Hz. All the three components of the seismic response
on the NS- and EW-arrays extending from 850 m left to 850 m right of the ridge
axis along the flanks of the topography and with equidistant 341 receiver points (5 m
horizontally apart) were simulated (Fig. 1). The seismic response of homogeneous
half-space model on a rectangular array with 5 m spacing between two consecutive
receiver points was also simulated to use it as reference ground motion to quantify
the topography effects.
4 Seismic Response of 2D Cross-Section of 3D Topography
4.1 SH-Wave Response
In this section, the effect of 2D elliptical cross-section of the 3D ellipsoidal ridge

models on the SH-wave characteristics has been calculated. Figure 2a, b shows the
change of average spectral amplification (ASA) with varying velocity and thickness
Fig. 2 Diagram showing variation of ASA in the case of 2D equivalent to a cross-section of a

WERM1-WERM5, b WERM6-WERM10 models in low-frequency range (0.1–1.5 Hz) (left) and
high-frequency range (1.5–10.0 Hz) (right) with SH-wave as input motion
of the weathered layer in low-frequency (0.1–1.5) and high-frequency ranges (1.5–

10 Hz). It is clear that in both low- and high-frequency range, ASA is increasing
with decrease in the velocity and an increase in thickness of the weathered layer. The
effect of weathering is more in the high-frequency range where the wavelength of
wave is comparable or less than the width of base of topography.
4.2 SV-Wave Response
The variation of ASA for the horizontal and vertical components of the SV-wave
has been shown in Figs. 3 and 4, respectively. It has been observed that in the
high-frequency range, there is an increase in ASA with increase in thickness and
decrease in velocity of weathering. However, in the low-frequency range, there is
Fig. 3 Variation of ASA of the horizontal component for the 2D equivalent of a WERM1-WERM5,
b WERM6-WERM10 models in low-frequency range (0.1–1.5 Hz) and high-frequency range (1.5–
0.0 Hz) for SV-wave as input motion
no substantial changes in ASA. It is interesting to note that the amplification in the

vertical component is more than the horizontal component in high-frequency range
near the base of the 2D ridge.
5 Effect of 3D Topography
In this section, the effect of 3D weathered ellipsoidal topography on the ground

motion characteristics has been calculated. The WERM1-WERM10 models with
specification as shown in Table 3 have been considered. The models have been
excited with the plane S-wave front having NS-polarity. Figure 5 shows the variation
of ASA for the NS-component of EW-array of in the case of WERM1-WERM10
models. This is very much similar to that of the SH-wave response. ASA is showing an
Fig. 4 Variation of ASA of the vertical component of the 2D equivalent of a WERM1-WERM5, b

WERM6-WERM10 models in low-frequency range (0.1–1.5 Hz) and high-frequency range (1.5–
0.0 Hz) for SV-wave as input motion
increase with decrease in velocity of weathered layer along the slope of topography.
Further decrease in ASA for both low- and high-frequency range has been observed
up to 200 m from top of topography. Overall, a correlation between the thickness of
the weathering layer and the ASA has been detected.
Figures 6 and 7 show the variation of ASA for the NS-component and vertical
component of receivers along the NS-array of on the WERM1-WERM10 models.
This is very much similar to that of the SV-wave response. The same pattern as in
the NS-component of the EW-array has been observed. But, the amplitude is less in
this case. The increase in ASA with decrease in velocity of weathered layer has been
observed along slope of topography. In addition, there is a significant rise in ASA in
the vertical component on the NS-array receivers, especially around the base of the
ridges. A comparison of Figs. 5 and 6 shows the notable variation in ASA from top
toward the base. Although, it is same at the top of ridge in both the cases, but it is
somewhat larger in Fig. 6.
Fig. 5 Variation of ASA for the NS-component of the EW-array of a WERM1-WERM5, b

WERM6-WERM10 models in low-frequency range (0.1–1.5 Hz) and high-frequency range
(1.5–10.0 Hz) for plane S-wave with NS-polarity
6 Comparison of 2D and 3D Weathering Effect
In this section, a comparison of 2D and 3D average weathering effect (AWE) due to

weathering effect in high-frequency range is presented. AWE is obtained by dividing
the ASA obtained for weathered models (WERM1-WERM10) to without weathered
model (ERM).
Figure 8a, b show the obtained AWE for the SH-wave response of an elliptical
ridge and the NS-component of an elliptical ridge response along the EW-array,
respectively. The weathering is aggravating the topographic effects from the base to
the top of the 2D ridge. In contrast to this, it is aggravating topography effects only
near the base of the 3D topography. Figure 9a, b show the AWE for the horizontal
components of the SV-wave response of a 2D ridge and the NS-component of the
response of the 3D ridge along the NS-array, respectively. Similarly, in Fig. 10a, b the
AWE for the vertical component of the SV-wave response of the 2D topography
Fig. 6 Variation of ASA for the NS-component of the NS-array of a WERM1-WERM5, b

WERM6-WERM10 models in low-frequency range (0.1–1.5 Hz) and high-frequency range
(1.5–10.0 Hz) for plane S-wave with NS-polarity
and the vertical component of the response of the 3D topography along the NS-
array. The analysis of these figures also reveals that weathering is aggravating the
topography effects from the base to the top of the 2D ridge, and in contrast to this, it
is aggravating 3D topography effects only near the base. Further, the AWE increases
with the decrease in velocity of the weathering layer and increases with the increase
in thickness. Furthermore, in the case of 3D topography, a de-amplification near the
top of the ridge can be inferred. It is important to note that the AWE is greater in
vertical component than in the horizontal component.
Fig. 7 ASA obtained for the vertical component of NS-array of a WERM1-WERM5, b WERM6-
WERM10 models in low-frequency range (0.1–1.5 Hz) and high-frequency range (1.5–10.0 Hz)
for plane S-wave with NS-polarity
7 Conclusions
The analysis of the simulated results reveals that the dimensionality of the topog-
raphy plays an important role in the amplitude amplification and de-amplification
patterns along the flanks of the ridge. The obtained ASA in the case of 3D ellipsoidal
topography is more as compared to 2D elliptical topography. Further, the increase of
thickness and decrease of velocity of the weathered layer increases the topography
Fig. 8 Comparison of AWE obtained for a SH-wave response, b NS-component of 3D response

along EW-array for the WERM1-WERM5 (left) and WERM6-WERM10 (right) models
Fig. 9 Comparison of AWE obtained for a horizontal component of SV-wave response, b NS-
component of 3D response along NS-array for the WERM1-WERM5 (left) and WERM6-WERM10
(right) models
Fig. 9 (continued)
Fig. 10 Diagram showing the comparison of AWE for a vertical component of SV-wave response,
b vertical component of 3D response along NS-array for the WERM1-WERM5 (left) and WERM6-
WERM10 (right) models
effects. In the low-frequency range, where the wavelength of the wave exceeds the
width of the topography, the effect of weathering is not very prominent. While, in
the case of high frequency range, the effect of weathering is significant. SH-wave
horizontal component amplification is greater than SV-wave horizontal component
amplification due to mode conversion in SV-wave response. The computed AWE
reflects that weathering aggravates the topographic amplification throughout the 2D
topography. But, it aggravates the topographic amplification only near the base of
the 3D topography. An increase of AWE with a decrease of weathering velocity is
obtained. It is concluded that the consideration of weathering effects is very important
for an accurate estimation of the topographic amplification.
References
1. Trifunac, M.: Scattering of plane SH waves by a SemiCylindrical rigid foundation seismic

energy distribution during soil-structure interaction view project. Earthq. Eng. Struct. Dyn. 1,
267–281 (1973)
2. Spudich, P., Hellweg, M., Lee, W. H. K.: Directional topographic site response at Tarzana
observed in aftershocks of the 1994 Northridge, California, earthquake: implications for
mainshock motions. Bull. Seismol. Soc. Am. 86(1 SUPPL. B) (1996)
3. Celebi, M.: Topographical and geological amplifications determined from strong-motion and
aftershock records of the 3 march 1985 chile earthquake (1987)
4. Çelebi, M.: Topographical and geological amplification: case studies and engineering impli-
cations. Struct. Saf. 10(1–3), 199–217 (1991). https://doi.org/10.1016/0167-4730(91)900
15-2
5. Kawase, H., Aki, K.: Topography effect at the critical SV-wave incidence: possible explanation
of damage pattern by the Whittier narrows, California, earthquake of 1 October 1987. Bull.
Seismol. Soc. Am. 80(1) (1990)
6. Wang, F., et al.: Effects of topographic and geological features on building damage caused by
2015.4.25 Mw7.8 Gorkha earthquake in Nepal: a preliminary investigation report. Geoenviron.
Disasters 3(1) (2016). https://doi.org/10.1186/s40677-016-0040-2
7. Pitarka, A., Akinci, A., De Gori, P., Buttinelli, M.: Deterministic 3D ground-motion simula-
tions (0–5 Hz) and surface topography effects of the 30 October 2016 Mw 6.5 Norcia, Italy,
Earthquake. Bull. Seismol. Soc. Am. 112(1), 262–286 (2022). https://doi.org/10.1785/012021
0133
8. Geli, L., Bard, P.-Y.: The effect of topography on earthquake ground motion: a review and new
results (1988)
9. Poursartip, B.: Topographic Amplification of Seismic Motion. University of Texas at Austin
(2017)
10. Massa, M., Barani, S., Lovati, S.: Overview of topographic effects based on experimental
observations: meaning, causes and possible interpretations. Geophys. J. Int. 197(3), 1537–1550
(2014). https://doi.org/10.1093/gji/ggt341
11. Lee, S.J., Komatitsch, D., Huang, B.S., Tromp, J.: Effects of topography on seismic-wave
propagation: an example from Northern Taiwan. Bull. Seismol. Soc. Am. 99(1), 314–325
(2009). https://doi.org/10.1785/0120080020
12. Grelle, G., et al.: Topographic effects observed at Amatrice hill during the 2016–2017 Central
Italy seismic sequence. Earthq. Eng. Eng. Vib. 20(1), 63–78 (2021). https://doi.org/10.1007/
s11803-021-2005-z
13. Narayan, J.P., Prasad Rao, P.V.: Two and half dimensional simulation of ridge effects on the
ground motion characteristics. Pure Appl. Geophys. 160(8), 1557–1571 (2003). https://doi.
org/10.1007/s00024-003-2360-x
14. Narayan, J.P., Kumar, V.: A numerical study of effects of ridge-weathering and ridge-shape-
ratio on the ground motion characteristics. J. Seismol. 19(1), 83–104 (2014). https://doi.org/
10.1007/s10950-014-9452-1
15. Narayan, J.P., Kumar, V.: A fourth-order accurate finite-difference program for the simulation
of SH-wave propagation in heterogeneous viscoelastic medium. Geofizika 30(2) (2013)
16. Narayan, J.P., Kumar, V.: Study of combined effects of sediment rheology and basement
focusing in an unbounded viscoelastic medium using P-SV-wave finite-difference modelling.
Acta Geophys. 62(6), 1214–1245 (2014). https://doi.org/10.2478/s11600-013-0199-9
(2014). https://doi.org/10.1007/s10596-014-9442-y
18. Emmerich, H., Korn, M.: Incorporation of attenuation into time-domain computations of
seismic wave fields. Geophysics 52(9), 1252–1264 (1987). https://doi.org/10.1190/1.1442386
19. Kristek, J., Moczo, P.: Seismic-wave propagation in viscoelastic media with material disconti-
nuities: a 3D fourth-order staggered-grid finite-difference modeling. Bull. Seismol. Soc. Am.
93(5), 2273–2280 (2003). https://doi.org/10.1785/0120030023
20. Zeng, C., Xia, J., Miller, R.D., Tsoflias, G.P.: An improved vacuum formulation for 2D
finite-difference modeling of Rayleigh waves including surface topography and internal
discontinuities. Geophysics 77(1) (2012). https://doi.org/10.1190/geo2011-0067.1
21. Israeli, M., Orszag, S.A.: Approximation of radiation boundary conditions. J. Comput. Phys.
41(1), 115–135 (1981). https://doi.org/10.1016/0021-9991(81)90082-6
22. Narayan, J.P., Kumar, S.: A fourth order accurate SH-wave staggered grid finite-difference
algorithm with variable grid size and VGR-stress imaging technique. Pure Appl. Geophys.
165(2), 271–294 (2008). https://doi.org/10.1007/s00024-008-0298-8
Developing a Comprehensive Historical
Tsunami Database for the Indian Ocean
Nazeel Sabah and Daya Shanker
Abstract The Indian Ocean is one of the most active tsunamigenic zones on the
earth. The destructive tsunami of 26th December, 2004 and 28th September, 2018
are standing testimonies of the tsunamicity in the region. The occurrence of two
strong tsunami in consecutive decades is a driving force behind the compilation of
a comprehensive historical catalog for the region. Unlike the other tsunamigenic
regions of the earth, there exist no historical tsunami catalog in a concretized form
for the Indian Ocean. The importance of this study is that it gives impetus toward
developing a probabilistic tsunami hazard analysis methodology for long-term risk
reduction against tsunami. For this study, the geographic area between 45° S–27° N
and 24° E–120° E has been considered as the extent of the Indian Ocean. All the
significant tsunami events from 416 AD till 2021 have been collected from various
online catalogs globally. A few regional catalogs have also been referred to. All
magnitudes of tsunamigenic earthquakes are homogenized into moment magnitude
scale (M W ) using global correlation equations. Statistical regression tools like stan-
dard linear regression (SLR), inverse standard linear regression (ISR), and orthogonal
standard regression (OSR) are used to developed correlation between tsunami magni-
tude (M t ) and earthquake magnitude (M w ). The missing data on tsunami intensity of
several events are calculated on the Soloviev scale using already existing relations.
An updated catalog for tsunami events of the Indian Ocean is hence developed which
fills several missing values in the already existing catalog. The catalog so compiled
is presented on a webpage.
Keywords Tsunami catalog · Magnitude homogenization · Tsunami intensity ·

Tsunami magnitude
N. Sabah (B) · D. Shanker

Indian Institute of Technology, Roorkee, India
e-mail: nsabah@eq.iitr.ac.in
https://doi.org/10.1007/978-981-99-1459-3_28
354 N. Sabah and D. Shanker
1 Introduction
The Indian Ocean region has experienced around 200 notable tsunami from 416
A.D. till date killing several thousands [1]. The Indian Ocean region bound between
45° S–27° N and 24° E–120° E [2] is one of the most active tsunamigenic zones
on earth contributing to about 5% of global tsunamicity. There exists no well docu-
mented catalog for the Indian Ocean region unlike the other tsunamigenic zones.
Also, the existing global catalogs of [1, 3] have a number of missing entries for the
Indian Ocean region. This paper deals with improving the quality of the existing
catalogs by completing the missing values wherever possible. 194 tsunami events
were fetched from [1, 3, 4] for the Indian Ocean region between 416 A.D. till date for
the Indian Ocean region. Figure 1 presents a color-coded plot epicentral locations of
tsunamigenic earthquakes between 416 A.D. till 2021. The magnitudes of tsunami-
genic earthquakes are homogenized using the existing global correlation equations
of [5, 6]. There exists no correlation equations between real magnitude-type tsunami
scale (M t ) as proposed by [7, 8] and moment magnitude (M W ) of tsunamigenic earth-
quakes. Here, we develop a correlation between M t and M W using statistical regres-
sion tools, viz. standard linear regression (SLR), inverse standard linear regression
(ISR), and orthogonal standard regression (OSR). All these models are compared,
and the best suited regression model is chosen based on coefficient of determination
(R2 ).
Fig. 1 Historical locations of tsunamigenic earthquakes in the Indian Ocean between 416 A.D. till
2021 with magnitudes ranging from 4.5 to 9.2 MW. Developed using [9] and data obtained from
[1, 3]
Developing a Comprehensive Historical Tsunami Database … 355
There exist several intensity-magnitude relationships for earthquake events. But

there is only a little work done to link the tsunami intensity with earthquake magni-
tude. Tsunami magnitude is expressed generally on Soloviev-Imamura scale [10].
In this study, we find the missing Soloviev-Imamura tsunami intensity of tsunami
events in the existing global catalog. The formula suggested by [11] is used to predict
tsunami intensity from the moment magnitude of earthquake (M W ). In this study, the
existing catalog is improved by filling in the missing values either using the devel-
oped conversion equations or the existing predictive relations. The updated catalog
so compiled is updated on a webpage for public access. The catalog can be viewed on
the following link: https://sites.google.com/eq.iitr.ac.in/comprehensive-tsunami-io.
2 Magnitude Homogenisation
There are several global and regional equations to convert magnitude from one
scale to another. Different historical tsunami catalog presents the magnitude data
in different units. Since, moment magnitude (M W ) is the most scientific measure
of earthquake magnitude which depends on the seismic moment generated at the
source, all the magnitudes of tsunamigenic earthquakes are converted into moment
magnitude. The two most widely used magnitude homogenization equations are the
ones suggested by [5, 12]. In this study, we use the relations suggested by [12] for
magnitude homogenization since it is more recent one and is developed by incor-
porating a larger number of data into regression models (Eq. 1). Also, it is found to
have a better accuracy than other homogenization equations.
MW = 1.06(±0.0002)M S − 0.38(±0.006)
6.2 ≤ M S ≤ 8.4, R 2 = 0.89 (1)
Of all the 194, tsunami events extracted from various catalogs for the Indian
Ocean, only 63 events had magnitudes presented on MW scale. Using Eq. 1, the
missing moment magnitude data for 112 events have been calculated and updated
on the Webpage mentioned in the introduction.
3 Calculation of Tsunami Magnitude (Mt )
Abe [7, 8] had developed a realistic scale for tsunami magnitude, which depends
on the instrumental records of tsunami height. The readings of coastal tide-gages
were used to read the maximum and average amplitudes of tsunami waves [10]. The
moment magnitude of tsunami is calculated mathematically as per Eq. 2.
Mt = a log H + b log R + D (2)

where,
H Highest amplitude of tsunami wave (in meters) as obtained from tide
gage.
R Distance to epicenter (in km).
a, b, and D Constants determined to make M t scale closely related to M W scale.
It may not be practically possible to ascertain these parameters, particularly “H”
due to the absence of tide gage readings at site. Due to this limitation, out of 194 events
obtained from various catalogs, only about 4 events had magnitudes expressed on M t
scale. Here, we develop correlation equations to determine the tsunami magnitude
(M t ) from earthquake moment magnitude (M W ). For this purpose, we use statistical
regression tools SLR, ISR, and OSR as already mentioned in the introduction. The
detailed mechanism and procedure used for this procedure have been adopted from
[13]. The detailed procedure for standard linear regression (SLR) has been referenced
from [14–17]. Draper and Smith [17] explains the detailed procedure for carrying out
inverse standard regression (ISR). The procedure for orthogonal standard regression
(OSR) has been adopted from the work done by [18–21].
To successfully develop a correlation between two parameters, one dependent and
the other independent, we need a good amount of statistically reliable data, but as per
the available online catalogs, only four events have their tsunami magnitude presented
as Mt . Hence, it is difficult to develop a good correlation between M t and M W by
considering just the data for the Indian Ocean. To get rid of this practical difficulty, we
collect the worldwide data of tsunami events from [1, 3, 4] and develop a correlation
between M t and M W . We develop this global correlation using 173 values of M t and
M W extracted from the global database. The Indian Ocean region considered in this
study, being a subset of the global database, the developed correlation equations fit
well for the Indian Ocean region as well. Figures 2, 3, and 4 show the correlation
equations developed between M t and M W for the Indian Ocean region using SLR,
ISR, and OSR regression methods.
The correlation equations are presented in terms of the slope and intercepts, with
the errors within brackets. The developed correlation equations using three regression
methods are compared based on their coefficients of determination (R2 ) since it is the
best statistical metric for the goodness of any correlation for any scientific domain
[22]. In this study, we assume a linear correlation between M t and M W, and hence,
no attempt has been made to develop a bilinear correlation between the variables.
Table 1 presents the summary of the correlation equations developed for the
study area using 3 correlation methods along with the errors. It also compares the
coefficients of determinations of the developed correlation equations.
From Table 1, it can be inferred that the correlation equations developed using
orthogonal standard regression have the highest value of coefficient of determination.
Hence, OSR is the best regression tool for correlation between M t and M W . The
missing values of tsunami magnitude (M t ) of tsunami events in the Indian ocean
using the relation M t = 0.96 (± 0.0299) M W + 0.37 (± 0.2262) are completed and
added to the updated catalog as mentioned in the link presented in the introduction.
This way, the tsunami magnitude (M t ) of 112 events have been estimated.
Mt = 0.89 (± 0.0274) M + 0.87 (± 0.207)

W
R2 = 0.862
Fig. 2 M t versus M W correlation developed using standard linear regression (SLR)
Mt = 1.04 (± 0.0296) M - 0.21 (± 0.207)

W
R2 = 0.862
Fig. 3 M t versus M W correlation developed using standard linear regression (ISR)
Figure 5 presents a comparison of the correlations developed between M t and

M W for the Indian Ocean using SLR, ISR, and OSR.
Mt = 0.96 (± 0.0299) M + 0.37 (± 0.2262)

W
R2 = 0.9283
Fig. 4 M t versus M W correlation developed using standard linear regression (OSR)
Table 1 Summary of correlation equations developed

Indian ocean region
Regression type Equation R2
SLR M t = 0.89 (± 0.0274) M W + 0.87 (± 0.207) 0.862
ISR M t = 1.04 (± 0.0296) M W − 0.21 (± 0.207) 0.862
OSR M t = 0.96 (± 0.0299) M W + 0.37 (± 0.2262) 0.928
4 Tsunami Intensity-Magnitude Relationship
With the advent of instrumental means of earthquake and tsunami observation, some
researchers tried means to link the tsunami run-up heights to the source magnitude of
the causative earthquake. The tsunami intensity scale adopted for most studies is the
Soloviev-Imamura tsunami scale which is developed using the average values of run-
up at the coastal region nearest to the site. The source magnitude scale of the causative
earthquake event is either M S or M W . Chubarov and Gusiakov [11] developed a
formula to deduce tsunami magnitude from the earthquake source magnitude. The
formula is presented as Eq. 3.
I = 3.55MW − 27.1 (3)
Figure 6 presents the predicted tsunami intensity of 684 tsunami events between
416 A.D. till date using the formula suggested by [11].
Fig. 5 Comparison of the developed correlation equations between M t and M W
Fig. 6 Tsunami intensity versus moment magnitude of 684 tsunami worldwide from 416 A.D. till
date
Figure 6 shows that the tsunami intensity M t and moment magnitude M W of the
source earthquake show a very little dependency. But one can note a general trend of
an increasing tsunami intensity with an increase in moment magnitude of the source
earthquake. Relying on this trend, we use the tsunami prediction equation of [11]
as expressed in Eq. 3 to roughly calculate the values of the missing tsunami events
(112 tsunami events) in the existing online catalogs for the Indian Ocean region. The
tsunami magnitude is computed and uploaded on the Webpage as mentioned in the
introduction part.
5 Conclusions
The present work has attempted to update the existing tsunami catalog for the Indian
Ocean region by filling in the missing values or by deducing a relationship to esti-
mate/predict the same. Wherever possible, existing relationships have been used to
predict the unknown value of parameters (like M w ) for historical tsunami of the Indian
Ocean region. 112 missing values of moment magnitude (M W ) which were missing
in the existing catalogs were filled by using the existing global relations to convert
surface-wave magnitude (M S ) to moment magnitude (M W ). This way magnitude
homogenization has been performed on the tsunami catalog of the Indian Ocean.
A relationship has been developed between tsunami magnitude (M t ) and moment
magnitude (M W ) of the earthquake source for the Indian Ocean region by regressing
173 globally observed values of M t and M W , obtained from the existing historical
tsunami catalogs. This way 112 missing values of tsunami magnitude were added to
the updated tsunami catalog of the Indian Ocean. It is also observed that orthogonal
standard regression (OSR) is the best regression tool for developing a relationship
between tsunami and earthquake magnitude for the Indian Ocean region as OSR
provided the highest value of coefficient of determination (R2 ). Also, 112 missing
values of tsunami intensity were added to the updated catalog using the existing
tsunami intensity prediction equation suggested by [11]. Figure 7 shows the intensity
of tsunami across the years as per the update tsunami catalog. Figure 8 presents a
histogram of tsunami intensity as per the update catalog for time intervals of 50 years.
The updated tsunami catalog for the Indian Ocean region can be found on the
following link: https://sites.google.com/eq.iitr.ac.in/comprehensive-tsunami-io.
Fig. 7 Tsunami intensity across years as per the updated catalog. The size of circle is in proportion
to the earthquake magnitude in M w scale
Fig. 8 Histogram of tsunami occurrence in the Indian Ocean between 1600 and 2020 at intervals
of 50 years as per the updated tsunami catalog. Dark tone indicates the total number of tsunami
events. Whereas light tone shows the number of strong and damaging tsunami with I > 1
References
1. National Geophysical Data Center/World Data Service (NGDC/WDS): Global Historical

Tsunami Database. In: National Geophysical Data Center, NOAA. p. Boulder, CO, USA,
2015. https://doi.org/10.7289/V5PN93H7
2. Gusiakov, V.K.: Historical tsunami catalog for the Indian ocean region. In: Proceedings of the
22nd International Tsunami Symposium, 2005, p. 268
3. Gusiakov, V.K.: Historical tsunami database for the pacific, 47 B.C to present. In: Novosibirsk
Tsunami Laboratory (NTL), 2005. Accessed 08 Nov 2018 [online]. Available http://tsun.sscc.
ru/htdbpac/
4. Ekström, G., Nettles, M., Dziewoński, A.M.: The global CMT project 2004–2010: centroid-
moment tensors for 13,017 earthquakes. Phys. Earth Planet. Inter. 200–201, 1–9 (2012). https://
doi.org/10.1016/J.PEPI.2012.04.002
5. Scordilis, E.M.: Empirical global relations converting MS and mb to moment magnitude. J.
Seismol. 10(2), 225–236 (2006). https://doi.org/10.1007/s10950-006-9012-4
6. Das, R., Wason, H.R., Sharma, M.L.: Magnitude conversion to unified moment magnitude
using orthogonal regression relation. J. Asian Earth Sci. 50, 44–51 (2012). https://doi.org/10.
1016/j.jseaes.2012.01.014
7. Abe, K.: Size of great earthquakes of 1837–1974 inferred from tsunami data. J. Geophys. Res.
Solid Earth 84(B4), 1561–1568 (1979). https://doi.org/10.1029/JB084iB04p01561
8. Abe, K.: Physical size of tsunamigenic earthquakes of the northwestern Pacific. Phys. Earth
Planet. Inter. 27(3), 194–205 (1981). https://doi.org/10.1016/0031-9201(81)90016-9
9. Wessel, P., et al.: The generic mapping tools version 6. Geochem. Geophys. Geosyst. 20(11),
5556–5564 (2019). https://doi.org/10.1029/2019GC008515
10. Gusiakov, V.K.: Tsunami history—recorded V. K. Gusiakov. Most, November, pp. 1–30 (2007)
11. Chubarov, L.B., Gusiakov, V.K.: Tsunamis and earthquake mechanism in the island arc region.
Sci. Tsunami Hazards 3(1), 3–21 (1985)
12. Das, R., Wason, H.R., Sharma, M.L.: Global regression relations for conversion of surface
wave and body wave magnitudes to moment magnitude. Nat. Hazards 59(2), 801–810 (2011).
https://doi.org/10.1007/s11069-011-9796-6
13. Davis, J.C.: Statistics and Data Analysis in Geology, 3rd edn. John Wiley and Sons, Kansas
(2002)
14. Bindi, D., Spallarossa, D., Eva, C., Cattaneo, M.: Local and duration magnitudes in North-
western Italy, and seismic moment versus magnitude relationships. Bull. Seismol. Soc. Am.
95, 592–604 (2005)
15. Chen, K.-P., Ben Tsai, Y.: A catalog of Taiwan earthquakes (190P_2006) with homogenized
Mw magnitudes. Bull. Seismol. Soc. Am. 98, 483–489 (2008)
16. Yadav, R.B.S., Bormann, P., Rastogi, B.K., Das, M.C., Chopra, S.: A Homogeneous and
complete earthquake catalog for northeast India and the adjoining region. Seismol. Res. Lett.
80(4), 609–627 (2009). https://doi.org/10.1785/gssrl.80.4.609
17. Draper, N.R., Smith, H.: Applied Regression Analysis, vol. 326. John Wiley & Sons (1998)
18. Carroll, R.J., Ruppert, D.: The use and misuse of orthogonal regression in linear errors-in-
variables models. Am. Stat. 50(1), 1–6 (1996). https://doi.org/10.1080/00031305.1996.104
73533
19. Fuller, W.A.: Measurement Error Models. New York (1987)
20. Kendall, M.G., Stuart, A.: The Advanced Theory of Statistics, vol. 2, no. 1. New York (1961)
21. Stefanski, L.A.: Measurement error models. J. Am. Stat. Assoc. 95(452), 1353–1358 (2000).
https://doi.org/10.1080/01621459.2000.10474347
22. Chicco, D., Warrens, M.J., Jurman, G.: The coefficient of determination R-squared is more
informative than SMAPE, MAE, MAPE, MSE and RMSE in regression analysis evaluation.
PeerJ Comput. Sci. 7, 1–24 (2021). https://doi.org/10.7717/PEERJ-CS.623
Seismically Induced Landslide Hazard
Analyses for a Road Corridor
in the Lower Himalayas
A. Tyagi , R. R. Nath , M. L. Sharma , and J. Das
Abstract Seismically induced landslide hazard assessment for important road

networks in the hilly regions constitutes a significant component of planning and
management of road infrastructures. Such multi-hazard assessment, though compli-
cated, carries even more significance for seismically active mountains like the
Himalayas. However, traditional landslide hazard zonation (LHZ) practice generally
lacks in incorporating seismic factors. An endeavour has been made in the present
study to carry out seismically induced LHZ mapping for 85 km stretches of Kalka-
Shimla highway (NH-5) in the lower Himalayan belt under a scenario earthquake
condition. Eight different landslide preparatory factors, viz. lithology, slope angle,
aspect, elevation profile, distance form fault, distance from drainage, distance from
road, and land-use-land-cover patterns are integrated with the landslide-triggering
seismic factors in a Geographical Information System (GIS). The seismic factor is
included in terms of peak ground acceleration (PGA) parameters generated for a
10% exceedance probability in 50 years using probabilistic seismic hazard assess-
ment (PSHA). The study develops a heuristic model of landslide hazard assessment
coalescing semi-quantitative and bivariate frameworks to generate the seismically
The presentation of material and details in maps used in this chapter does not imply the expression
of any opinion whatsoever on the part of the Publishers or Author concerning the legal status of
any country, area or territory or of its authorities, or concerning the delimitation of its borders. The
depiction and use of boundaries, geographic names and related data shown on maps and included
in lists, tables, documents, and databases in this chapter are not warranted to be error free nor do
they necessarily imply official endorsement or acceptance by the Publisher or Author.
A. Tyagi · M. L. Sharma · J. Das

Indian Institute of Technology, Roorkee 247667, India
e-mail: atyagi1@eq.iitr.ac.in
M. L. Sharma
J. Das
e-mail: josodhir.das@eq.iitr.ac.in
R. R. Nath (B)
Adani Institute of Infrastructure Engineering, Ahmedabad, Gujarat 382421, India
e-mail: riturajnath14@gmail.com
https://doi.org/10.1007/978-981-99-1459-3_29
364 A. Tyagi et al.
induced LHZ map. The prepared LHZ map demarcates more than 40% of the study
area as the zones of high to very high landslide hazard under the scenario earth-
quake. The prediction accuracy of the proposed method is estimated as 80% for the
study. Based on the findings, the study recommends that the seismic element be taken
into consideration for a more realistic evaluation of the current landslide threats in
seismically active mountain ranges like the Himalayas.
Keywords Seismically induced landslide hazard zonation · Earthquakes ·

Probabilistic seismic hazard assessment · Analytical hierarchy process (AHP) ·
Frequency ratio (FR) method
1 Introduction
Past experiences imply earthquakes as one of the major landslide-triggering factors

in the Himalayas [1–3]. Seismically induced landslides in mountainous terrain have
reported to cause over 70% of all earthquakes-related fatalities that were not directly
brought on by ground shaking [4], and more than forty thousand seismically induced
landslide fatalities were reported between 2004 and 2010 [5]. Data showed that road
corridors in general have higher propensity to failure from landslides due to toe
cutting [6–8] and require a robust assessment of the prevailing as well as prospective
landslide hazard for proper planning, maintenance, and management of road infras-
tructures [9, 10]. In this context, inclusion of seismic factor(s) in landslide hazard
zonation (LHZ) models is important for seismically active Himalayas as the govern-
ment has now invested significant resources in developing the road infrastructure in
the trans-Himalayan belt. Yet, only few studies [1–3, 11] considered seismic indi-
cators in LHZ mapping of the Himalayas, and there is a requirement of even more
studies for the lower Himalayas considering the increased volume of rapid devel-
opment and road widening projects in this region. An endeavour has been made in
this study to integrate landslide-triggering seismic factor with the landslide control-
ling factors using a heuristic approach for an important road corridor in the lower
Himalayas. The resultant seismically induced LHZ maps for this important highway-
corridor will be of practical importance for disaster preparedness and sustainable
development.
One of the major challenges in creating seismically generated landslide hazard
zonation (LHZ) maps is the selection of a well-known technique. Different statis-
tical techniques that are frequently used to determine the susceptibility of landslides
typically fall short in incorporating seismic indicators [1, 12]. This is mainly due to
scarcity of records of seismically induced landslides, which may further be attributed
to the rare occurrence of strong ground motions. A statistical technique produced
from an inventory of earthquake-induced landslides (developed for a single seismic
event) is also often insufficient for a diverse tectonic environment due to regional
and distinctive predispositions. These issues make studying seismically generated
landslide hazard zones more difficult. Different researchers have been trying to
Seismically Induced Landslide Hazard Analyses for a Road Corridor … 365
address these issues by including peak ground acceleration (PGA) parameter of

a scenario earthquake generated through probabilistic seismic hazard assessment
(PSHA) as a landslide-triggering seismic factor in the traditional bivariate models of
LHZ mapping [1, 2, 11, 13, 14]. In this context, this paper presents a case study on
the use of probabilistically generated PGA (for a scenario earthquake with a return
period of 475 years) as a seismic factor in preparing seismically induced LHZ map
for a part of National Highway-5 (NH-5) India (connecting two important establish-
ments of lower Himalayas: Kalka and Shimla in Himachal Pradesh) using bivariate
framework of LHZ mapping.
2 Study Area
A stretch of about 85 km of the NH-5 starting at Kalka and ending at Shimla, India
with a buffer of 1 km, is selected as the study area (Fig. 1). The development processes
such as urbanization, road construction, and other civil infrastructures have rapidly
been established in the study area in the last two decades. About 64 landslides have
been identified for the study area using Google Earth historical images and field
reconnaissance surveys.
Fig. 1 Study area map displaying the map of India and the study area buffer for a part of NH-5
366 A. Tyagi et al.
3 Methodology
3.1 Probabilistic Seismic Hazard Assessment
For the study area, seven seismogenic sources are identified based on the hetero-
geneity of their seismicity and geological attributes. The widely used Gutenberg–
Richter (G-R) relationship is used to define the annual rate of seismicity. Compilation
and treatment of the earthquake catalogue is carried out as per the standard practice
of probabilistic seismic hazard assessment (PSHA). The NGA-WEST-1 attenuation
relationship [15] is observed to be the most appropriate attenuation relationship,
which is used to generate the PGA parameter of a scenario earthquake with 10%
exceedance probability in 50 years (corresponding to the return period of 475 years).
It is to be noted that a detailed discussion on the PSHA methodology is not provided
here considering the lack of both space and scope.
3.2 Generation of Landslide Potential Index (LPI)
For the present study, eight landslide preparatory factors are identified, which are then
integrated with the PGA parameter generated through PSHA in a GIS environment.
As the study envisages to integrate two sets of temporarily variable parameters, it
adopts a heuristic approach of LSZ modelling. For that purpose, the widely accepted
map combination method [1, 16] is used where all the landslide causative parame-
ters (both preparatory and triggering factors) are included as thematic layers. Each
layer is further categorized into subclasses for establishing intra-class heterogeneity.
Each thematic class is assigned a rank (Ri ), and its subclasses are assigned a weight
(W i ) based on the perceived influence on the overall landslide susceptibility, terrain
characteristics, and the professional judgement of the investigator. Availability of the
pertinent dataset also plays a major role. The landslide potential index (LPI) for each
pixel is produced by numerically integrating the specified rankings and weights in the
GIS environment. LPI [1, 16] indicates the likelihood of landslide occurrence based
on the prevalent causative factors in an area. Greater landslide hazard susceptibility
is indicated by a higher LPI rating and vice versa. Mathematically, it may be defined
as,

n
m
LPI = Ri × Wi, j (1)
i=1 j=1
where Ri denotes the rank of the ith thematic layer of landslide causative factor and
W i,j denotes the weight of the jth thematic subclass of the ith layer. The LPI values
are then consolidated and classified in GIS environment to generate the final LSZ
map of the study area.
The subjective nature of how different thematic layers and their subclasses are
ranked and weighted is the main drawback of the map combination approach. The
study adopts the semi-quantitative analytical hierarchy process (AHP) to reduce the
subjectivity of the rank assignment process [1]. By generating the consistency ratio
(CR) of the resulting decision matrix, AHP employs pair-wise comparisons to assess
the consistency of the decision made (ranks assigned in this scenario). CR is defined
as the ratio of the consistency index (CI) to the random consistency index (RI).
Mathematically,
λmax − N
CI = (2)
N −1
CR = CI/RI (3)
where λmax is the largest normalized principal eigenvector and N is order of the
decision matrix. Values of RI are predefined for different matrices orders [17]. If
CR ≤ 10%, the decisions made (i.e. the ranks assigned) are consistent, and if it
not so, then the process should be repeated. The AHP scale is carefully chosen
from 1 to 9 where 1 signifies equal importance; 3 signifies moderate importance;
5 signifies strong importance; 7 signifies very strong importance, and 9 signifies
extreme importance between a pair of thematic layers. The values of 2, 4, 6, and 8
signify intermediate importance. The same scale is further extended for assignment
of weights to the subclasses of various thematic layers.
The weights of different subclasses of all the thematic layers are assigned based
on the frequency ratio (FR) analyses with respect to the prepared landslide database.
The FR is a bivariate statistical model of LHZ mapping which correlates the ratio
of the landslide numbers in a particular subclass to the total landslide numbers in a
thematic layer with the ratio of the area of that subclass to the total thematic area (in
terms of pixel counts). It is mathematically expressed as
Number of landslides in subclass/Total landslides in the thematic layer

FR =
No. of pixels in the subclass/Total no. of pixels in the thematic layer
(4)
Based on the computed FR values, the weights of the subclasses are assigned in the
same AHP scale which is used for assignment of ranks of thematic layers. Therefore,
the assigned ranks and weights are not based on the investigators’ perception or expe-
rience but are derived objectively. Thus, the present study reduces the subjectivity
of the map combination method which can be considered as a distinct advantage of
the study.
368 A. Tyagi et al.
Table 1 Details about the data used in the preparation of thematic maps
Data Format Source
Landslide points Vector Filed surveys/Google Earth platform
DEM & its Derivatives (Slope, aspect etc.) Raster ALOS PALSAR (https://search.asf.ala
ska.edu/)
LULC Raster 10-m resolution Global land cover 2017
(https://doi.org/10.1016/j.scib.2019.
03.002.)
Faults Vector Seismotectonic Atlas of India and its
environ/GSI/Bhukosh
Lithology Vector Ancillary data/Field surveys
Roads Vector Ancillary data
4 Preparation of Thematic Layers of Landslide Controlling

Parameters
In this study, eight different landslide preparatory factors, viz. lithology, slope angle,
aspect, elevation profile, distance form fault, distance from drainage, distance from
road, and land-use-land-cover patterns are integrated with the landslide-triggering
seismic factors in a GIS environment. The seismic factor is included in terms of peak
ground acceleration (PGA) parameters generated for a 10% exceedance probability
in 50 years using PSHA. The data required for this study have been obtained from the
satellite imageries, Survey of India (SOI) toposheets, field surveys, and Geological
Survey of India (GSI) practical sheets. A 12.5-m ALOS PALSAR high-resolution
digital elevation model (DEM) was used in this study. Thematic maps for all the
DEM derivatives were prepared in GIS environment. All data used for the study are
resampled at a scale of 1:12,500 corresponding to the DEM resolution. The format
and sources of data collected is enumerated in Table 1. The prepared thematic maps
of parameters and their subclasses are shown in Fig. 2.
.
5 Seismically Induced Landslide Hazard Zonation: Results

and Discussion
5.1 Assignment of Ranks (Ri ) to Landslide Thematic Layers
In this study, the impact of the slope angle and geological formations on the overall
landslide susceptibility is observed to be high to very high. Therefore, the thematic
Fig. 2 a–i Thematic maps prepared for all the considered controlling parameters
370 A. Tyagi et al.
layers of slope angle and geological formations are assigned ranks of 9 and 8, respec-
tively. The PGA thematic layer is assigned a rank of 7 considering the active seis-
micity of the study area and the past incidences of major earthquake (most notably
the 1905 Kangra earthquake, M s = 7.8) in the region. Widening of the existing NH-5
in the study area has rendered additional instability due to toe cutting, and therefore,
the road Euclidean distance thematic layer is assigned a rank of 7. All these causative
parameters are perceived to have extreme to very strong influence on the overall land-
slide hazard of the study area. The slope aspect thematic layer, the LULC thematic
layer, and the fault Euclidean distance thematic layers are assigned ranks of 6, 5,
and 5, respectively, considering their strong influence on the landslide hazard. The
elevation thematic layer is assigned a relatively moderate rank of 3 in this present
study. Although the effects of drainage density on the landslide susceptibility are
well recognized for the Himalayas, the drainage Euclidean distance thematic layer is
assigned a low rank of 1 mainly due to the lack of major rivers/drainage in the study
area. The consistency of the rank assignment process is checked by formulating the
decision matrix using AHP which is shown in Table 2. 36 pair-wise comparisons
are made for the formulated decision matrix shown in Table 2, with 5 Eigen vector
solutions (λmax equals to 9.338) and an error of 4.9E-8. The consistency ratio (CR)
for the computed decision matrix is 2.98%. Since CR < 10%, the decision made, i.e.
the ranks assigned are found to be consistent.
5.2 Assignment of Weights (Wi ) to Subclasses of Landslide

Thematic Layers
The frequency ratios (FRs) of the subclasses of various thematic layers excluding
the PGA thematic layer are computed as discussed in Sect. 3. The computed FR
values range from a minimum of 0 to a maximum of 6.76 (discarding one value
of 26.19 as a statistical outlier), implicating the significant variations of the effects
of different thematic subclasses on the overall landslide susceptibility of the study
area. For example, in the slope angle thematic layer, two subclasses have significant
variations in their computed FR values. The subclass of slopes with less than 15° angle
has a very low FR of 0.08 indicating that these slopes have very low susceptibility
for landslides. Only 1 landslide is observed in 171,351 pixels for this subclass which
constitute almost 21% of the total pixels in the study area. On the other hand, the
slopes with greater than 45° angle have extremely high landslide susceptibility as
10% of total the identified landslides occur in this subclass which constitutes only
1.6% of the total pixels of this thematic layer. Based on their perceived influence on
the overall landslide susceptibility of the study area, the FR values are categorized
into 5 classes of [0–0.39], [0.40–0.69], [0. 70–1.09], [1.10–1.49], and [1.50–6.76],
which are then assigned a weight of 1, 3, 5, 7, and 9, respectively. This allows for the
extrapolation of the bivariate model into the heuristic model in the same AHP scale
which is used for assignment of ranks (Ri ) to the landslide thematic layers. This is
Table 2 Assigned ranks (Ri ) to the landslide thematic layers and formation of the AHP decision matrix for the evaluation of the consistency of the rank
assignment process
Landslide Assigned AHP decision matrix Normalized
controlling ranks (Ri ) Slope angle Lithology PGA Road ED Slope aspect LULC Fault ED Elevation Drainage ED eigen values
parameters
Slope Angle 9 1 0.293
Lithology 8 0.50 1 0.202
PGA 7 0.33 0.50 1 0.132
Road ED 7 0.33 0.50 1.00 1 0.132
Slope 6 0.25 0.33 0.50 0.50 1 0.085
Aspect
LULC 5 0.20 0.25 0.33 0.33 0.50 1 0.056
Fault ED 5 0.20 0.25 0.33 0.33 0.50 1.00 1 0.056
Elevation 3 0.14 0.17 0.20 0.20 0.25 0.33 0.33 1 0.028
Seismically Induced Landslide Hazard Analyses for a Road Corridor …
Drainage 1 0.11 0.12 0.14 0.14 0.17 0.20 0.20 0.33 1 0.017
ED
371
372 A. Tyagi et al.
Table 3 Computation of FR values for thematic subclasses and their weights (W i,j ) in AHP scale
Slope No. of Total Ratio of Pixels Total Ratio Frequency Final
angle landslides the no. of pixels ratio Weight
subclass landslides
< 15° 1 64 0.015 171,351 827,001 0.2071 0.08 1
15°–25° 16 0.250 293,324 0.3547 1.42 7
25°–35° 20 0.313 266,336 0.3221 0.97 5
35°–45° 20 0.313 82,608 0.0999 3.13 9
>45° 7 0.109 13,382 0.0162 6.76 9
a crucial task in multi-hazard integration as it provides with the scope to fit in both
the preparatory and triggering causative factors in the same scale. The same scale
is then used to assign the weights to the subclasses of the PGA thematic subclasses
facilitating multi-hazard integration. Table 3 shows computation of FR values for
slope angle subclasses and assignment of the final weights in the AHP scale. The
same process is used to assign the weights of other subclasses of various thematic
layers, but not shown here due to constrain of space.
The PGA thematic class is categorized into 5 different subclasses, each of which
represent a different level of seismic hazard for the same scenario earthquake. Higher
values of PGA represent a higher level of seismic hazard, and therefore, their weights
can also be assigned in a linear scale. For example, a PGA value of 0.147 is less severe
than a PGA of 0.163, and therefore, the weight of the former should be less than
the latter. Based on this simple understanding, the PGA subclasses are assigned the
following weights in the same AHP scale which is shown in Table 4. It may be noted
that computation of FRs for PGA subclasses seems irrational as the PGA thematic
layer is derived based on a particular exceedance probability in a finite time. Thus,
the study refrains from carrying out proximity analyses for the PGA thematic layer
which is believed to be one of many probable scenarios of future earthquakes in
the lower Himalayan region. Readers may further refer to Nath et al. [1] for a more
elaborate discussion on the implications of scenario earthquakes on landslide hazard
zonation for the Himalayas.
Table 4 Assigned weights to

Range of PGA values (in Weight Perceived hazard level
the subclasses of the PGA
g)
thematic layer in AHP scale
0.1477–0.1539 5 High
0.1539–0.1582 6 High to very high
0.1582–0.1618 7 Very high
0.1618–0.1638 8 Very high to extremely
high
>0.1638 9 Extremely high
5.3 Generation of the Seismically Induced Landslide Hazard

Zonation (LHZ) Map of the Study Area
Numerical integration of ranks (Ri ) and weights (W i,j ) of the thematic layers and
their subclasses in the GIS environment can produce the LPI of every pixel, which
is consolidated to generate the seismically induced landslide hazard zonation (LHZ)
map of the study area. The LPI values range from a minimum of 101 to a maximum
of 445. They are further classified using the “Jenks natural breaks” optimization
technique for maximizing each class’s deviation from the means of other classes.
The prepared seismically induced LHZ map is categorized into 5 separate zones of
landslide susceptibility from very low to very high. Figure 3 shows the prepared
seismically induced LHZ map of the study area.
From Fig. 3, it is observed that almost 27% of the study area falls in the low to very
low landslide hazard zones that contains only 3% of the total landslides observed.
However, almost 43% of the study area is susceptible to high to very high seismically
induced landslides, which contains approximately 85% of the observed landslides.
The remaining 30% of the study area shows moderate susceptibility towards landslide
hazard with 12% of observed landslides falling in this zone.
Fig. 3 Prepared seismically

induced landslide hazard
zonation map of the study
area showing five separate
classes of landslide hazard
374 A. Tyagi et al.
Table 5 Frequency ratio analysis (spatial validation) of the prepared seismically induced landslide
hazard zonation map of the study area
Landslide Range of No. of Ratio of the No. of Ratio of the Frequency
hazard zone LPI values Pixels class pixels landslides no. of ratio of the
to total landslides in landslide
pixels the class to hazard zone
total no. of
landslides
Very Low 101–194 62,754 0.0763 0 0 0
Low 194–249 167,324 0.2034 2 0.0313 0.154
Moderate 249–292 237,842 0.2891 8 0.1250 0.432
High 292–339 244,437 0.2971 24 0.3750 1.262
Very High 339–445 110,335 0.1341 30 0.4688 3.495
5.4 Validation of the Seismically Induced Landslide Hazard

Map
Frequency ratio analysis of the prepared LHZ map is used for spatial validation,
results of which are shown in Table 5. The frequency of observed landslides gradu-
ally declines from very high level to very low landslide hazard zones with significant
division. Thus, it can be stated that the computed and identified susceptibility zones
are in good accord with the incidences of prior landslides. With a FR value of 3.495,
the extremely high landslide hazard zone has the highest density of reported land-
slides (47% of all observed landslides), despite its short spatial extent (around 13%
of the overall area).
Similarly, an efficient technique for testing an LSZ map with regard to model
prediction success is the examination of the success rate curve [18]. The success
rate curve for the prepared LHZ map is presented in Fig. 4 that shows the model
prediction accuracy as ~ 79%.
The prepared seismically induced landslide hazard zonation map indicates that for
the considered earthquake scenario, almost 43% of the study area has high to very high
landslide susceptibility. This implies that there is a high probability of widespread
landslides induced by the particular earthquake scenario. Similar observations were
made by Pareek et al. [2], where pattern shift of landslide zones from lower to
higher susceptibility occurred post the 1999 Chamoli earthquake and Nath et al.
[1], where 51% of the study area in the lower Himalayas was demarcated as very
highly susceptible zone for a scenario earthquake with same exceedance probability.
Such scenarios, though extremely rare (10% exceedance probability in 50 years),
would invariably induce slope failure of extreme multitude, considering the fragile
geology and other favourable conditions of the lower Himalayan belt. Therefore,
it is important for planners to realize the actual risk associated with seismically
induced landslides and incorporate seismic factor(s) in the traditional practice of
LHZ mapping.
4
(a) Frequency Ratio (FR)
Analysis 3.5
3
Frequency Ratio
1.26
0.43
0.15
0
0
Very Low Low Moderate High Very High
1
Cumulative Percentage of Landslide Susceptibility Area
(b) Success rate curve
0.8
0.6
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
Landslides in Cumulative Percentage
Fig. 4 Spatial and statistical validation of the prepared seismically induced landslide hazard
zonation map using a Frequency ratio (FR) analysis and b Success rate curve
6 Conclusion
The present study proposes a heuristic approach that coalesces semi-quantitative

and bivariate frameworks of landslide hazard zonation for assessment of seismically
induced landslide hazard for an important road corridor in the Lower Himalayas.
376 A. Tyagi et al.
The study considers PGA of an earthquake scenario (return period of 475 years) as
the landslide-triggering seismic factor. Using the proposed heuristic approach, the
seismic factor is successfully integrated with eight landslide controlling parameters of
the study area to generate a landslide hazard zonation map under the specific seismic
condition. The results of the study demonstrate that application of probabilistically
generated PGA factor can enhance the current practice of landslide hazard zonation
in terms of evaluating the co- and post-seismic landslide hazard for an expected
earthquake scenario. The findings of the study, interpreted in terms of hazard zones,
show that the susceptibility of seismically induced landslides is high to very high for
more than 40% of the study area and demonstrates the impact of major earthquakes
on the overall landslide susceptibility of the lower Himalayan belt. Thus, the study
recommends the inclusion of seismic factor for a more realistic assessment of the
prevailing landslide hazards in the seismically active hilly regions like the Himalayas.
References
1. Nath, R.R., Pareek, N., Sharma, M.L.: Implications and inclusion of size-dependent scenario
earthquakes on landslide hazard zonation: a case study of the Indian Himalayas. CATENA 212,
106027 (2022)
2. Pareek, N., Sharma, M.L., Arora, M.K.: Impact of seismic factors on landslide suscep-ibility
zonation: a case study in part of Indian Himalayas. Landslides 7(2), 191–201 (2010)
3. Nath, R.R., Pal, S., Sharma, M.L.: Use of probabilistically generated scenario earthquakes in
landslide hazard zonation: a semi-qualitative approach. In: Impact of Climate Change, Land Use
and Land Cover, and Socio-economic Dynamics on Landslides. Springer, Singapore, pp. 247–
274 (2022)
4. Marano, K.D., Wald, D.J., Allen, T.I.: Global earthquake casualties due to secondary effects:
a quantitative analysis for improving rapid loss analyses. Nat. Hazards 52(2), 319–328 (2010)
5. Kennedy, I.T., Petley, D.N., Williams, R., Murray, V.: A systematic review of the health impacts
of mass earth movements (landslides). PLoS Curr. 7 (2015)
6. Bhambri, R., Mehta, M., Singh, S., Jayangondaperumal, R., Gupta, A.K., Srivastava, P.: Land-
slide inventory and damage assessment in the Bhagirathi Valley, Uttarakhand, during June 2013
flood. Himalayan Geol. 38(2), 193–205 (2017)
7. Singh, K., Kumar, V.: Hazard assessment of landslide disaster using information value method
and analytical hierarchy process in highly tectonic Chamba region in bosom of Himalaya. J.
Mt. Sci. 15(4), 808–824 (2018)
8. Panchal, S., Shrivastava, A. K.: Landslide hazard assessment using analytic hierarchy process
(AHP): a case study of National Highway 5 in India. Ain Shams Eng. J. 13(3), 101626 (2022)
9. Ali, S., Biermanns, P., Haider, R., Reicherter, K.: Landslide susceptibility mapping by using GIS
along the China–Pakistan economic corridor (Karakoram Highway), Pakistan. Nat. Hazards
Earth Syst. Sci., 1–40 (2018)
10. Alsabhan, A.H., Singh, K., Sharma, A., Alam, S., Pandey, D.D., Rahman, S.A.S., Khursheed,
A., Munshi, F.M.: Landslide susceptibility assessment in the Himalayan range based along
Kasauli-Parwanoo road corridor using weight of evidence, information value, and frequency
ratio. J. King Saud Univ. Sci. 34(2), 101759 (2022)
11. Tanyaş, H., et al.: Presentation and analysis of a worldwide database of earthquake-induced
landslide inventories. J. Geophys. Res. Earth Surf. 122(10), 1991–2015 (2017)
12. Nath, R.R., Sharma, M.L., Tyagi, A.: Review of the current practice on inclusion of seismicity
in landslide susceptibility zonation: a case study for Garhwal Himalaya. Himalayan Geol.
41(2), 222–233 (2020)
13. Nath, R.R., Sharma, M.L., Goswami, A., Sweta, K., Pareek, N.: Landslide susceptibility zona-
tion with special emphasis on tectonic features for occurrence of landslidesin Lower Indian
Himalaya. J. Indian Soc. Remote Sens. 49(5), 1221–1238 (2021)
14. Nadim, F., Kjekstad, O., Peduzzi, P., Herold, C., Jaedicke, C.: Global landslide and avalanche
hotspots. Landslides 3(2), 159–173 (2006)
15. Boore, D.M., Atkinson, G.M.: Ground-motion prediction equations for the average horizontal
component of PGA, PGV, and 5%-damped PSA at spectral periods between 0.01 s and 10.0 s.
Earthq. Spectra 24(1), 99–138 (2008)
16. Sarkar, S., Kanungo, D.P.: An integrated approach for landslide susceptibility map-ping using
remote sensing and GIS. Photogramm. Eng. Remote. Sens. 70(5), 617–625 (2004)
17. Saaty, T.L.: The Analytic Hierarchy Process: Planning, Priority Setting, Resource Allocation.
McGraw-Hill Book Co, New York (1980)
18. Chung, C.J.F., Fabbri, A.G.: Probabilistic prediction models for landslide hazard mapping.
Photogramm. Eng. Remote. Sens. 65(12), 1389–1399 (1999)
Assessment of Proxy-Based V s30
Estimation in Roorkee, Uttarakhand
M. Srivastava and M. L. Sharma
Abstract A well-liked measure for site characterization and classification is the

averaged seismic shear wave velocity with time for the top 30 m below the surface,
abbreviated V s30 . It is associated with material property and thus has a strong influ-
ence on ground motions. V s30 is an important aspect of earthquake site effects and
a crucial element for ground response analyzes (GRA) and localized site-specific
effects. To examine shear wave velocity (V s ), standard penetration test and cone
penetration test are a few in situ field tests in practice. Additionally, there are surface
wave techniques like continuous surface wave systems, multichannel analysis of
surface waves test, and spectral analysis of surface waves test (SASW) that provide a
non-intrusive, non-invasive, economical, and expedient way to conduct site investiga-
tions. Complex geomorphology, geology, and lack of in situ measurements emphasize
establishing a more suitable approach for V s30 based site classification. To deal with
such issues, various proxy-based techniques have been proposed, such as topography
slope, surface geology, and terrain-based approach. Among the three proxy-based
approaches, the topography slope method is gaining importance for preparing site
condition maps for seismic damage assessments. The goal of the current study is
to carry out an MASW test at various locations, enumerate the shear wave velocity
profile at each location, and compare those results to those from the proxy-based
V s30 estimation. Both MASW and proxy-based V s30 estimations have emerged as an
important tools. This paper presents important aspects of both techniques and their
reliability for future site investigations and effectiveness in geotechnical seismic site
characterization.
Keywords V s30 · MASW · Slope proxy · Site characterization
M. Srivastava (B) · M. L. Sharma

e-mail: msrivastava@eq.iitr.ac.in
M. L. Sharma
https://doi.org/10.1007/978-981-99-1459-3_30
380 M. Srivastava and M. L. Sharma
1 Introduction
Since ancient times, inquisitiveness to explore the subsurface to deliver the expla-
nation for seismic activities has been a topic of interest among researchers. Seismic
exploration and geotechnical investigation using passive sources (earthquake waves)
and active sources (explosions) emerged as a progressive domain for the study.
Modern portable equipment and various technologies have facilitated seismic studies
even in the complex Himalayan regions. The behavior of the soil for a region under
dynamic loading accounts for potential risk and damageability. The behavior of soil is
determined by dynamic soil features, namely shear wave velocity (V s ), poison’s ratio,
shear modulus (G), and damping characteristics [6]. Ground motions are precepted
by V s , as it is associated with stiffness and material property of the soil profile.
According to Anderson et al. [2] and statements from various researchers, features
in the top 30 m are important for local site-effect analyzes of earthquake ground
motions. As a result, V s30 came into the limelight for site characterization and profile
description. To examine shear wave velocity (V s ), high strain tests such as standard
penetration test (SPT-N), cross-hole test (CHT), cone penetration test (CPT), and
down-hole test (DHT) are a few in situ field tests in practice. Additionally, there
are surface wave techniques like continuous surface wave systems (CSWS), multi-
channel analysis of surface waves test (MASW), seismic reflection/refraction test,
and spectral analysis of surface waves test (SASW) that can be used to investigate a
site without being intrusive or invasive, as well as being inexpensive and quick.
Standard penetration test (SPT) for 30 m depth provides numbers of blows (N),
constructive details of the soil profile, relative density, and approximate shear strength
but it is difficult to conduct the test in rocky structures. SPT tests are expensive
and time-consuming. MASW consists of a three-step procedure in which Rayleigh
surface wave data are acquired, phase velocity against frequency is plotted (disper-
sion curve), V s is inverted, and then, a shear wave velocity (V s ) profile is obtained
after Rayleigh-type surface wave analysis on a multichannel record [12]. MASW is
a widely used approach due to its non-invasive and time-effective application, but
it is also a laborious and tough job in complex site conditions. Maheshwari et al.
[8] conducted field tests at selected sites in the Himalayan region and provided an
applicable relationship between V s and N. Reliable correlation between V s estimated
from MASW and “N” from SPT test has been also presented by Muley et al. [7] for
Roorkee. It can be inferred from Roy et al. [11] that the V s30 values may differ upon
a random selection of V s -N correlation and can lead to pseudo site classification
and therefore, pinned the need for site-specific correlations. Various studies have
been made for the region for V s30 measurement, Sharma et al. [13], provided V s30 =
267 m/s and engineering bedrock around 300 m of depth for Roorkee. The amplifica-
tion ratio of 7.0 was also observed, due to the presence thick alluvium deposit in the
region. Considering geological studies and records, thick alluvial cover indicates the
presence of three layers, i.e., soft, medium, and hard soil columns followed by hard
rocks. Soil characterization using specific methodologies and information available,
the area was divided into three: Solani River (V s30 = 190 m/s), Roorkee West &
Assessment of Proxy-Based V s30 Estimation in Roorkee, Uttarakhand 381
East (V s30 = 210 m/s & 240 m/s, respectively). It can be inferred that deep deposits
should be included for more precise local site effects studies [14].
For such geologically challenging regions, proxy-based techniques such as topog-
raphy slope, surface geology, and hybrid slope geology to determine V s30 have been
introduced [1]. The topography slope proxy approach has proven to be a flexible tool
for analyzing seismic site conditions in the first order to deliver V s30 . Since competent
materials sustain a sharp slope and while on the contrary deep sediments are gener-
ally formed in conditions with extremely low gradients, shear velocity down to 30 m
observations are tie-up against the topographic slope, and factors for V s30 predic-
tions are created [18]. As part of a geology-based approach to site classification, Wills
et al. [19] linked V s30 with geological parameters and created a site category map
of California. The terrain-based classification was first introduced by Iwahashi and
Pike [5] as a result of geomorphic categories based on the slope’s steepness/slope
gradient, the roughness of the surface/surface texture, and surface curvature/local
convexity. Stewart et al. [15] conducted a study in Greece that found that both the
geology-based proxy and the terrain-based proxy assessment were equally useful. In
California, the terrain-based method is ideally suited for site classification [20].
The National Earthquake Hazard Reduction Programme’s “Design criteria” are
collated with the respective versus soil profiles of the various sites in this research
generated from, MASW and the proxy-based method for site categorization [3].
2 Study Area
Roorkee is a small city situated in the Gangetic belt close to the foothills of the
Himalayas and lies in high seismic zone IV. It comprises a settlement of recent
alluvium with a gentle slope. Roorkee city comes under the Haridwar district of the
state Uttarakhand having 30 km from the district headquarters, 65 km from the state
capital, Dehradun, and 184.3 km north of the national capital, New Delhi (Fig. 1).
3 Methodology
3.1 Geotechnical Method—Multichannel Analysis

of Surface Waves (MASW)
At three locations within the campus of IIT Roorkee, the Department of Earthquake
Engineering DEQ, ABN ground, and convocation ground, respectively, MASW tests
were carried out to evaluate the shear wave velocity (V s ) variation with depth, V s30
profile. The inversion of the dispersion curve yields the shear wave velocity (V s )
variation in a layered soil structure [4]. For shallower inquiry depth, the active MASW
approach is used, and the passive Nakamura method/horizontal-to-vertical spectral
Fig. 1 Site locations within IIT Roorkee Campus Roorkee, Uttarakhand
ratio method (HVSR) is used for deeper depth. Finding a comprehensive dispersion
curve for the site is encouraged by active as well as passive data collected from
the site location. Thus, by integrating dispersion curves and improving the general
nature of extended frequencies and phase velocity ranges, more trustworthy results
are obtained for the locations [9]. To derive V s30 , the velocity profile of the locations
and joint inversion of HVSR with MASW have been used in the study (Fig. 2).
3.2 Proxy-Based Vs30 Estimation and Site Classification
V s30 estimation based on topography slope as a proxy was extracted from the USGS
V s30 map with grid resolution provided by the map of 30 arc seconds [17]. With
the aid of geographic information system mapping tools, the analysis execution was
performed (Fig. 3).
The term “slope,” which is extracted from the digital elevation model (DEM)
data, refers to, slope of vertical height to the horizontal distance ratio, also known
as the gradient. Competent high-velocity materials maintain steep slopes and deep
basin sediments with low gradients. Park and Elrick [10] place special emphasis on
the correlation between lower shear wave velocity to the slope based on sediment
(a) (b) (c)
Fig. 2 V s profile at test sites a DEQ, b ABN ground, and c convocation ground
Fig. 3 Site classification: a the topographic slope and b the geology-based proxy-based estimation
fineness as a proxy. Among the physical parameters, the void ratio affects the shear
modulus, hence, more pertinent to shear velocity. Void ratio measurement can be
well obtained by soil texture and relative grain size distribution, Fumal and Tinsley
[16]. Shear velocity is found to be proportional to grain size explaining the support
for good correlation (Table 1).
Meghalayan and Holocene, Miocene-Pliocene, Pliocene–Pleistocene, and
Middle-Late Pleistocene are the several geological eras that make up the region. V s30
was associated with geological units (Rock types/Age, Strata, Lithology) by Wills
et al. in 2000 (Table 2). Site classification has been made in the current study for the
area based on the geological/stratigraphical/chronological dataset and V s30 ’s estab-
lished correlation. According to Yong et al. [20], the terrain-based strategy entails
the creation of a collection of geomorphic categories based on surface texture, slope
gradient, and local convexity generated from surface curvature (Table 3).
Table 1 NEHRP’s summary of slope ranges and V s30

Site class V s30 (m/s) Slope ranges (m/m)
Active region Stable region
E <180 <1.0E−4 <2.0E−5
180–240 1.0E−4–2.20E−3 2.0E−5–2.0E−3
D 180–240 2.20E−3–6.30E−3 2.0E−3–4.0E−3
300–360 6.30E−3–0.018 4.0E−3–7.2E−3
360–490 0.018–0.05 7.2E−3–0.013
C 490–620 0.05–0.10 0.013–0.018
620–760 0.1–0.138 0.018–0.025
B >760 >0.138 >0.025
Source Wald and Allen [18]
Table 2 Accordance between geological age and site class

Geological age Site class Lithological description
Meghalayan/Holocene D Hard soil, sand, clay, and gravel
Pleistocene/Pliocene/Miocene CD
Oligocene/Eocene/Paleocene C Hard to very hard soil, mostly gravel, and
standard soft rock
Late Cretaceous/Early Cretaceous B Rocks
Late Jurassic/Middle Jurassic
Early Jurassic/Late Triassic
Middle Triassic/Early Triassic
…
Source Wills et al. [19]
4 Results
According to S-wave velocity profiles derived from the MASW test, shear wave
velocity (V s ) variation with depth is V s30 = 291 m/s at DEQ, V s30 = 302 m/s at
ABN ground, and V s30 = 303 m/s at convocation ground, respectively. The study
region is designated as site class D by the NEHRP criteria for site classification,
with a V s30 value ranging 180–360 m/s. Site class D for DEQ as well as convocation
ground is found in the region using the topographic slope-based technique, with
V s30 values of 327 and 304 m/s, respectively. According to the NEHRP criteria,
ABN ground V s30 , which was calculated using this method, is 368 m/s, placing it in
site class C. The area is categorized as site class CD [Miocene-Pliocene, Pliocene–
Pleistocene, Middle-Late Pleistocene] V s30 values ranging 360–760 m/s and site
class D [Meghalayan and Holocene] V s30 values ranging 180–360 m/s from the
geology-based proxy technique (Fig. 4).
Table 3 Terrain types and correspondence mean V s30

Terrain types Terrain description Mean V s30 value
TT 1 Well-dissected alpine summits, mountains, etc. 519
TT 2 Large volcanos, high block plateaus, etc. 393
TT 3 Well dissected, low mountains, etc. 547
TT 4 Volcanic fans, foot slope of high block plateaus, etc. 459
TT 5 Dissected plateaus, etc. 402
TT 6 Basalt lava plains, glaciated plateaus, etc. 345
TT 7 Moderately eroded mountains, lava flow, etc. 388
TT 8 Desert alluvial slope, volcanic fans, etc. 374
TT 9 Well eroded plains of weak rocks, etc. 497
TT 10 Valleys, till plains, etc. 349
TT 11 Eroded plains of weak rocks, etc. 328
TT 12 Eroded alluvial fans, till plains, etc. 297
TT 13 Incised terraces, etc. –
TT 14 Eroded alluvial fans, till plains, etc. 209
TT 15 Dunes, incised terraces, etc. 363
TT 16 Fluvial plains, alluvial fans, low-lying flat plains, etc. 246
Source Yong et al. [20]
Fig. 4 Terrain-based site classification

Table 4 NHERP soil classification

Soil class Rock and soil description Average shear wave velocity (V s30 ) m/s
A Hard rocks >1500
B Rocks 760–1500
C Hard, stiff to very stiff soil, most gravel 360–760
D Sand, silt, stiff to very stiff clays, and 180–360
some gravel
E Soft clays <180
Source NEHRP [3]
Terrain-based classification of site type in the study area region was detailed and
diverse. There are six types of terrain in the area. TT3: well-dissected structural hills
and valleys/dam & reservoirs, TT1: snow-covered/moderately well-dissected struc-
tural hills and valleys/glacial nunatak/arete, in addition, TT16: river/active flood
plains/younger alluvial plains/water body lake, TT12: older flood plains/abandoned
quarry, TT14: older alluvial plains/mass-wasting products, and TT9 piedmont allu-
vial plain land/dissected structural hills and valleys are examples. Roorkee constitutes
TT14 and TT16 site classes, respectively (Table 4).
5 Conclusions
This study made an effort to compare the applicability of the three proxy methods.
Site classification in the study area is done by collecting the data on a regional scale
and observing the variation from the results obtained from the MASW test done at
site locations. It can be inferred that the topographic slope-based approach has a
very well match with the MASW results. The geology-based approach has a close
correlation with the site data as compared to Terrain-based classification. Although
the site is well classified and diverse in the case of the Terrain approach, results are
underestimated for the plain area and emphasize localized correction for the area.
Among the three proxy-based approaches, the topographic slope method appeared
well correlated for the region and can be taken into account for preliminary studies
as a quick investigation tool.
References
1. Ahdi, S.K., Stewart, J.P., Ancheta, T.D., Kwak, D.Y., Mitra, D.: Development of VS profile
database and proxy-based models for VS 30 prediction in the Pacific Northwest region of North
America. Bull. Seismol. Soc. Am. 107(4), 1781–1801 (2017)
2. Anderson, J.G., Lee, Y., Zeng, Y., Day, S.: Control of strong motion by the upper 30 meters.
Bull. Seismol. Soc. Am. 86(6), 1749–1759 (1996)
3. Committee, B.S.S.: Draft National Earthquake Hazard Reduction Programme (NEHRP)

Recommended Provisions for Seismic Regulations for New Buildings and Other Structures.
National Science Foundation, Washington (1997)
4. Foti, S., Parolai, S., Bergamo, P., Di Giulio, G., Maraschini, M., Milana, G., Puglia, R.: Surface
wave surveys for seismic site characterization of accelerometric stations in ITACA. Bull. Earthq.
Eng. 9(6), 1797–1820 (2011)
5. Iwahashi, J., Pike, R.J.: Automated classifications of topography from DEMs by an unsuper-
vised nested-means algorithm and a three-part geometric signature. Geomorphology 86(3–4),
409–440 (2007)
6. Jhinkwan, H., Jain, P.K.: Prediction of shear wave velocity using SPT-N value. Int. J. Curr.
Eng. Sci. Res. 3, 28 (2016)
7. Kirar, B., Maheshwari, B.K., Muley, P.: Correlation between shear wave velocity (Vs) and SPT
resistance (N) for the Roorkee region. Int. J. Geosynthetics Ground Eng. 2(1), 1–11 (2016)
8. Maheshwari, B.K., Mahajan, A.K., Sharma, M.L., Paul, D.K., Kaynia, A.M., Lindholm, C.:
Relationship between shear velocity and SPT resistance for sandy soils in the Ganga basin. Int.
J. Geotech. Eng. 7(1), 63–70 (2013)
9. Park, C.B., Miller, R.D., Xia, J., Ivanov, J.: Multichannel analysis of surface waves (MASW)—
active and passive methods. Lead. Edge 26(1), 60–64 (2007)
10. Park, S., Elrick, S.: Predictions of shear-wave velocities in southern California using surface
geology. Bull. Seismol. Soc. Am. 88(3), 677–685 (1998)
11. Roy, N., Shiuly, A., Sahu, R.B., Jakka, R.S.: Effect of uncertainty in VS –N correlations on
seismic site response analysis. J. Earth Syst. Sci. 127(7), 1–21 (2018)
12. Sairam, B., Rastogi, B.K., Aggarwal, S., Chauhan, M., Bhonde, U.: Seismic site character-
ization using Vs30 and site amplification in Gandhinagar region, Gujarat, India. Curr. Sci.
754–761 (2011)
13. Sharma, P., Sharma, M.L., Sawant, V.: Estimation of seismic hazard and amplification of strong
ground motions in Indo-Gangetic Plains. J. Seismol. Earthq. Eng. 22(1), 15–30 (2020)
14. Srivastava, A.K., Sharma, M.L., Paul, D.K., Das, J., Jakka, R.: Deep Soil Characteristics in the
Vicinity of Himalayas, vol. 1991, pp. 74–86 (2014)
15. Stewart, J.P., Klimis, N., Savvaidis, A., Theodoulidis, N., Zargli, E., Athanasopoulos, G.,
Pelekis, P., Mylonakis, G., Margaris, B.: Compilation of a local VS profile database and its
application for inference of VS 30 from geologic-and terrain-based proxies. Bull. Seismol.
Soc. Am. 104(6), 2827–2841 (2014)
16. Fumal, T.E., Tinsley, J.C.: Mapping shear wave velocities of near-surface geological materials.
In: Ziony, J.I. (ed.) Predicting Area Limits of Earthquake Induced Land Sliding: In Evaluation of
Earthquake Hazards in the Los Angeles Region–An Earth Science Prospective, US Geological
Survey Paper 1360, pp. 127–150 (1985)
17. Unites States Geological Survey: Vs 30 Models and Data (USGS, 2020). Retrieved
from https://usgs.maps.arcgis.com/apps/webappviewer/index.html?id=8ac19bc334f747e486
550f32837578e
18. Wald, D.J., Allen, T.I.: Topographic slope as a proxy for seismic site conditions and
amplification. Bull. Seismol. Soc. Am. 97(5), 1379–1395 (2007)
19. Wills, C.J., Petersen, M., Bryant, W.A., Reichle, M., Saucedo, G.J., Tan, S., Taylor, G., Treiman,
J.: A site-conditions map for California based on geology and shear-wave velocity. Bull.
Seismol. Soc. Am. 90(6B), S187–S208 (2000)
20. Yong, A., Hough, S.E., Iwahashi, J., Braverman, A.: A terrain-based site-conditions map of
California with implications for the contiguous United States. Bull. Seismol. Soc. Am. 102(1),
114–128 (2012)
Ground Motion Prediction Equation
for NW Himalaya Region
Vandana, Harendra K. Dadhich, Himanshu Mittal, and O. P. Mishra
Abstract An attenuation relationship, also known as ground motion prediction

equation (GMPE), has been developed for Northwest Himalaya and its neighboring
region using the strong-motion data recorded by National Center for Seismology
Network. In the present study, we have used the 40 records of 18 earthquakes
from the 2010 to 2020 having magnitudes (5.0 < Mw < 5.6) that occurred in the
region. To develop the new attenuation relationship, we have used the maximum peak
ground acceleration (PGA) of two horizontal components. Two-step stratified-based
regression approach is followed. The newly developed relationship is as follows:
Log10 PGA(gal) = 0.36512+0.55599∗M−0.95736·Log10 (HD + exp(1.27581∗M) )
where PGA is peak ground acceleration in gal, M is the magnitude, and HD is
hypocentral distance. The standard error is of the order of 0.0093. The estab-
lished GMPE is proposed to be useful for seismic hazard estimation, ground motion
simulation, site-specific studies, and other engineering applications.
Keywords Attenuation relationship · Regression · GMPE · Amplitude
1 Introduction
Estimation of ground motion in earthquake engineering is one of the primary

aspects. The accurate prediction of ground motion at a site is consistent for seismic
hazard analysis. This can be obtained by deriving ground motion predictive equa-
tion (GMPE) using a large set of observed data. Many equations are available in the
regions where the large amount of strong-motion data is accessible. So, the best way
to estimate the GMPEs is using the recorded strong-motion data which is always
Vandana (B) · H. K. Dadhich · H. Mittal · O. P. Mishra

National Centre for Seismology, Ministry of Earth Sciences, New Delhi, India
e-mail: vandana.ghangas@gmail.com
O. P. Mishra
e-mail: omp.mishra@nic.in
https://doi.org/10.1007/978-981-99-1459-3_31
390 Vandana et al.
uncertain for a region with shortage of observed strong-motion data like India. On
the other hand, the other way to estimate the GMPEs is to simulate the data which is
widely used in different parts of the world by using different simulation techniques
involving empirical Green’s function technique and stochastic simulation approach
[2, 8, 9].
GMPEs represent an empirical relationship between ground shaking parameters
and earthquake parameters. Singh et al. [12] described that the ground shaking param-
eters are the main requirement to carry out the analysis of seismic hazard of a region.
The region-specific GMPEs play an important role for macro- and micro-level hazard
mapping [1]. GMPEs are useful in designing of important structures like: nuclear
power plants, hydroelectric projects, etc. So, the selection of appropriate GMPEs may
have a great dominance on the design and safety of a structure. So, the developed
attenuation relationship is necessary in prediction of strong ground motion, in seismic
hazard assessment and in the reduction of seismic risk. To develop the appropriate
GMPEs, ground motion data recorded from various magnitude earthquakes and at
various distances should be available. In addition, the use of site and path parameters
authenticates the development of GMPEs. A number of GMPEs or attenuation rela-
tionships are available for Himalayan region. Region-specific GMPEs developed by
Singh et al. [12], Sharma [13], Iyenger and Ghosh [4], Nath et al. [11], Nath et al.
(2009), Sharma and Bungum [14], Sharma et al. [15], Anbazhagan et al. [1] and
Kumar et al. [6, 7] for the Himalayan region used recorded data, data from other
regions having same geological conditions, and the simulated data.
In the present study, an attenuation relationship, also known as GMPE, has been
developed for the Northwest Himalaya and the neighboring region by using the
strong-motion data of National Center for Seismology (NCS) Network. In the present
study, 40 records of 18 earthquakes from the 2010 to 2020 having magnitudes (5.0 <
M W < 5.6) that occurred in the region are considered for the study. The established
GMPE will be very useful in many studies, involving, site-specific studies, hazard
assessment, simulation of ground motion, and other engineering applications.
2 Geology and Tectonics
About 50 Ma years ago, the nucleation of the Indian plate with the Eurasian plate
gave rise to the formation of the Himalaya [20]. Himalaya is alienated into four main
tectonic from north to south, namely, the Tethys Himalaya, the Great Himalaya, the
Lesser Himalaya, and the Sub-Himalaya. The Indus–Tsangpo suture zone (ITSZ) is
the most northern boundary of the Indian plate. The Trans-Himadri fault separates
the Tethys Himalaya and the Great Himalaya. Earlier the Trans-Himadri fault was
known as the Malari thrust [16, 17], and later was changed as Trans-Himadri fault
[18, 19]. The Greater Himalaya and the Lesser Himalaya are separated by a major
tectonic boundary known as Main Central Thrust (MCT). The other prominent thrust
fault, the Main Boundary Thrust (MBT), separates the Lesser Himalaya from the
Sub-Himalaya. The distinction of Sub-Himalaya from the Indo-Gangetic Plains is
Ground Motion Prediction Equation for NW Himalaya Region 391
Fig. 1 Tectonic map of study region with location of earthquakes and stations recording these
earthquakes
marked by another thrust known as Himalaya Frontal thrust (HFT). HFT is visible at
many places. The southward relocation of the main deformation front is supported
by these primary thrusts, namely MCT, MBT, and MFT, due to its young age and
shallow depth. Some limited portions of these thrusts exhibit active faulting and
neotectonic activity on the surface [10]. The two set-off structures are divided by
the major tectonic segregation, or MCT, into the Lesser Himalaya and the Higher
Himalaya, which have different stratigraphy and tectonic features (Fig. 1).
3 Data
The earthquake data used in present study are recorded at 15 stations of the digital
strong-motion network of the National Centre for Seismology (NCS), Ministry of
Earth Sciences (MoES) from 2010 to 2020. The used data consist of 40 records
from 18 earthquakes having magnitudes (5.0 < M W < 5.6). The epicenters of these
18 earthquakes are taken from International Seismological Centre (ISC) bulletin,
and the location of the earthquakes is plotted on the study area (Fig. 1). The data
392 Vandana et al.
with sampling rate of 100 samples per second (sps) were recorded at different
stations, namely, Dehradun (DDI), Dharamshala (DHRM), Shimla (SMLA), Talwara
(TLWR), Thein Dam (THN), Uttarkashi (UTK), Joshimath (JOSI), Kalpa (KLP),
Tissa (TSSA), Jammu (JMU), Hanley (HNLY), and Alchi (leh). The digital accelero-
graphs of TSA-100S type (Metrozet, model TSA-100S) are installed at Dehradun
(DDI), Dharmshala (DHRM), and Shimla (SMLA) stations. On the other hand, the
digital accelerographs of 147A-01/3 (Reftek, model147A-01/3) type are used at
Talwara (TLWR), Thein Dam (THN), Uttarkashi (UTK), Joshimath (JOSI), Kalpa
(KLP), Tissa (TSSA), Jammu (JMU), Hanley (HNLY), and Alchi (leh) stations. In
the present study, we did not use the dataset of Bhakra (BHK), Sundernagar (SNDN),
Srinagar (SRIN), and Chandigarh (CGRH) stations because of the poor signal-to-
noise ratio. To develop the new attenuation relationship, we have used the maximum
PGA out of two components (Horizontal components). The earthquakes considered
for attenuation regression are shown in Table 1.
Table 1 Hypocentral parameters used for attenuation regression analysis

Eq. No. Date Time Latitude Longitude Depth Magnitude No. of records
MW
1 22-02-2010 23:45.6 30.0106 80.0846 16.5 5.3 1
2 28-05-2010 25:03.9 31.2301 77.8568 26.4 5 1
3 20-06-2011 27:19.9 30.5513 79.3192 26.6 5.2 2
4 09-02-2012 17:32.9 30.9444 78.2981 24.8 5.1 2
5 17-02-2012 44:24.1 32.3928 82.7188 27.1 5.3 1
6 05-03-2012 41:04.4 28.7746 76.6928 11.5 5 2
7 02-10-2012 34:54.5 32.3996 76.4856 14 5 2
8 08-10-2012 46:22.3 31.8432 78.4864 19.3 5 2
9 27-11-2012 15:14.9 30.896 78.3957 10 5.4 2
10 11-02-2013 48:51.6 30.9987 78.3247 6.3 5 2
11 04-06-2013 34:46.5 32.6036 76.582 14 5 2
12 09-07-2013 49:17.9 32.6902 78.2044 35 5 2
13 21-08-2014 11:19.2 32.2604 76.321 15.1 5 2
14 01-04-2015 23:55.3 30.3749 79.4818 17 5 2
15 06-02-2017 03:05.8 30.5773 79.0787 14.6 5.6 1
16 06-12-2017 19:49.8 30.595 79.0957 12.8 5.1 5
17 09-09-2019 40:33.5 32.9804 76.181 43.9 5.1 7
18 26-02-2020 04:09.0 32.9752 76.2474 10 5 2
4 Development of Ground Motion Relationships
The basic form of the attenuation relationship has been considered in previous studies
[1, 4, 7] is as follows:
Log(Y ) = F1(M) + F2(R, E) + F3(R, M, E) + F4(F)+ ∈ (1)
where Y is the estimated parameter and F1 to F4 are the functions of magnitude,

distance, tectonic environment, and fault geometry. M W is the magnitude; R is
hypocentral distance; E is the tectonic environment; F is the fault factor, and ∈
is the random variable accounting for uncertainties in estimation.
For each earthquake, the decay parameter b is estimated using the following linear
regression:
log(Y ) = −b log(X ) + c (2)
where Y is the predicted PGA (gals) in our case. X is hypocentral distance, and b
and c are the regression coefficients. The generalized form of Campbell [3] model is
considered for further study by discarding the fault mechanism term. The considered
GMPE has the form:
Log(Y ) = c1 + c2M − b log(X + ec3M) + ∈ (3)
where c1, c2, and c3 are the regression coefficients to be determined. Following
Joyner and Boore [5], the b value is fixed to 0.95736. ∈ represents the standard error
term. Using this form of nonlinear regression for the whole dataset, the values of
regression coefficients c1, c2, and c3 are found to be 0.36512, 0.55599, and 1.2758,
respectively. The standard error is equal to 0.0093. The logarithm term considered
in all the above equations has base 10. So, considering the values of the above
coefficients, the final GMPE obtained using recorded ground motion data can be
written as:
Log10 PGA (gal) = 0.36512 + 0.55599 ∗ M

− 0.95736 log 10 HD + exp(1.27581∗M)
5 Results and Conclusions
A dataset of 40 records of 18 earthquakes from the 2010 to 2020 having magnitudes

(5.0 < M W < 5.6) has been used in the present study. Two-step stratified approach
has been used in this study. The proposed attenuation relationship is as follows:
394 Vandana et al.
Log10 PGA(gal) = 0.36512 + 0.55599 ∗ M

− 0.95736 log 10 HD + exp(1.27581∗M)
where PGA is peak strong ground acceleration in gal, HD is hypocentral distance,

and M is magnitude. The standard error of the proposed relationship is 0.0093. The
derived GMPE is strictly valid for a narrow magnitude range and hypocenter distance
between 150 and 400 km. Again, the used earthquakes are recorded sparsely at few
stations, and for some earthquakes, only one record is available. Considering all the
above factors, the computed GMPE may not be exactly true for higher magnitude
earthquakes and in near-source regions. Considering the sparsity of data and narrow
magnitude range, we skip the comparison of our derived GMPE with the other
relations. However, the derived GMPE may be useful for site-specific studies as well
as for hazard assessment. Being valid for narrow magnitude range, the relation may
not be much useful for engineering applications but may provide useful information
for ground motion simulations.
Acknowledgements The authors are thankful to the Secretary, Ministry of Earth Sciences, Govern-
ment of India for consistent support and encouragement to do this study. Authors are grateful to the
Director of National Center for Seismology, Ministry of Earth Sciences, New Delhi, for providing
earthquake waveform data.
References
1. Anbazhagan, P., Kumar, A., Sitharam, T.G.: Ground motion prediction equation considering
combined dataset of recorded and simulated ground motions. Soil Dyn. Earthq. Eng. 53, 92–108
(2013). https://doi.org/10.1016/j.soildyn.2013.06.003
2. Boore, D.M., Joyner, W.B.: The empirical prediction of ground motion. Bull. Seismol. Soc.
Am. 72, S43-60 (1982)
3. Campbell, K.W.: Near-source attenuation of peak horizontal acceleration. Bull. Seismol. Soc.
Am. 71(6), 2039–2070 (1981)
4. Iyengar, R.N., Ghosh, S.: Microzonation of earthquake hazard in Greater Delhi area. Curr. Sci.
87, 1193–1202 (2004)
5. Joyner, W.B., Boore, D.M.: Measurement, characterization, and prediction of strong ground
motion. In: Earthquake Engineering and Soil Dynamics II, Proc. of American Society of Civil
Engineers Geotechnical Engineering Division Specialty Conference, Park City, Utah, pp. 43–
102 (1988)
6. Kumar, A., Mittal, H., Sachdeva, R., Kumar, A.: Indian strong motion instrumentation network.
Seismol. Res. Lett. 83, 59–66 (2012). https://doi.org/10.1785/gssrl.83.1.59
7. Kumar, A., Mittal, H., Kumar, R., Ahluwalia, R.S.: Empirical attenuation relationship for peak
ground horizontal acceleration for North-East Himalaya. Vietnam J. Earth Sci. 39, 46–56 (2017)
8. Mittal, H.: Estimation of Ground Motion in Delhi. Doctoral dissertation, Roorkee, India (2011)
9. Mittal, H., Kumar, A., Wu, Y.M., Kumar, A.: Source study of M w 5.4 April 4, 2011 India–
Nepal border earthquake and scenario events in the Kumaon–Garhwal Region. Arab. J. Geosci.
9(5):1–15 (2016)
10. Nakata, T.: Active faults of the Himalaya of India and Nepal. Geol. Soc. Am. Pap. 232, 243–264
(1989)
11. Nath, S.K., Vyas, M., Pal, I., Singh, A.K., Mukherjee, S., Sengupta, P.: Spectral attenuation
models in the Sikkim Himalaya from the observed and simulated strong motion events in the
region. Curr. Sci. 88(2), 295–303 (2005)
12. Singh, R.P., Aman, A., Prasad, Y.J.J.: Attenuation relations for strong seismic ground motion
in the Himalayan Region. Pure Appl. Geophys. 147(1), 161–180 (1996)
13. Sharma, M.L.: Attenuation relationship for estimation of peak ground horizontal acceleration
using data from strong-motion arrays in India. Bull. Seismol. Soc. Am. 88(4), 1063–1069
(1998)
14. Sharma, M., Bungum, H.: New strong ground-motion spectral acceleration relations for the
Himalayan Region. In: Proceedings of First European Conference on Earthquake Engineering
and Seismology (A joint event of the 13th ECEE& 30th General Assembly of the ESC). Paper
Number 1459 (2006)
15. Sharma, M.L., Douglas, J., Bungum, H., Kotadia, J.: Ground-motion prediction equations based
on data from the Himalayan and Zagros Regions. J. Earthquake Eng. 13(8), 1191–1210 (2009)
16. Valdiya, K.S.: An outline of the structural set-up of the Kumaon Himalaya. J. Geol. Soc. India
20, 145–157 (1979)
17. Valdiya, K.S., Joshi, D.D., Sanwal, R., Tandon, S.K.: Geomorphologic envelopment across
the active main boundary Thrust, an example from the Nainital Hills in Kumaun Himalaya. J.
Geol. Soc. India 25(12), 761–774 (1984)
18. Valdiya, K.S.: Trans-Himadri Thrust and domalupwarps immediately south of collision zone
and tectonic implications. Curr. Sci. 56, 200–209 (1987)
19. Valdiya, K.S.: Trans-Himadri intra crustal fault and basement upwarps south of the Indus-
Tsangpo Suture Zone. Geol. Soc. Amer. Spec. 153–168 (1989)
20. Valdiya, K.S.: Dynamic Himalaya. University Press, Hyderabad, India, p. 178 (1998)
Identification of Strong Motion
Generation Area of the 2019 Hualien
Earthquake
Saurabh Sharma, A. Joshi, S. Singh, C. M. Lin, C. H. Kuo, and K. L. Wen
Abstract Strong motion generation areas are the extended areas with large uniform
slip velocity within the entire rupture plane. The high-frequency ground motion due
to these areas need to be studied to model the contribution of stress released with
respect to the entire rupture plane. The study presented in this paper examines the
high frequency ground motion resulting from the strong motion generation area
(SMGA) located within the entire rupture plane that caused the Hualien earthquake.
The eastern coast of Taiwan was hit by a strong earthquake of M w (6.2) on April
18, 2019. The seismic activity of this earthquake was recorded by 107 stations of
the Central Weather Bureau (CWB) around entire Taiwan region. The epicentral
distance of stations recording this earthquake varies from 5 to 242 km. A total of
10 near-field stations have been used to investigate the source characteristics of this
earthquake. The records from these near-field stations show the presence of two
prominent phases in the horizontal components. These prominent phases have been
termed as SMGA and are the source of high-frequency radiations (Miyake et al in Bull
Seismol Soc Am 93(6):2531–2545 2003, [6]). The SH component obtained by the
S. Sharma (B) · A. Joshi · S. Singh

Department of Earth Sciences, Indian Institute of Technology Roorkee, Roorkee, India
e-mail: ssharma1@es.iitr.ac.in
A. Joshi
e-mail: joshi.anand@es.iitr.ac.in
S. Singh
e-mail: sandeep.singh@es.iitr.ac.in
C. M. Lin · C. H. Kuo
National Center for Research On Earthquake Engineering, No. 200, Section 3, Xinhai Roadz,
Taipei 10668, Taiwan
e-mail: cmlin@ncree.narl.org.tw
C. H. Kuo
e-mail: chkuo@ncree.narl.org.tw
K. L. Wen
Department of Earth Sciences, National Central University, No. 300, Zhongda Road, Taoyuan
City 32001, Taiwan
e-mail: wenkl@cc.ncu.edu.tw
https://doi.org/10.1007/978-981-99-1459-3_32
398 S. Sharma et al.
horizontal components has been numerically integrated to find the size of SMGAs.
The displacement record obtained clearly shows the presence of two SMGAs. The
seismic moment (M o ) related to these SMGAs is (3.60 ± 0.82) × 1017 Nm and (4.55
± 0.7) × 1017 Nm, respectively.
Keywords 2019 Hualien earthquake · Strong motion generation area · Source

parameters
1 Introduction
Earthquakes have been occurring for millions of years around world’s seismically
active regions. The region facing a particular earthquake undergoes devastation and
loss of life depending on the intensity and magnitude of that earthquake. When an
earthquake occurs, a sudden release of energy propagates in the form of seismic waves
which traverse through the earth’s lithosphere. The shaking due to earthquakes being
recorded within the epicentral range of 50 km is known as strong ground motion.
Recently, the eastern coast of Taiwan was hit by a strong earthquake of (M w
6.2) on April 18, 2019, which produced a strong ground motion around Hualien
region. Figure 1 illustrates the observed accelerograms at the nearby stations used
for this study.
Actually, the Philippine Sea Plate (PSP) is subducting underneath Eurasian plate
(EP) with a velocity of 80 mm/yr [9]. This collision is the main reason behind long
history of earthquakes in Taiwan. Apart from this collision, there are several local
faults in western and southern Taiwan which play an efficient role in the frequent
earthquake activities in and around Taiwan.
According to the finite fault model, the earthquakes occur along the finite-sized
rupture plane. Strong ground motion (SGM) due to earthquakes occurs mainly along
particular patches within the entire rupture plane, known as strong motion generation
areas (SMGAs). These SMGAs are responsible for generating high frequency ground
shaking [6]. The presence of SMGAs is indicated by the spatio-temporal distribution
of aftershocks as well as visual inspection of the observed horizontal components
(NS and EW) of SGM. Visual inspection of SMGAs on the observed records can
be done properly using the secondary waves in purely horizontal plane in which
SH waves travel. The source parameters of the identified SMGAs play a vital role
in understanding the kinematical and dynamical properties of earthquake source.
The source displacement spectra of observed horizontal records of the identified
SMGAs have been used to study the source parameters like seismic moment, corner
frequency, and stress drop. Furthermore, the rupture responsible for this earthquake
propagated northward therefore, indicates a strong directivity effect [4].
Identification of Strong Motion Generation Area of the 2019 Hualien … 399
Fig. 1 The surface records and locations of the used near-field stations [8]
2 Geology and Seismotectonics of Hualien Region
An earthquake of moment magnitude (M w ) 6.2 hit the Hualien region lies on the
eastern coast of Taiwan (121.54 °E 24.06 °N) on April 18, 2019 (05:01:07 UTC).
The PSP is moving with the velocity of approximately 80 mm/yr. toward the EP [9].
The Coastal Range (CR) is situated on the eastern side of the Longitudinal Valley
(LV) and consists of solidifying andesitic magma and volcanic rocks originating from
the Philippine Sea Plate (PSP). Meanwhile, the Central Range (CR), located on the
western side of the LV, is composed of metamorphic terrane belonging to the Eurasian
Plate (EP). The collision between PSP and EP releases the accumulated stress which
leads to earthquake around Taiwan Island. Most of earthquakes in Taiwan occurs due
to active seismic faults located in the southern and western regions. However, the
Hualien region which lies on the eastern coast of Taiwan experiences earthquakes
due to the oblique subduction of the PSP underneath the EP [8]. This region still has
several faults such as the Milun and LV faults having a high rupture probability from
the Taiwan earthquake model (TEM) [1, 2, 7]. There are numerous earthquakes of
magnitude greater than 4 due to this subduction. The subduction of the Philippine
Sea Plate (PSP) underneath the Eurasian Plate (EP) occurs obliquely in the Hualien
region that leads to categorization of earthquakes into two types on the basis of
focal mechanisms: normal faulting shallow earthquakes caused by the bending of
the PSP, and west-dipping thrust shallow to deep focus earthquakes caused by the
subduction of the PSP. Therefore, focal mechanism can be the basis of clustering of
these earthquakes [3].
3 Data
The April 18, 2019 Hualien earthquake was felt throughout Taiwan and was detected
by 107 stations of the Central Weather Bureau (CWB). The ground shaking was
mainly strong near the source area, and Hualien county observed a maximum inten-
sity of 7 (> 400 Gal) on the Central Weather Bureau (CWB) intensity scale. The
region around northern Taiwan observed the shaking of intensity 5 (>80 Gal) [4].
This earthquake produced the highest intensity of 4 (>25 Gal) and duration of 10 s
in Taipei city. The station TWD, which is located at epicentral distance of 6 km,
recorded a peak ground acceleration of 110 Gal for the component in E-W direction
and 167 Gal for the component in N-S direction. The Station ETM, located 12 km
from the epicenter, recorded the highest peak ground acceleration with 379 Gal for
the component in E-W direction and 515 Gal for the component in N-S direction [8].
CWB developed and maintain Geophysical Database Management System (GDMS)
which provided the acceleration records utilized in this study. The preprocessing
steps like baseline correction and scaling has been applied on the observed records
according to the recording sensor. USGS and Global CMT have provided information
regarding the magnitude, location and fault plane solution of this earthquake.
4 Methodology
The Hualien earthquake was succeeded by many aftershocks between April 18, 2019,
and May 3, 2019. During the 15 days after the mainshock, a total of 1906 after-
shocks occurred along the eastern coast of Taiwan. The spatiotemporal distribution
of aftershocks has been shown in Fig. 2.
The network of accelerographs deployed by Central Weather Bureau (CWB)
recording strong-motion reported the spatio-temporal aftershocks distribution of the
2019 Hualien earthquake between April 18, 2019, and May 3, 2019, which indicates
the existence of SMGAs. The rupture plane of both identified SMGAs is shown by
two small rectangles on the rupture plane of the mainshock. The spatio-temporal
Fig. 2 Source model of Hualien Earthquake showing identified SMGAs within the plane of rupture.
The epicenter and aftershocks of this earthquake are represented by black star and solid circles,
respectively [8]
distribution of most of aftershocks are aligned along the nodal plane1, which has a
dip of 46° and strike of 215° associated with the mainshock. According to Lee et al.
[4], it was concluded that the plane of rupture of this earthquake dips towards west
direction.
The presence of SMGAs is also confirmed by the observed records on the surface.
The acceleration records on the surface shown in Fig. 1 has been used to compute the
displacement records for this earthquake. A double numerical integration has been
applied to the observed accelerograms to get the displacement records on the surface.
Figure 3 shows the displacement records which clearly confirms the presence of two
SMGAs.
The formula used to calculate the shear horizontal (SH component) of the
acceleration record (ASH ) at the stations is given as:
ASH = ANS cos δ + AEW sin δ (1)
where NS and EW component of the acceleration record are represented by ANS and
AEW . The angle δ denotes the anticlockwise between the ray path that joins the source
and the station and the NS (North-South) direction. The graphical representation of
numerical treatment for SH wave has been shown in Fig. 4. The station has been
shown in the figure by red triangle, and the source is represented by the red star in
Fig. 4.
Fig. 3 NS component of observed displacement records at near-field stations. Red ellipse shows
the envelope from SMGA1 and black ellipse shows the envelope from SMGA2 on the observed
displacement records
Fig. 4 Pictorial representation to compute shear horizontal component (ASH ) in the horizontal
plane using NS and EW components of observed records
Suitable numerical integration schemes are applied to transform the particle

motion due to propagation of SH wave to displacement records. Figure 5 shows
the obtained displacement records of SH components.
5 Analysis of Strong Motion Generation Areas
The clustering of aftershocks is observed in the spatiotemporal distribution that

results from two SMGAs positioned within the plane of rupture of this earthquake
which is shown in Fig. 2. Moreover, two distinct wave packets observed at a nearby
station to the epicenter specifies the presence of two SMGAs within the plane of
rupture that were accounted for the mainshock. Once the SMGAs have been identi-
fied both spatially as well as temporally, their properties in terms of seismic moment,
corner frequency, and stress drop are computed from the identified phases in the
observed records to know their role in the rupture process of the Hualien earthquake.
Fig. 5 SH component obtained after numerical treatment of horizontal components of observed

displacement records at near-field stations. Red ellipse shows the envelope from SMGA1 and black
ellipse shows the envelope from SMGA2 on the SH component records
Visual inspection of observed acceleration records suggests that second envelope

indicates SMGA having large dimensions than SMGA shown by first envelope.
Source parameters have been assessed by analyzing the source displacement spectra
(SDS) of the waveform that represents two SMGAs at different near-field stations.
Figures 6 and 7 show the SDS obtained from the phase denoting the SMGA1 and
SMGA2 of NS component of the observed records. The red line shows the theoretical
spectra fit on SDS.
The theoretical spectra fit on SDS helps to estimate the corner frequency (f c )
and low frequency spectral level (Ωo ) which are utilized to compute various source
Fig. 6 Source displacement spectra (SDS) obtained from the phase denoting the SMGA1 of NS
component of the observed records. The red line shows the theoretical spectra fit on SDS
Fig. 7 Source displacement spectra (SDS) obtained from the phase denoting the SMGA2 of NS
component of the observed records. The red line shows the theoretical spectra fit on SDS
parameters like seismic moment (M o ), stress drop (Δσ ), radius (r), and moment
magnitude (M w ) for both SMGAs as shown in Tables 1 and 2. The following equations
are used to calculate the source parameters of both SMGAs:

Mo = 4πρβ 3 Ω0 R / F S × Rθϕ × PRIT (2)
r = 2.34β / 2π f c (3)
Δσ = 7Mo / 16r 3 (4)
The symbols in the equation are defined as follows: ρ represents the density of the
medium, Rθϕ represents the radiation pattern coefficient of the S phase, β depicts the
velocity of secondary wave of the medium, free-surface effect is denoted by FS, PRIT
Table 1 Source parameters obtained from SDS from NS component of SMGA1 at various stations
Parameters M o (×1017 ) (Nm) Δσ (×105 ) (N/m2 ) f c (Hz) Radius (r) (km) Mw
EAH 0.61 10.63 0.43 2.93 5.12
EHP 4.3 70.3 0.42 3 5.69
ETL 2.9 46.8 0.42 3 5.57
TWD 2.3 40 0.43 2.93 5.51
ETM 2.3 40.7 0.43 2.93 5.51
ESL 9.9 13.94 0.4 3.15 5.27
EHY 1.1 18.76 0.42 3 5.31
WHF 6 88.76 0.4 3.15 5.8
ETLH 4.7 107.61 0.47 2.68 5.72
SHUL 2.2 30.07 0.39 3.23 5.51
Table 2 Source parameters obtained from SDS from NS component of SMGA2 at various stations
Parameters M o (×1017 ) (Nm) Δσ (×105 ) (N/m2 ) f c (Hz) Radius (r) (km) Mw
EAH 6.9 70.7 0.36 3.5 5.83
EHP 7.7 107.72 0.4 3.15 5.86
ETL 1.7 24.97 0.4 3.15 5.43
TWD 3.4 27 0.33 3.82 5.62
ETM 2.5 134.7 0.62 2.03 5.54
ESL 7.1 61 0.36 3.71 5.83
EHY 1.8 36.61 0.45 2.8 5.44
WHF 2.7 148.3 0.63 2 5.56
ETLH 6 97.6 0.42 3 5.79
SHUL 5.7 58.2 0.36 3.5 5.77
is partitioning of energy into two components in horizontal direction, long-term flat

level of SDS is represented by Ω 0 .
The average seismic moment (M o ), moment magnitude (M w ), stress drop (Δσ ),
radius and corner frequency (f c ) for SMGA1 are (3.60 ± 0.82) x 1017 Nm, 5.5 ± 0.06,
and (46.7 ± 9.77) x 105 N/m2 , 3 ± 0.04 km and 0.42 ± 0.006 Hz, respectively. The
average seismic moment (M o ), moment magnitude (M w ), stress drop (Δσ ), radius
and corner frequency (f c ) for SMGA2 are (4.55 ± 0.7) x 1017 Nm, 5.6 ± 0.05, (76.68
± 13.14) x 105 N/m2 , 3.06 ± 0.19 km and 0.43 ± 0.03 Hz respectively.
6 Discussions and Conclusions
It is assumed that the Hualien earthquake is triggered by the Milun fault due to
its focal mechanism and juxtaposition to the fault, which resulted in SGM along the
eastern coast of Taiwan [5]. The spatiotemporal distribution of aftershocks and visual
inspection of observed displacement records suggests the presence of two strong
motion generation area within the rupture plane responsible for this earthquake.
The numerical treatment of observed records produced the SH records at each near-
field station which confirms the presence of SMGAs. To investigate the rupture
mechanism of this earthquake, it is necessary to compute the source parameters of
the both identified SMGAs. The source displacement spectra of observed records
at various near-field stations have been computed, and low-frequency spectral level
(Ωo ) and corner frequency (fc ) have been estimated. The average seismic moment
(Mo ), moment magnitude (Mw ), stress drop (Δσ), radius and corner frequency (fc )
for SMGA1 are (3.60 ± 0.82) x 1017 Nm, 5.5 ± 0.06, and (46.7 ± 9.77) x 105
N/m2 , 3 ± 0.04 km and 0.42 ± 0.006 Hz, respectively. The average seismic moment
(Mo ), moment magnitude (Mw ), stress drop (Δσ), radius and corner frequency (fc )
for SMGA2 are (4.55 ± 0.7) x 1017 Nm, 5.6 ± 0.05, (76.68 ± 13.14) x 105 N/m2 ,
3.06 ± 0.19 km and 0.43 ± 0.03 Hz respectively.
Acknowledgements The acceleration waveform data and aftershocks data of the April 18, 2019
Hualien earthquake are provided by Geophysical Database Management System (GDMS). This is a
Web-based data service platform in Taiwan that is constructed by Central Weather Bureau (CWB).
The data provided by CWB for this work are highly acknowledged. The authors would like to offer
special thanks to Indian Institute of Technology Roorkee for the support required for the research
work shown in this paper. Project grant No. GITA/DST/TWN/P-75/2017 approved by Department
of Science and Technology (DST), Government of India has been highly acknowledged.
References
1. Chan, C.H., Wang, Y., Wang, Y.J., Lee, Y.T.: Seismic-hazard assessment over time: modeling
earthquakes in Taiwan. Bull. seism. Soc. Am. 107(5), 2342–2352 (2017)
2. Chan, C.H., Ma, K.F., Lee, Y.T., Wang, Y.J.: Rethinking seismic source model of probabilistic
hazard assessment in Taiwan after the 2018 Hualien, Taiwan, earthquake sequence. Seismol.
Res. Lett. 90(1), 88–96 (2019)
3. Kuo-chen, H., Vu, Y.M., Chang, C.H., Hu, J.C., Chen, W.S.: Relocation of eastern Taiwan
earthquakes and tectonic implications. Terr. Atmos. Ocean. Sci. 15, 647–666 (2004)
4. Lee, S.J., Wong, T.P., Liu, T.Y., Lin, T.C., Chen, C.T.: Strong ground motion over a large area
in northern Taiwan caused by the northward rupture directivity of the 2019 Hualien earthquake.
J. Asian Earth Sci. 192, 104095 (2020)
5. Mittal, H., Yang, B.M., Tseng, T.L., Wu, Y.M.: Importance of real-time PGV in terms of lead-
time and shakemaps: results using 2018 ML 6.2 & 2019 ML 6.3 Hualien, Taiwan earthquakes.
J. Asian Earth Sci. 220, 104936 (2021)
6. Miyake, H., Iwata, T., Irikura, K.: Source characterization for broadband ground-motion simula-
tion: Kinematic heterogeneous source model and strong motion generation area. Bull. Seismol.
Soc. Am. 93(6), 2531–2545 (2003)
7. Wang, Y.J., Chan, C.H., Lee, Y.T., Ma, K.F., Shyu, J.B.H., Rau, R.J., Cheng, C.T.: Probabilistic
seismic hazard assessment for Taiwan. Terr. Atmos. Ocean. Sci. 27(3), 325–340 (2016)
8. Sharma, S., Joshi, A., Sandeep et al. Modeling of rupture using strong motion generation area:
a case study of Hualien earthquake (Mw 6.1) occurred on April 18, 2019. Acta Geophys. 71,
1–28 (2023).
9. Smoczyk, G.M., Hayes, G.P., Hamburger, M.W., Benz, H.M., Villaseñor, A.H., Furlong, K.P.:
Seismicity of the Earth 1900–2012 Philippine Sea Plate and Vicinity (No. 2010-1083-M). US
Geological Survey (2013.
Role of SH Wave in the Mapping
of Shallow Subsurface
Jyoti Singh, A. Joshi, Saurabh Sharma, and Mohit Pandey
Abstract SH waves are directly related to soil matrix which makes SH wave refrac-
tion/reflection methods effective in site investigation. This paper presents the genera-
tion of purely SH wave by using a specially designed SH wave source; this generator
plate is made of aluminum metals, and the wave is generated by giving impact on the
generator plate in the horizontal direction, and it has nails on its base which keeps
it fixed to the ground so that only horizontal motion is recorded. Since the SH wave
propagates in a plane transverse to the ray path, this SH wave generator is kept fixed
in a direction perpendicular to the geophone spread. The seismic waves have been
recorded using geophones with a natural frequency of 10 and 4.5 Hz. The propaga-
tion of SH waves in shallow subsurface can be modeled by the finite-difference (FD)
method based on a staggered algorithm. The developed software for FD modeling
of the medium has been tested for SH wave propagation in a purely elastic medium
in terms of numerical stability, dispersion, and boundary conditions. The obtained
near-surface properties have been used as an input in the FD method to simulate SH
waves and to compare with the real-data records acquired in the region surrounding
North Almora Thrust (NAT).
Keywords Finite-difference modeling · MASW · Ground motion simulation
1 Introduction
Site effects can significantly affect the nature of strong ground motion. Local condi-
tions can generate large amplifications and important spatial variations of seismic
ground motion. These effects are of particular significance in the assessment of
seismic risk, in studies of microzonation and planning, and in the seismic design of
important facilities [4, 16]. The effect of soil conditions on the ground motion has
J. Singh (B) · A. Joshi · S. Sharma · M. Pandey

Department of Earth Sciences, Indian Institute of Technology Roorkee, Roorkee, Uttarakhand,
India
e-mail: jsingh2@es.iitr.ac.in
https://doi.org/10.1007/978-981-99-1459-3_33
412 J. Singh et al.
been observed in well-documented earthquakes [11, 19]; recent work has emphasized
the physical understanding of site effects so that quantitative predictions can be made
[2], but there is still a lack of numerical methods for dealing with the problem taking
into account source, path, and local conditions. As SH waves are directly related
to soil matrix, they can be analyzed separately from body waves and thus is very
effective in site investigation studies.
Finite-difference method (FDM) is one of the most popular methods used in
seismic modeling (e.g., [1, 5, 9, 14, 23]) and migration [3]. The FDM is capable to
handle a very complex seismic model.
The fact that surface SH waves are observed in nature has been used to infer that
the Earth’s crust is layered. Also, recent development in the oil industry, such as oil
sand exploration and monitoring, involves more and more multi-component surveys
[12]. The set of equations that is used for generating synthetic seismograms can be
solved using finite-difference (FDM) methods on the numerical grid of nodal points
both in space and time. Depending on the choice of nodal points, the grid scheme can
be classified into two broad categories: staggered and non-staggered. Virieux [22,
23] successfully used a staggered finite grid scheme in wave modeling phenomena.
The method is popularly known as explicit finite-difference time domain staggered
grid. In the staggered FDM, stress and particle velocity are defined at different points
of the grid. A staggered grid with the location of points defining particle velocity and
stress is shown in Fig. 1.
The finite-difference method has been used in this paper to simulate SH waves.
The material through which the wave propagates is assumed to be perfectly elastic,
isotropic, and homogeneous.
Fig. 1 Location of stress and particle velocity in the staggered grid used for SH wave modeling
Role of SH Wave in the Mapping of Shallow Subsurface 413
Fig. 2 Geological map of North Almora Thrust (NAT). Modified after [21]
2 Study Area
The Almora klippe has apparently formed by the erosion of a large synclinal fold
in the crystalline Munsiari sheet, tectonically underlying the Main Central Thrust
sheet of the Higher Himalaya [20, 21]. The NAT marks the northern boundary of
the Almora klippe; its southern margin is bound by the South Almora Thrust (SAT).
The study area is shown in Fig. 2 which maps the north Almora Thrust.
3 Data
Two sets of data have been acquired, one for calculating real-data records, and the
second set of data has been acquired to calculate near-surface properties, i.e., shear-
wave velocity structure and density have been used as an input in the FD method to
simulate the records.
The source used was a specially designed SH wave generator, and a sledgehammer
of 5 kg has been used for the survey. The specially designed source for SH wave is
made up of an aluminum plate; the whole source generator is made up of an aluminum
plate, and one piece makes the base of the plate with ground coupling spikes in it.
On both sides, there is an impact plate, and this base plate with spikes is fixed to the
ground so that no vertical motion is recorded. The source is effective in recording
the horizontal motion of the ground, i.e., SH waves and thus this specially designed
source is used to generate SH waves.
414 J. Singh et al.
Fig. 3 SH field setup by using specially designed SH wave source. Modified after [26]
MASW technique is employed for investigating shallow subsurface features based

on variation in the seismic velocity of the subsurface lithological unit [6–8, 10, 13,
17, 18]. The technique involved in estimating the shear-wave velocity and delineation
of horizontal and vertical variation of shallow surface material properties [15, 24]. A
linear array of 24 receivers (4.5 Hz vertical geophones) is deployed in a horizontal
96 m, ground spread, and connected to a multichannel seismograph recorder. Two
parameters source offset and receiver spacing are important for data acquisition, and
a SH wave source and a sledgehammer of 5 kg are used for generating the seismic
wave. The data were collected by using 24 geophones and a seismograph; this has
been used to generate a 2D shear-wave velocity model. A field setup for generating
and recording SH wave is used which is shown in Fig. 3.
One more set of data has been finally collected for real-data records of SH wave
using TROMINO as receiver and source as specially designed SH wave source on
which hammer has made impact from both sides on impact plates by fixing the
source at the middle of the profile. Data have been acquired by using TROMINO as
receivers and keeping the source as fixed and recorded at different positions of the
spread of the profile line by hitting on impact plates from both sides assuming one
side as negative (West to East) and the other as positive (East to West) side.
4 Methodology
In this study, 2D MASW has been used to collect the data by using OYO seis-
mograph and 24 geophones of 4.5 Hz which were deployed at 4-m intervals and
oriented in the perpendicular direction to wave propagation. Seismic waves have
been recorded using geophones and horizontal geophones in the OYO seismograph.
MASW processing involves the transformation of raw data from time to frequency
Fig. 4 An example of processing of the 2D MASW method. a Multichannel raw field data. b
Rayleigh wave phase velocities are extracted from field data in the F-K domain. c) Phase velocities
are inverted for a shear-wave velocity profile (vs. depth). d 2D S-wave velocity map generated from
geoplot. Modified after [25]
domain by Fourier analysis. The “energy accumulation pattern recognition tech-

nique” is employed to extract the fundamental mode Rayleigh wave for generating
the dispersion curve, which represents the relationship between Rayleigh wave, phase
velocity, and frequency. Finally, this extracted dispersion curve is used to calculate the
S-wave velocity variations with depth. To calculate velocity variations with a depth,
we match the theoretically developed dispersion curve and the field dispersion curve,
the model for which it best fits is the velocity layer model. These three subsurface
methods are used in combination to image and analyze the shallow subsurface fault
and its extension through the width and depth of the fault line. The data processing
steps for the data recorded in field are shown in Fig. 4 with an example of how data
have been processed for 2D MASW method.
This data have been processed in SeisImager software to calculate the 2D MASW
velocity model and density which is further used in the FD method to simulate the
SH waves.
The SH waves are generated by hitting the wave generator plate in the E-W
direction and vice-versa. As one shot was from one face and the other was from the
reverse side, these two shots at the same position are added to cancel out any vertical
component, and thus only the horizontal component is recorded.
5 2D Shear-Wave Velocity Model
In this section, the two-dimensional velocity model obtained using the active source
MASW method has been presented. The specially designed SH wave generator
has been used to generate the SH wave. Figure 5 shows the results obtained after
the inversion of dispersion curves obtained from experimental records. SeisImager
software has been used to calculate the velocity model. The analysis of Fig. 5 shows
416 J. Singh et al.
Fig. 5 Shear-wave velocity obtained from 2D MASW survey
that the shear-wave velocity in 30 m is varying from 260 to 570 m/s near the north
Almora Thrust.
6 Comparison of Experimental and Numerical Study
In this section, the results obtained from the experimental and numerical study have
been presented. To obtain the result using the MASW technique, a specially designed
SH wave generator as described above has been used. In order to generate the purely
SH wave, the shot has been taken from the impact from East–West and West–East
direction. The results obtained from these two shots have been further averaged
out to remove the effect of motion in the vertical direction. Figure 6a, b shows the
record obtained from the impact in E–W direction and W–E direction, respectively.
Figure 6c shows the record obtained after averaging the E–W and W–E recorded
components.
Figure 7 shows the comparison of results obtained from experimental and numer-
ical studies. The results from experimental and numerical studies are presented in
blue and black color, respectively. The analysis of Fig. 7 shows that the simulated
Fig. 6 a Records obtained from impact in E–W direction. b Records obtained from impact in W–E
direction. c Added E–W component of the profile
data are in good agreement with the recorded data. To further validate this argument,
the first arrival time from simulated and experimental records has been computed and
analyzed. Figure 8 shows the first arrival time with respect to offset for experimental
and numerical records. The analysis of the results shows that the results from the
numerical simulation are in good match with the experimental records.
418 J. Singh et al.
Fig. 7 Records obtained from observed data in McSEIS-SX and simulated from finite-difference
method using 2D shear-wave velocity calculated from refraction survey
7 Conclusions
In the present work, the subsurface velocity model near North Almora Thrust in
Garhwal Himalaya has been presented. Further, the results obtained from the exper-
imental and numerical studies have been matched and validated. In the numerical
study, the velocity model obtained from the experimental McSEIS-SX technique has
been used. An accurate second order in time and 12th order in space finite-difference
code have been used to obtain the numerical results. The results obtained from the
numerical study show a good match with the experimental study results. In order
to further validate the results, the first arrival times of both the real data and the
simulated data have been matched. It has been observed that the first arrival time
obtained from the numerical study matches well with the experimental study results.
In the conclusion of this work, it can be stated that the presented code can be effi-
ciently used in simulating the active source results. Further, it can be stated that the
developed instrument can be used in the generation of SH waves.
Fig. 8 Observed and simulated first arrival time with respect to different channel numbers
References
1. Bansal, R., Sen, M.K.: Finite-difference modelling of S-wave splitting in anisotropic media.
Geophys. Prospect. 56(3), 293–312 (2008)
3. Claerbout, J.F.: Imaging the Earth’s Interior, vol. 1. Blackwell scientific publications, Oxford
(1985)
4. Esteva, L.: Microzoning: models and reality. In: Proceeding of 6th World Conference on
Earthquake Engineering, pp. 27–43 (1977)
5. Etgen, J.T.: A tutorial on optimizing time domain finite-difference schemes: “Beyond Holberg.”
Stanford Explor. Proj. Report 129, 33–43 (2007)
6. Guadagno, F.M., Nunziata, C.: Seismic velocities of fractured carbonate rocks (Southern
Apennines, Italy). Geophys. J. Int. 113(3), 739–746 (1993)
7. Haberland, C., Maercklin, N., Kesten, D., Ryberg, T., Janssen, C., Agnon, A., Weber, M.,
Schulze, A., Qabbani, I., El-Kelani, R.: Shallow architecture of the Wadi Araba fault (Dead
Sea Transform) from high-resolution seismic investigations. Tectonophysics 432(1–4), 37–50
(2007)
8. Improta, L., Zollo, A., Bruno, P.P., Herrero, A., Villani, F.: High-resolution seismic tomography
across the 1980 (Ms 6.9) Southern Italy earthquake fault scarp. Geophys. Res. Lett. 30(10)
(2003)
420 J. Singh et al.
9. Igel, H., Weber, M.: SH-wave propagation in the whole mantle using high-order finite
differences. Geophys. Res. Lett. 22(6), 731–734 (1995)
10. Jeanne, P., Guglielmi, Y., Cappa, F.: Multiscale seismic signature of a small fault zone in a
carbonate reservoir: relationships between VP imaging, fault zone architecture and cohesion.
Tectonophysics 554, 185–201 (2012)
11. Jennings, P.C.: Engineering Features of the San Fernando earthquake of February 9, 1971
(1971)
12. Jiang, X.: A review of physical modelling and numerical simulation of long-term geological
storage of CO2 . Appl. Energy 88(11), 3557–3566 (2011)
13. Karastathis, V.K., Ganas, A., Makris, J., Papoulia, J., Dafnis, P., Gerolymatou, E., Drakatos,
G.: The application of shallow seismic techniques in the study of active faults: the Atalanti
normal fault, central Greece. J. Appl. Geophys. 62(3), 215–233 (2007)
14. Kelly, K.R., Ward, R.W., Treitel, S., Alford, R.M.: Synthetic seismograms: a finite-difference
approach. Geophysics 41(1), 2–27 (1976)
15. Park, C.B., Miller, R.D., Xia, J.: Multichannel analysis of surface waves. Geophysics 64(3),
800–808 (1999)
16. Ruiz, S.E.: Influencia de las condiciones locales en las características de los sismos. Universidad
Nacional Autónoma de México, Instituto de Ingeniería (1977)
17. Sanny, T.A., Sassa, K.: Detection of fault structure under a near-surface low velocity layer by
seismic tomography: synthetics studies. J. Appl. Geophys. 35(2–3), 117–131 (1996)
18. Sheley, D., Crosby, T., Zhou, M., Giacoma, J., Yu, J., He, R., Schuster, G.T.: 2-D seismic
trenching of colluvial wedges and faults. Tectonophysics 368(1–4), 51–69 (2003)
19. Sozen, M.A. (ed.): Engineering Report on the Caracas Earthquake of 29 July 1967. National
Academy of Sciences (1968)
20. Srivastava, P., Mitra, G.: Thrust geometries and deep structure of the outer and lesser Himalaya,
Kumaon and Garhwal (India): implications for evolution of the Himalayan fold-and-thrust belt.
Tectonics 13(1), 89–109 (1994)
21. Valdiya, K.S.: The two intracrustal boundary thrusts of the Himalaya. Tectonophysics 66(4),
323–348 (1980)
22. Virieux, J.: SH-wave propagation in heterogeneous media: velocity-stress finite-difference
method. Geophysics 49(11), 1933–1942 (1984)
23. Virieux, J.: P-SV wave propagation in heterogeneous media: velocity-stress finite-difference
method. Geophysics 51(4), 889–901 (1986)
24. Xia, J., Miller, R.D., Park, C.B.: Estimation of near-surface shear-wave velocity by inversion
of Rayleigh waves. Geophysics 64(3), 691–700 (1999)
25. Xia, J., Miller, R.D., Park, C.B., Ivanov, J.: Construction of 2-D vertical shear-wave velocity
field by the multichannel analysis of surface wave technique. In: 13th EEGS Symposium on the
Application of Geophysics to Engineering and Environmental Problems, pp. cp–200. European
Association of Geoscientists & Engineers (2000)
26. Yordkayhun, S., Sujitapan, C., Chalermyanont, T.: Joint analysis of shear wave velocity from
SH-wave refraction and MASW techniques for SPT-N estimation. Songklanakarin J. Sci.
Technol. 36(3) 2014
A Variational Mode Decomposition
Approach for Modal Identification
of Structures
S. Gupta and S. Kaloni
Abstract Modal identification describes the behavior of the structure under dynamic
loading conditions such as earthquake and wind. A newly developed variational mode
decomposition (VMD) method identifies the modal parameter of structures based
on dynamic responses. The VMD algorithm decomposes the acceleration response
from the structure under the earthquake excitation into a series of a finite number
of the mono-components. The decaying amplitude of the extracted modal responses
is then utilized to find the natural frequencies by using a fast Fourier transform.
Finally, the mode shape in the modal space for each decomposed modes in the
structure is recognized from the obtained modal response data. This paper presents the
applicability of the variational mode decomposition for modal identification purposes
and its comparison with the earlier developed Hilbert spectral analysis (HSA) and
empirical mode decomposition (EMD) method. To demonstrate the efficiency of
VMD algorithm, two case studies were considered. First, modal parameters of a
numerical model of a three-story shear-beam type building are computed using VMD
technique followed by an experimental study on multistory building model. The
results demonstrate that the proposed VMD approach can be easily identified the
modal parameter of building with better accuracy.
Keywords Variational mode decomposition · Empirical mode decomposition ·

Modal identification
S. Gupta (B) · S. Kaloni

Dept of Civil Engineering, NIT Uttarakhand, Srinagar, India
e-mail: mt20civ007@nituk.ac.in
S. Kaloni
e-mail: smitakaloni@nituk.ac.in
https://doi.org/10.1007/978-981-99-1459-3_34
422 S. Gupta and S. Kaloni
1 Introduction
In the fields of mechanical and aeronautical engineering, it is widely used, and there is
also a great scope in the health monitoring of structures and identify the spot damage
at the initial stage, and also, identifying the dynamic properties of civil structures
from ground data captured during strong ground shaking has been a research topic in
civil engineering for few decades. Techniques for modal identification are useful not
only for assessing the present conditions of the building, but also for enhancing future
performance, confirming retrofit processes, vibration serviceability assessment, and
validating predictive models. Research on determining the dynamic properties of
civil structures from the response observed during intense ground shaking has been
ongoing for more than three decades. The basic idea for damage detection based on
structural dynamics is variation in physical properties such as mass, damping, and
stiffness due to damage that causes modal properties such as frequency, damping
ratio, and mode shape would be detectable. Therefore, the variation in the modal
parameter implies the variation of the structure state of health, that is, damage.
Therefore, the modal identification is an experimental measurement and determi-
nation of the dynamic behavior of the structure. Since real-world structures may
exhibit nonlinear behavior even in their healthy state, it is very difficult to identify the
behavior of such structures by using traditional time domain and frequency domain
method. Over the past few decades, many modal information-based damage detec-
tion time–frequency techniques such as Hilbert-Huang transform (HHT), wavelet
transform (WT), short-time Fourier transform (STFT), synchro-squeezed wavelet
transform (SSWT), empirical mode decomposition (EMD), and local mean decom-
position (LMD) for structural damage identification have been developed. The WT
and HHT approaches, which have been widely used to identify the dynamic behavior
of structures subjected to earthquake excitation, may be the two most popular time-
frequency analysis tools among these time-frequency techniques. A newly developed
modal identification technique, namely, variational mode decomposition (VMD), has
been successfully utilized for estimation of modal parameter of the civil structures
under the dynamic loading condition in recent time.
2 Modal Identification Algorithm
Variational mode decomposition (VMD) proposed by Dragomiretskiy et al. [7] which

is non-recursive alternative to the EMD algorithm and applied to the modal identi-
fication of the structure systems. There are three main steps in the procedure of the
VMD algorithm: (1) The VMD is used to extract the natural frequencies and modal
response from a recorded acceleration response by applying fast Fourier transform
(FFT), (2) For each modal response, a linear least-squares fitting function is suggested
to determine the damping ratio from the response’s decaying amplitude of system, (3)
A Variational Mode Decomposition Approach for Modal Identification … 423
Mode shape vectors are obtained after repeating the same procedure for all available
sensors on the structure.
2.1 Extracting Modal Responses and Frequencies
The VMD is used to decompose an acceleration signal S(t) into a series of discrete
number of sub-signals (modes), sk (t), k = 1, 2, 3, …, K, each with a small spectral
bandwidth. Each sub-signal is supposed to be compacted around a respective central
frequency which can be determined along with the decomposition. The VMD decom-
position process is a constrained optimization problem that minimizes the sum of
the estimated mode bandwidths:
⎧ ⎫
k ⎬
k
⎨ j
k
min α ∂t δ(t) + ∗ s (t) e− j ωt
subjected to sk (t) = s(t)
{sk },{ω K }⎩ πt
k ⎭
k=1 k=1
where * indicates the convolution, α represents the variance of the noise, δ and k
indicate the Dirac function and number of sub-signal of response, respectively, {sk } =
{s1 , s2 , . . . sk } and {ωk } = {ω1 , ω2 , . . . ωk } are the modes and center frequencies,
respectively.
2.2 Identification of Modal Damping Ratios
Since most of the civil engineering structures have low inherent damping, the
damping ratios for each mode can be computed once the modal response has been
specified. The damping ratio can be calculated from the modal response’s decaying
amplitude. √
The modal damping ratios are calculated using the 1/ 2 Method, which has been
√ spectrum. Figure 1 depicts the damping ratio evaluation
modified to fit the Fourier
method utilizing the 1/ 2 method on a Fourier spectrum of one data. The nth modal
damping is calculated using f n as the nth modal natural frequency identified by the
half power bandwidth approach and the damping ratio is computed as:
1 δf
Damping ratio(ξ ) =
2 fn
√
where f is the bandwidth in frequency domain at X max / 2 method as half power
bandwidth method. X max is the amplitude of the nth modal natural frequency f n on
Fourier spectrum.
Fig. 1 Evaluation method of

each mode damping ratio
2.3 Identification of Mode Shapes
Using VMD, we examine the decomposed measured response into its constituent
parts. The mode shapes are then estimated using these separate parts. The ratio of
recognized harmonic sources from the recorded data at different floor levels obtained
by using VMD with the one corresponding to the higher energy level (generally
roof level) can be used to determine the corresponding mode shape vector. Now,
we considered the response of n-degree of freedom of structural system and its
constituent parts, as determined by VMD:
xk (t) = sk1 (t) + sk2 (t) + · · · + skm (t)
where ski (t) signifies the ith VMD recognized component of the kth signal and xk (t)
denotes the response of the kth degree of freedom of the system. We obtain the
following equation using the modal expansion theorem (with mode truncation):
xk (t) = ∅k1 q1 (t) + ∅k2 q2 (t) + · · · + ∅km qm (t)
Dynamic response, x = ∅ki qi (t), where ∅i is mode shape matrix at the ith floor
level and q represents the modal coordinate i = 1, 2, …, m

SK i (t)
∅i = mean
Sni (t)
SK i (t) denotes identified constitute component at the ith floor level and Sni =
normalized constituent components at i = 1, 2, …, m.
2.4 Modal Assurance Criteria (MAC)
Modal assurance criteria (MAC) are used to assess the degree of correlation between
the identified mode shape from analytical models and modes from the VMD method,
and hence, mode shape vectors are evaluated for accuracy using the modal assurance
criterion (MAC). It is easy to use, and there is no requirement of the formation of
matrix for the estimation of mode shape. The range of MAC value lies between 0 and
1, where value of 1 indicates the good consistency of mode shape, and if the value
is close to 0, the mode shapes are not consistent. It has been shown that the modal
assurance criterion is a straightforward statistical idea that has the potential to be a
very effective tool for experimental modal analysis.
|{ϕ A }rT {ϕx }q |2

MAC (r, q) =
{(ϕ A }rT {ϕ A }r )({ϕx }qT qT{ϕx }q )
2.5 Effect of Measurement Noise
Some simulated white Gaussian noises (i.e., 5 and 10% white noises) introduced to
the response of the structural system in order to investigate the performance of VMD
algorithm in noise signal processing. When VMD is employed for signal decom-
position, the same procedure is performed. The input noise signal is scaled using
simulated signals collected from experiments, and the signal is then decomposed
into the number of IMFS using VMD, and the modal parameters from these noisy
signals are found.
3 Numerical Study and Validation
VMD method attempts to split an acceleration signal into a number of mono-

component with gradually changing amplitudes and compacted around their respec-
tive center frequencies, so that when their cumulative modes are combined, then
responses reproduce the original acceleration response of the structural system.
To validate the proposed algorithm, two examples were presented. In first example,
three-story shear-beam type building model is to be considered. In the next example,
vibration of six-story building recorded by an accelerometer was studied. Table 1
shows the consistency and accuracy of frequencies between acquired from the present
algorithm and theoretical values. Additionally, the effectiveness of the suggested
technique was examined at various signal-to-noise ratio levels (SNR) ranges from
having low-to-high SNRs. The noisy signal in Fig. 2b was first introduced with
white Gaussian noise and then proceed with the VMD algorithm and identified the
Table 1 Natural frequencies and damping ratios of 3-DOF building model from the proposed
method and HSA method
Mode Theoretical value Identified value HSA method [9]
Frequency Damping Frequency Damping Frequency Damping
(Hz) ratio (%) (Hz) ratio (%) (Hz) ratio (%)
1 2.22 2.0 2.20 2.2 2.22 2.1
2 6.22 5.6 6.20 5.4 6.22 5.7
3 8.98 8.1 8.90 7.3 8.94 7.8
(a) (b)
Fig. 2 Acceleration impulse response of 3-DOF system: a response of third floor without noise, b
response with noises (5%), and response with noise (10%) of third floor
desirable modal parameters results. Table 2 lists the identified frequencies for low-
to-high SNRs, it is clear that all the natural frequencies were identified with efficient
and better accuracy.
3.1 Identification of 3-DOF Shear-Beam Type Building
First, a 3-DOF shear building structure was used to assess the capabilities of the
suggested method in structural identification problem. The following properties of
a three-story shear-beam building model are taken into consideration in which all
stories have the same mass, stiffness, and viscous damping with m i = 1000 kg, ki
= 980 KN/m, ci = 2.814 KN-s/m, respectively, (j = 1; 2; 3). Suppose an impact
loading is applied to the second floor. The analytical modal frequencies of shear
building were computed first, then compared to the experimental modal frequencies.
For a three-degree of freedom structure, the mass matrix M, stiffness matrix K, and
damping matrix C were formed.
The acceleration responses of each story are collected using the accelerometers
which are installed in the building and recorded the response. Now, recorded signals
are processed to identify the modal properties of the structures. Finally, the VMD
algorithm is used on the response in order to decompose the relevant IMFs and
find out the modal parameters. Figure shows the derived IMFs from the vibration
data without and with noise. After this, natural frequency of three-story building is
calculated by fast Fourier transform. The acceleration response of the building is
obtained with the help of time-stepped integration method (Newmark-beta method).
The acceleration response of the top floor is selected for the signal decomposition
which is shown in Figs. 2 and 4 (Tables 1 and 2).
By computing the modal assurance criterion (MAC), consistency for the first
three mode of the finite element model is checked and comparing them to those
estimated by the VMD and determined to be in the range [1.0–0.90]. The uniformity
of calculated mode shapes from vibration data is confirmed by large MAC values. The
mode shapes identified by the VMD processes are extremely similar and confirming
the accuracy of the mode shape of the structure system (Table 3; Fig. 5).
(a) (b)
Fig. 3 Decomposed IMFs using VMD corresponding to the response: a without noise, b with
noises (5%), and with c noise (10%) of third floor
(a) (b) (c)
Fig. 4 Fourier transforms: a Fourier transform of third floor without noises, b Fourier transform
with 5% noise and c Fourier transform with 10% noises to peak response ratio
Table 2 Natural frequencies and damping ratios of 3-DOF building model with noise 5 and 10%
Mode Identified value with 5% Identified value with 10% HSA method with 5%
noise noise noise [9]
Frequency Damping Frequency Damping Frequency Damping
(Hz) ratio (Hz) ratio (%) (Hz) ratio (%)
1 2.20 2.0 2.20 2.2 2.21 1.9
2 6.20 5.4 6.20 5.4 6.19 5.5
3 8.90 7.3 8.90 7.3 8.94 7.6
Table 3 MAC values for identified modes using the proposed method and the HSA method
Mode Identified values HSA method [9]
Noise = 0% Noise = 5% Noise = 10% Noise = 0% Noise = 5%
1 0.99 0.91 0.91 1.00 1.00
2 0.96 0.96 0.96 1.00 0.99
3 0.91 0.91 0.91 0.99 0.99
Mode Shape 1 Mode Shape 2 Mode Shape 3

3.5 3.5 3.5
3 3 3
Height of floor (m)
Height of floor (m)

Height of floor (m)
2.5 2.5 2.5

2 2 2
1.5 1.5 1.5
1 1 1
0.5 0.5 0.5
0 0 0
0 2 4 -2 0 2 -5 0 5
Fig. 5 Identified mode shape of the 3-story building
3.2 Identification of Multistory Building
A 6-story RC frame structure is considered with the following properties:

Beam size = 350 × 500 mm, Column size = 500 × 500 mm, Clear cover for beams
= 25 mm, Clear cover for columns = 40 mm, Shape of members = Rectangular,
Story Height = 3.3 m, Live loads = 4 kNm2
, Live load on terrace = 1 kN
m2
, Floor finish =
1 m2 , f ck = 25 mm2 , and f y = 415 mm2 . . An earthquake loading condition (PEER
kN N N
NGA strong ground data, Northridge-01, 1/17/1994, Tarzana—Cedar Hill A, 360)

is used as the input force on the considered building model.
Six accelerometer is employed to collect the response of each story of building
under the earthquake excitation. Finally, the VMD algorithm is used to proceed on
the response in order to produce the relevant IMFs. Fast Fourier transform (FFT) is
(a) (b) (c)
Fig. 6 Acceleration impulse response of 6-DOF system: a response of top floor without noise
pollution, b response with noises (5%), and response with noise (10%) of top floor
applied to each IMF of an acceleration response to transfer the modal response into
frequency domain, and then, modal parameters are collected.
Acceleration response of a building is obtained by time history analysis using
finite element software which is shown in Figs. 6 and 7 (Tables 4 and 5).
(a) (b) (c)
Fig. 7 Fourier transforms: a Fourier transform of top floor of 6-DOF system without noises, b
Fourier transform with 5% noise and c Fourier transform with 10% noises to peak response ratio
Table 4 Natural frequencies and damping ratios of 6-DOF building

Mode No. Frequency (Hz)
Theoretical Without noise Noise (5%) Noise (10%)
1 0.823 0.825 0.825 0.825
2 2.66 2.65 2.65 2.65
3 4.99 5.35 5.35 5.35
4 6.01 6.9 6.9 6.9
5 7.88 8.00 8.00 8.00
6 9.38 9.625 9.625 9.625
Table 5 Damping ratios of

Mode No. Damping ratio (%)
6-DOF building by propose
method Without noise Noise (5%) Noise (10%)
1 4.6 4.6 4.2
2 1.46 1.46 1.46
3 1.40 1.40 1.40
4 0.81 0.81 0.81
5 0.78 0.78 0.78
6 0.77 0.77 0.77
4 Conclusions
This study presents the variational mode decomposition technique as a potential

approach for modal identification of civil structures. The method uses the measured
free vibration time histories response and decomposed into the finite number of
mono-component (IMFs), and then, fast Fourier transform is used to determine modal
properties, such as natural frequencies and mode shapes of the decomposed modes.
When the response of each story of a structure is measured, the proposed algorithm
is efficient to identifying the mode shapes of the structure. The proposed method
allows for the extraction of natural frequencies and damping ratios of all decomposed
modes from one measurement only. The advantages of the proposed method for the
modal identification have been illustrated through a numerical example, and these
numerical simulations also considered the effect of different noise in the signal. The
identification outcomes from the suggested method were compared with those from
HSA method, and these results show the better performance of VMD method. The
proposed VMD-based modal identification method avoids characterizing individual
modes as signals with explicit IMFs, in contrast to existing decomposition methods
such as the HSA and EMD. Therefore, the methodology proposed in this research
provides a unique and effective tool for identifying linear MDOF structures’ dynamic
features.
References
1. Liu, S., Zhao, R., Yu, K., Zheng, B., Liao, B.: Output-only modal identification based on the
variational mode decomposition (VMD) framework. J. Sound Vib. 522, 116668 (2022)
2. Zhang, M., Xu, F.: Variational mode decomposition based modal parameter identification in
civil engineering. Front. Struct. Civ. Eng. 13(5), 1082–1094 (2019)
3. Yang, K., Wang, G., Dong, Y., Zhang, Q., Sang, L.: Early chatter identification based on an
optimized variational mode decomposition. Mech. Syst. Signal Process. 115, 238–254 (2019)
4. Bagheri, A., Ozbulut, O.E., Harris, D.K.: Structural system identification based on variational
mode decomposition. J. Sound Vib. 417, 182–197 (2018). https://doi.org/10.1016/j.jsv.2017.
12.014
5. Sadhu, A.: An integrated multivariate empirical mode decomposition method towards modal
identification of structures. J. Vib. Control 23(17), 2727–2741 (2017)
6. Kaloni, S., Shrikhande, M.: ScienceDirect Output Output only only system system identifica-
tion identification based based on on synchrosqueezed synchrosqueezed transform transform.
Procedia Eng. 199, 1002–1007 (2017). https://doi.org/10.1016/j.proeng.2017.09.235
7. Dragomiretskiy, K., Zosso, D.: Variational mode decomposition. IEEE Trans. Signal Process.
62(3), 531–544 (2014)
8. Nagarajaiah, S., Basu, B.: Output only modal identification and structural damage detection
using time frequency & wavelet techniques. Earthq. Eng. Eng. Vib. 8(4), 583–605 (2009)
9. Yang, J.N., Lei, Y., Pan, S., Huang, N.: System identification of linear structures based on
Hilbert–Huang spectral analysis. Part 1: normal modes. Earthq. Eng. Struct. Dyn. 32(9), 1443–
1467 (2003)
10. Paultre, P.: Dynamics of structures. Wiley & Sons (2013)
11. Yang, Y., Nagarajaiah, S.: Time-frequency blind source separation using independent compo-
nent analysis for output-only modal identification of highly damped structures. J. Struct. Eng.
139(10), 1780–1793 (2013)
12. Ni, P., Li, J., Hao, H., Xia, Y., Wang, X., Lee, J., Jung, K., Street, K., Hom, H., Kong, H.
(n.d.). Time-Varying System Identification using Variational Mode Decomposition. Vmd, 0–2.
https://doi.org/10.1002/stc.2175
13. Baker, J.W.: Measuring bias in structural response caused by ground motion scaling. In: Pacific
Conference on Earthquake Engineering, vol 56, pp 1–6 (2007). https://doi.org/10.1002/eqe
14. Sun, M.M., Li, Q. S., Zhou, K., He, Y.H., Zhi, L.H.: Modal identification from non-stationary
responses of high-rise buildings by variational mode decomposition and direct interpolation
techniques. Int. J. Struct. Stab. Dyn. 20(11) (2020). https://doi.org/10.1142/S02194554205
01151
15. Pastor, M., Binda, M., Harčarik, T.: Modal assurance criterion. Procedia Eng. 48, 543–548
(2012). https://doi.org/10.1016/j.proeng.2012.09.551
Geo-Factor Inference Modelling
with Empirical Susceptibility Weights
Approach for GIS-Based Seismic Hazard
Mapping of Thiruvananthapuram City
Madhu Mohan Velapgy and E. S. M. Suresh
Abstract Appropriate assessment of various site parameters influencing the spatial

distribution of seismic hazard was investigated to critically identify the suscep-
tible spots in the study area, Thiruvananthapuram city, a low to moderately active
seismic zone in southern Peninsular India (PI). The Geo-Factor Inference (GFI)
model developed in this study evaluates respective contribution level of local geo-
factors to estimate the peak ground acceleration (PGA), representing regional seismic
hazard. Considering the complexity and limitations in the widely accepted analytic
hierarchy process (AHP), mostly used in seismic studies, a simplified empirical
susceptibility weights approach (ESWA) was proposed to assess the attribute weights
for GFI model. The ESWA weights were mathematically calibrated by rank-order
centroid (ROC) tool and subsequently, compared with AHP weights to deduce vari-
ations in scale ranges. ESWA values from GFI were finally assigned to respective
geo-factor thematic layers for weighted overlay analysis in Geographical Informa-
tion Systems (GISs) platform. The solution from GFI model, integrated with GIS-
approach, provides distinctive distribution of susceptibility levels ranging from very
low to high in the seismic hazard mapping of the city.
Keywords Seismic hazard · Geo-factor inference (GFI) · Empirical susceptibility

weights approach (ESWA) · Rank-order centroid (ROC) · AHP
M. M. Velapgy (B)
Department of Civil Engineering, National Institute of Technical Teachers’ Training and
Research, Ministry of Education (MoE), Government of India, Chennai, India
e-mail: velapgymadhumohan@gmail.com
E. S. M. Suresh
Civil Engineering and Head-Department of Engineering Education, National Institute of
Technical Teachers’ Training and Research, MoE, Government of India, Chennai, India
https://doi.org/10.1007/978-981-99-1459-3_35
434 M. M. Velapgy and E. S. M. Suresh
1 Introduction
The seismic hazard of a site is mainly influenced by three interacting factors: seismic
wave characteristics of ground, local ground effects and built environment. However,
in most of the studies, only the severity of ground motion at a particular site is
commonly described by the seismic hazard without necessary consideration of conse-
quent factors [11]. Generally, the geo-spatial uncertainties of geological and geotech-
nical variances influence the seismicity of a region, and hence, it is essential to
characterise the site-specific geo-statistical assessments by trial-and-error, with a
high degree of spatial variability [10]. The present study focuses on the mapping
of seismic hazard of Thiruvananthapuram city, a low-seismicity area in the state of
Kerala in southern part of PI. This region is classified as Zone-III [4], indicating earth-
quake susceptibility of this city where majority of the buildings are low-to-high-rise
reinforced concrete construction. The seismic hazard map for this region is defined
in terms of PGA which is based on the earthquake magnitudes, signifying the ground
motion intensity [22] was selected as the single best alternative or target criterion. The
seven key site-related geo-parameters: geology, lineaments/faults, geomorphology,
elevation, slope, soil distribution and building density, influencing quantification of
ground motion were considered as sub-criteria. The GFI model developed for gener-
ating hazard map of the city attempts for a reasonable assessment of weights and
ranks of geo-factors to understand the respective contribution levels towards ground
motion characteristics of the site. Although the pair-wise comparison method, AHP
[18], is one of the well-known multi-criteria decision-making (MCDM) technique
to obtain the weights of different criteria in seismic studies, many researchers have
expressed their concerns on AHP, as a falsely precise weight elicitation method-
ology specifying its limitations, such as, ranking irregularities due to rank reversal
impossibility with multiplicative variant [21], interpretation issue of criteria weights
and difficulty in distinguishing amongst criteria due to artificial limitation of 9-point
scale. Considering these limitations and uncertainty to obtain a realistic solution even
after a lengthy AHP process, a simplified alternative method, empirical susceptibility
weights approach (ESWA) is proposed for weight assessment of selected geo-factors,
in this study. The ESWA weights of geo-factors were calibrated using an additive
multi-attribute method, rank-order centroid (ROC) for validating the proposed GFI
model, to ensure unbiased quantities and were further compared with corresponding
AHP weights to understand the scale of variations in both. The ESWA values were
finally assigned to respective thematic layers for weighted overlay analysis, using
GIS platform. This integrated GFI model with GIS-based method has a reason-
able analytical assessment capacity for providing distinctive distribution of suscep-
tibility scenarios, between very low to high levels in the seismic hazard mapping of
Thiruvananthapuram city.
Geo-Factor Inference Modelling with Empirical Susceptibility Weights … 435
2 Study Area and Geo-Factor Data
2.1 Regional Geographical and Geological Settings
Thiruvananthapuram (previously known as, Trivandrum) is a densely populated

capital city of Kerala state in India, which is one of the most important tourist
destinations in the world, with historical landmarks. The city is located at latitude
08° 30, N and longitude 76° 54, E covering an area of about 214.86 km2 , close to
Western Ghats, limited by the west coast of southern PI (Fig. 1). The toposheet from
Survey of India (SOI) was used for the base map and to locate built environment
in the city, using ArcGIS software. The western coastal tracts of geological part
form the city land profile, in the Southern Granulite Terrane (SGT) region of PI,
occupied largely by lateritised, tertiary and quaternary sedimentary formations of
Mio-Pliocene age. The SGT has three tectonic blocks separated by important shear
zone systems [7] known as, Cauvery Suture Zone (CSZ) and Achankovil Shear Zone
(AKSZ), separating Northern Granulite Block, Madurai Block and Trivandrum Block
(Kerala Khondalite Belt). Two major segments of Trivandrum and Nagercoil blocks
are located on the south of Achankovil fault (Fig. 2), revealing moderately active
seismicity of this area. Only two fault lines [9], Achankovil (N 60° W) and Kallada
(N 52° W) and five lineaments (medium and above) are close to the study area, which
are amongst those considered to be sensitive to the West Coast fault systems, identi-
fied near the study area, having direct effect on regional seismicity. The geological
datasets were based on Geology Map of Geological Survey of India (GSI), Chetty
et al. [7] and Thiruvananthapuram District Resource Map. Bhuvan’s resourcesat-2
data from the Linear Imaging Self-Scanning Sensor (LISS-III) with 23.5 m spatial
resolution were used for mapping the geology and lineament layers.
2.2 Regional Geomorphological, Geotechnical Aspects

and Seismicity
The geomorphological features of the city have a physiographically rugged topog-

raphy with three distinctive topographic units as, lowland (coastal plain), midland
and highland terrain, from west to east. Geomorphology map of GSI and Thiruvanan-
thapuram District Resource Map, Bhuvan’s resourcesat-2 data, were referred. The
ground terrain profile of this region largely constitutes rocks, with highly heteroge-
neous local soils varying from, red loams, coastal and riverside alluvium, lateritic
soil, brown hydromorphic soil and forest loam. Soil map of SOI-School of Atlas and
Geotechnical Investigation Reports of various locations in the city were used. Eleva-
tion and slope details were obtained from relief and slope map of SOI and Shuttle
Radar Topographic Mission (SRTM)-DEM data with a resolution of about 30 m. The
SOI, DEM and Google Maps were used, for sourcing building data. Many researchers
have scientifically established that, the collision of Indian tectonic plate with Asia
Fig. 1 Location map of study area [3]
Fig. 2 Geological features of SGT in PI (GSI & [7])

plate had resulted in a compression effect on the Indian crust. In consequence to this
collision effect, the PI shield experiences substantial earthquake hazards, due to the
existence of many critically loaded faults [19]. The seismic hazard evaluated from
the sizes (magnitudes) of potentially damaging earthquakes, occurring at multiple
locations, is uncertain in most of the situations [13]. The catalogue of historical
and instrumentally recorded earthquakes for the region was obtained from national
and international organizations including, India Meteorological Department (IMD)-
previously, Bhuvan-National Remote Sensing Centre (NRSC), National Centre for
Earth Science Studies (NCESS), Geological Survey of India (GSI), National Centre
for Seismology (NCS), Amateur Seismic Centre (ASC), International Seismological
Centre (ISC), International Seismological Summary (ISS) and United States Geolog-
ical Survey (USGS). Several compilations from published literatures [6, 16] were
used for historical earthquakes, which also include significant earthquakes of equiv-
alent moment magnitudes (M w ≥ 4.0), after 1950. The experiences of past seismic
studies showed that a site ground vibrates due to earthquakes originate anywhere
around 300 km radius, about the site. However, considering the influence of global
seismic activity rate on the regional seismic parameters and also due to the sparsely
available data within this radius, the seismic datasets within the lager geographical
influence zone of 500 km radius around Thiruvananthapuram city were collected
[12] in this study, which were reasonably informative, after eliminating duplicate
events. The maximum equivalent earthquake moment magnitude, M w = 5.8 (Coim-
batore, 1900), within the influence zone had been considered for the target criterion
in this investigation. Ganesha et al. [9], identified 22 events close to the 9 major
lineaments/faults near this region, amongst 31 earthquakes (with equivalent moment
magnitudes, 5 nos., M w > 5 and 26 nos., M w > 3) occurred in Kerala, since 1821.
These fault lines with lengths more than 20 km indicate possible correlation between
the fault systems and earthquakes in this zone.
3 Methodology and Data Analysis for GFI Model
Hazard mapping of the study area involves subjective assignment of weight values
for each of the influencing factors related to corresponding thematic layers and inte-
grating multiple overlays of various criteria and analysed using ArcGIS software, to
evaluate the target criteria. Based on the pilot survey and factor analysis, the main
ground-associated parameters including geology (GL), lineament density/fault distri-
bution (LD), geomorphology (GM), soil distribution (SD), elevation (EL), slope (SL)
and building density (BD) which contribute to the quantification of ground motion
were considered as sub-criteria set, for the study region. The decision about weight
determination of respective contributing factors has a high influence in the final
estimation of regional seismic hazard [17]. The most reasonable weights of identi-
fied geo-factors, specifically, GL, LD, GM, SD, EL, SL and BD, pertaining to the
study area were assessed using a three-step procedure. ESWA weights were initially
assigned, followed by testing and calibration of these values by ROC weighting
method, which were subsequently compared with AHP weights, to ascertain the
reasonably acceptable weights related to each of the selected GIS layers’ criterion.
The degree of belief in evolving GFI model will mainly depend on how well they
can represent the weight scenarios of ESWA values, which were validated by ROC
weights, with the similar constraints. The ESWA weights were assigned according
to the seismic sensitivity of respective layers, and the classes indicate the associated
seismic risk level attributed to each layer, regarded from very low to high, the GIS
layers were ranked, analysed, reclassified into a common scale of influence by the
importance of ESWA weights and finally, overlaid using weighted overlay tool in
ArcGIS software to generate the seismic hazard map of city.
3.1 Empirical Susceptibility Weights Approach (ESWA)

in GIS Layers
The ESWA methodology is a simplified, qualitative, non-iterative, mathematically

tested and calibrated weight assessment model developed for estimating the rational
contribution levels of seven identified key geo-factors (thematic layers) or sub-criteria
towards the evaluation of PGA (target criterion) proposed for the hazard assess-
ment for the study area, considering the limited number of attributes within the
smaller area extent of low-seismicity site. The parameters influencing the seismic
risk were studied for different level of susceptibility in the seismic hazard zone
[8]. Normally, the higher weight is assigned to the thematic layer which contributes
more to the hazard. Many previous investigations established the geology and soil
attributes, as the influencing factors for the earthquake events [23]. Based on the
earlier studies, the highest weights were assigned to the local geology and associated
lineament density factors, which are recognised as the main contributing attributes
to the target criterion representing regional seismicity. These identified geo-factors
were weighed sequentially, on percentage-basis with reasonable judgements on the
damage levels of these attributes, based on the understanding from the various avail-
able data sources, including site-related datasets, literatures from earlier research
studies, reliable internet information and expert opinions. A questionnaire along
with a query for comparison of each geo-factor (to get the geometric mean for the
final solution) was formulated and conducted to twelve expert opinions from varied
engineering disciplines, including civil, geotechnical, geology, geoinformatics, seis-
mology and geophysics for the pilot survey. Accordingly, the ESWA weights of the
selected geo-factors were empirically assigned as, geology (35%), lineament distri-
bution (20%), geomorphology (15%), soil distribution (10%), elevation (10%), slope
(5%) and building distribution (5%), according to the respective influence levels on
PGA, aggregated to 100% of cumulative for the whole sub-criteria set.
3.2 Testing and Calibration of ESWA Weights with ROC

Weighting
The ESWA geo-factor weights were validated using ROC weighting method, which
is a simple and reliable additive MAV weighting procedure for eliminating the possi-
bility of biased values and to ensure the appropriateness of these values in the concep-
tualisation of GFI model. It geometrically characterises the centroid of feasible
weight-space for each of the sub-criteria elements, providing reasonably accurate and
acceptable quantities. Understandably, the biases exist for several weight judgments,
and precise weight elicitation is subject to inconsistencies, which can be resolved
by the reconciliation method to get the final weights, based on ranking of attribute
ranges with an effective rank-based formula [5]. The additive multi-attribute ROC
weights, a surrogate weighting method [2] is a more systematic analysis of infor-
mation implicit in the ranks for computing the ROC weights, which are derived
from the simplex vertices (extreme points), w1 ≥ w2 ≥ … ≥ wn ≥ 0, limited to
Σ n
i=1 wi = 1 and indicated by the feasible weight-space, δ n . The vertices of the
simplex are defined as, v1 = (1, 0, …, 0), v2 = (1/2,1/2, 0, …, 0), vn = (1/n, 1/n,
…, 1/n). The respective coordinates of defining vertices are averaged to obtain the
coordinates of centroid
Σ(weights), and hence, the centroid weight for the most impor-
n
tant element is, 1/n i=1 1/i. The centroid weight corresponding to the ith most
important attribute, w(i,(ROC)) , is calculated using Eq. (2), given below:
Σ
n
w(i, (ROC)) = 1/n 1/j , where, i = 1, . . . , n (1)
j=i
Generally, the ROC weights order the alternatives according to average or expected
MAV, which is the largest. In this study, the allotted ranks and computed weights for
the ROC weights and corresponding ESWA weights, for each of the attributes in the
sub-criteria set, are given in Table 1, and the calibration graph Fig. 3a illustrates the
validity of the accuracy and consistency levels of ESWA weights.
3.3 Comparison of ESWA Weights with AHP
The calibrated ESWA weights were further compared with the corresponding AHP
values (scale 1–9) of geo-factor layers to understand the variations. By comparing
the alternatives with respect to a subjective criterion, a pair-wise comparison matrix
is formed first in AHP at each level of hierarchy (from second to the last level),
followed by the computation of relative weight for each parameter. After estimating
AHP weights, the consistency ratio (CR) was checked for each theme, which is equal
to 0.072, satisfying the required consistency (CR < 0.1), as shown in Table 2. It is to
be noted that the weights for the respective parameters, both in the proposed ESWA
Table 1 Comparison of ESWA weights for testing with corresponding ROC weights
Sub-criteria Weights by ESWA (%) ROC weights
Allotted ranks Weights Weights (%)
Geology 35 1 0.37 37.04
Lineament distribution 20 2 0.23 22.76
Geomorphology 15 3 0.16 15.61
Soil distribution 10 4 0.11 10.85
Elevation 10 5 0.07 7.28
Slope 5 6 0.04 4.42
Building distribution 5 7 0.02 2.04
Total 1.00 100.00
Fig. 3 a Calibration of ESWA geo-factor weights by ROC weighting; b Comparison of geo-factor

weights between ESWA and AHP methods
and in the well-accepted AHP methods, are evaluated based only on the subjective
judgements. Figure 3b shows the deviations in the resulting weights between ESWA
weights and corresponding normalised weight of each layer in AHP.
4 GIS Data for Spatial Analysis
The ESWA weights were assigned based on the tangible influence levels of respective
layers towards the PGA, and the classes specify the related seismic susceptibility of
each layer. After determining the value ranges for geo-factors, the ranking of class
values is assigned from 7 to 9 based on the regulations and assumptions of relevant
datasets from various disciplines of the attributes, using the standard equal interval
classification to emphasise the amount of an attribute value relative to other values,
class breaks and to set appropriate ranges of classes. Accordingly, seven classes
were considered for GL, five for GM and nine for LD, SD, EL, SL and BD layers
Table 2 Summary of pair-wise comparison matrix of layers in AHP
Sub-Criteria Geo-Factors Geology Lineament distribution Geomorphology Soil distribution Elevation Slope Building distribution Total
Normalised weight (wi ) 0.385 0.195 0.159 0.112 0.081 0.039 0.029 1
Order of the comparison matrix, n 7
Largest or principal eigenvalue of the matrix, λmax 7.582
Consistency index, CI = (λmax − n)/(n − 1) 0.097
Mean principal eigenvalue of the matrix, λmax = 2.7699n − 4.3513 15.038
Average random consistency index, RI = (λmax − n)/(n − 1) 1.339
Consistency ratio, CR = CI/RI 0.072
Geo-Factor Inference Modelling with Empirical Susceptibility Weights …
441
in GIS analysis. The GL, GM, SD and BD geo-factor layers were digitised based
on symbols and visual interpretation from related input maps. The lineament/fault
distances were determined for LD, using Euclidean distance tool, and the lineament
density distribution per unit area is calculated using density command in the ArcGIS
spatial analyst extension. The areas of high risk for SL were having the slope values
above 5%, and the scale of elevation ranges vary from <8 m to 174 m for EL, based
on altitude-level risks. The study area falls in the UTM-Projection_43P zone, and
the output cell size made for all the layers was 30 m, generated uniformly using the
same projection and output cell size. The reclassified values in the input raster layers
were entered into a common evaluation scale of risk or some similarly unifying scale,
multiplying the cell values of each input raster by the raster weight of importance.
5.1 Reliability of ESWA Weights for the GFI Model
Figure 3a shows the calibration graph of ESWA weights after testing with ROC
weights, which clearly illustrates the closeness and reliability of ESWA geo-factor
weights with ROC weights, demonstrating unbiased indicators of severity of the iden-
tified parameters. The ESWA weights for the highly influential parameters, particu-
larly GL, GM, SD and SL, are almost matching with those of ROC weights. However,
the deviations of ESWA weights with reference to ROC weights for LD, EL and BD
are notably attributed to the variations in the profile with assigned values of ESWA
polygonal, as against the polynomial profile of ROC weights. Hence, we highlight
that the assigned ESWA weights can provide reasonably accurate and reliable values
of GFI model for the study area, without undergoing any complex approximation
procedures, as in the case of conventional MCDM techniques. From Fig. 3b, the
influence of geomorphology, geology and soil distribution parameters calculated
by AHP was seen to be slightly on the higher side (6.13–12.00%), and lineament
distribution is marginally lower (2.50%), in comparison to related ESWA values.
However, the attribute weights corresponding to the elevation and slope are appre-
ciably varying with lower values of 19.40% and 22.60%, respectively, whereas the
resulting weight of building distribution factor in AHP is very much under-estimated
(41.40%), compared to that of ESWA.
5.2 Weighted Overlay Analysis
Using weighted overlay analysis tool, the resulting cell values of seven thematic map
layers were added together to generate the output raster in GIS, and finally, the seismic
hazard map was produced to visualise the contribution level of geo-factors in the city
Fig. 4 Geology
area. The output maps (Figs. 4, 5, 6, 7, 8 and 9) demonstrate that the overall effect
of identified geo-factors on PGA is the highest in a small portion in the north-west
and lowest in central and western parts of the city. Based on the GFI-GIS integrated
seismic hazard map (Fig. 9), four categories of potential risk zones had been identified
as, very low, low, moderate and high representing risk-level variations of local parts of
study region. The hazard map in Fig. 9 illustrates that, the central midland portion with
dense, low-to-high-rise built environment, including residential, commercial, office,
infra-structure, essential services and ground-transportation buildings, lies in the
very-low to low-level hazard parcel. The north-west city area sparingly displays high-
level hazard, where high-rise buildings are located in large numbers. The densely
populated north and south-east portion falls in the low-level hazard zone, in which,
large pockets of shallow buildings are present. The low-level hazard parcel in the
south-west part of the city has airport and shallow constructions.
5.3 Estimation of PGA for Susceptible Hazard Spot of Study

Area
The seismic hazard map generated in this study shows that the coastal areas in the
north-west part of the city are more susceptible and can be regarded as high-risk
zone, compared to the other parts of the city. Considering maximum earthquake
magnitude, M w = 5.8, the upper bound PGA value for this zone was estimated as
0.0476 g at bed-rock level, using attenuation Eq. (2) by Raghu Kanth and Iyengar
[15] for PI:
Fig. 5 Lineament density
Fig. 6 Geomorphology
ln(PGA) = 1.6858 + 0.9241 (Mw − 6) − 0.0760 (Mw − 6)2 − 0.0057 (R) − ln(R) + ln(∈) (2)
where PGA—refers to PGA (in g) at bed-rock level, R—hypocentral distance (in

km) and σ (lnε) = 0.4648. However, an optimally recommended value of 0.13 g
has been set in the initial estimate of PGA at rock level for Thiruvananthapuram
region, based on the investigations by Sitharam et al. [20], which defines the PGA
values corresponding to the site classifications for major cities in the country, as per
Fig. 7 Elevation
Fig. 8 Slope
National Earthquake Hazards Reduction Program (NEHRP), which is higher than

the value corresponding to the soil classification for the high-risk zone of study area.
Fig. 9 GIS seismic hazard

map
5.4 Comparison with Other Studies
Nath et al. [14] assigned weights for governing geo-hazard factors of geological
themes based on AHP as, surface geology (33.33%), soil cover/site class (26.66%),
slope (20.00%), landslide (13.33%) and rock outcrop (6.66%) for investigation in
Sikkim Himalaya and Guwahati city regions for the seismic hazard in Northeast
India, in which, the priorities set for the geology are closely matching with the
present study. However, PGA and associated attributes were included in the seismo-
logical themes and overlaid with the geological themes for the generation of AHP
integrated seismic microzonation map. Al-Dogom et al. [1] determined the weights
of geostatistical factors as, PGA(30%), distance from faults (30%), slope (5%), soil
(15%) and geology (20%) using AHP techniques with GIS mapping to investigate
their effect on the United Arab Emirates (UAE), considering the earthquake events
(1900–2015) throughout the Arabian plate. In the UAE study, where two major fault
systems, Zagros Fold and Makran subduction zone, which were considered to have
direct influence on seismicity of UAE, were included as the distance from faults
parameter, whereas fault/ lineament density had been included in this study, consid-
ering presently dormant clusters of these geological lines in the study zone. Differing
from the earlier researches, PGA which is the result of the other attributes is consid-
ered as the evaluation (target) criteria in the present study, whereas amplification
factor, shear wave velocity, liquefaction potential, etc., were regarded as tertiary
components, which are the seismicity modifiers which depend on the geophysical
factors: geology, geomorphology, soil (geotechnical), slope, etc. From the overall
comparisons with the other previous research studies, which were not included in this
discussion, although many of the researchers considered varied parameters, detailed
investigations indicate that the magnitude of local seismicity (causative factor) forms
the main component for evaluating PGA in most of the previous investigations,
whereas the attributes related to ground motion amplification, shear wave, liquefac-
tion and attenuation effects, which are the responses during earthquakes (reactive
factors), significantly modify the characteristics of target criterion (PGA) and can be
regarded as the site response modifiers. Hence, the target criteria, main component
and tertiary components were not included in GIS thematic layers, for the present
study. The reasons for these deviations in the factorisation need further investigations,
based on the local site conditions and sufficiently accurate records.
6 Conclusions
The GFI model developed in this study for Thiruvananthapuram city evaluated the
contribution levels of seven local geo-factors to estimate PGA of the region. In view
of the limitations in the well-known AHP weight assessment, the proposed simplified
ESWA weights for the GFI model were empirically evaluated as, geology (35%),
lineament distribution (20%), geomorphology (15%), soil distribution (10%), eleva-
tion (10%), slope (5%) and building distribution (5%), which represent the respec-
tive susceptibility levels towards PGA, considering the limited number of attributes
within the smaller area extent of low-seismicity site. The ESWA weights were tested
and calibrated by the mathematical tool, ROC weighting method and subsequently,
compared with AHP weights to understand the scale range variations. ESWA values
from GFI model were finally assigned to the respective geo-factor thematic layers,
for weighted overlay analysis in GIS for the mapping of study area. The solution from
the integrated GFI model with GIS-approach provides distinctive spatial distribution
of susceptibility levels ranging from very low to high, in the seismic hazard mapping
of the city. The main advantage of this integrated GFI model using ESWA weights
with GIS-based seismic hazard mapping with respect to other methodologies is that,
it includes pre-assigned levels or geo-factor weights, without any exhaustive weight
elicitation processes for identifying the most susceptible zone in the smaller study
area and to apply comprehensive attenuation relation corresponding only to the crit-
ical part of the city for estimating PGA. The city seismic hazard map shows that the
coastal areas in the north-west of city are more susceptible and regarded as high-risk
zone, compared to the other parts of the city, and the corresponding PGA for this part
of the city was estimated as, 0.0476 g at bed-rock level, which is lower than (0.13 g).
However, the accuracy of results in this study depends on the correctness of used
data and inferences from earlier studies in other areas and the proposed methodology
for the computation of influence weights.
References
1. Al-Dogom, D., Schuckma, K., Al-Ruzouq, R.: Geostatistical seismic analysis and hazard
assessment; United Arab Emirates. Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. ISPRS
Arch. 42(3W4) (2018). https://doi.org/10.5194/isprs-archives-XLII-3-W4-29-2018
2. Barron, H.F.: Selecting a best multiattribute alternative with partial information about
attribute weights. Acta Physiol. (Oxf) 80(1–3), 91–103 (1992). https://doi.org/10.1016/0001-
6918(92)90042-C
3. Bhuvan, NRSC-ISRO: Open Data Archive. National Remote Sensing Centre, India (2020).
https://bhuvanapp1.nrsc.gov.in/bhuvan2d/bhuvan/. Accessed 20 June 2020
4. Bureau of Indian Standards: IS 1893: 2016—Indian Standard Criteria for Earthquake Resistant
Design of Structures, Part 1—General Provisions and Buildings. Bureau of Indian Standards,
1893(December) (2016)
5. Borcherding, K., Schmeer, S., Weber, M.: Biases in Multiattribute Weight Elicitation. Caverni,
Jean Paul [Ed], pp. 3–28 (1995)
6. Chandra, U.: Earthquakes of Peninsular India—a seismotectonic study. Bull. Seismol. Soc.
Am. 67(5), 1387–1413 (1977)
7. Chetty, T.R.K., Yellappa, T., Santosh, M.: Crustal architecture and tectonic evolution of the
Cauvery Suture Zone, southern India. J. Asian Earth Sci. 130, 166–191 (2016). https://doi.org/
10.1016/j.jseaes.2016.07.003
8. Ganapathy, G.P., Rajawat, A.S.: Quantification of geologic hazard and vulnerability for Chennai
city, India. Int. J. Geomatics Geosci. 5(1), 32–42 (2014)
9. Ganesha Raj, K., Paul, M.A., Hegde, V.S., Nijagunappa, R.: Lineaments and seismicity of
Kerala—a remote sensing based analysis. J. Indian Soc. Remote Sens. 29(4), 203–211 (2001).
https://doi.org/10.1007/BF02995725
10. Kim, H.S., Sun, C.G., Cho, H.I.: Geospatial big data-based geostatistical zonation of seismic
site effects in Seoul metropolitan area. ISPRS Int. J. Geo Inf. 6(6), 174–191 (2017). https://
doi.org/10.3390/ijgi6060174
11. Kramer, S.L.: Geotechnical Earthquake Engineering. Prentice-Hall Civil Engineering and
Engineering Mechanics Series (1996). http://faculty.washington.edu/kramer/GEEbook.pdf
12. Madhu Mohan, V., Suresh, E.S.M.: Probabilistic seismic hazard assessment model for
GIS-based seismic risk study of Thiruvananthapuram City. In: Disaster Resilience and
Sustainability: Adaptation for Sustainable Development (2021)
13. Mohan, V.M., Suresh, E.S.M.: Seismic vulnerability assessment of existing buildings in India
using GIS techniques. In: 38th Asian Conference on Remote Sensing—Space Applications:
Touching Human Lives, ACRS 2017 (2017)
14. Nath, S.K., Thingbaijam, K.K.S., Raj, A.: Earthquake hazard in Northeast India—A seismic
microzonation approach with typical case studies from Sikkim Himalaya and Guwahati city.
J. Earth Syst. Sci. 117(SUPPL.2), 809–831 (2008)
15. Raghu Kanth, S.T.G., Iyengar, R.N.: Estimation of seismic spectral acceleration in Peninsular
India. J. Earth Syst. Sci. 116(3), 199–214 (2007). https://doi.org/10.1007/s12040-007-0020-8
16. Rajendran, C.P., John, B., Sreekumari, K., Rajendran, K.: Reassessing the earthquake hazard
in Kerala based on the historical and current seismicity. J. Geol. Soc. India 73(6), 785–802
(2009). https://doi.org/10.1007/s12594-009-0063-3
17. Roszkowska, E.: Rank ordering criteria weighting methods—a comparative overview.
Optimum. Studia Ekonomiczne 5:14–33 (2013). https://repozytorium.uwb.edu.pl/jspui/bitstr
eam/11320/2189/1/02_Ewa%20ROSZKOWSKA.pdf
18. Saaty, R.W.: The Analytic Hierarchy Process: Planning, Priority Setting, Resource Allocation
(Decision Making Series). Mathematical Modelling (1980)
19. Seeber, L., Armbruster, L.G., Jacob, K.H.: Probabilistic Assessment of Earthquake Hazard for
the State of Maharastra, p. 60. The Government of Maharastra Earthquake Rehabilitation Cell,
Mumbai, India (1999)
20. Sitharam, T.G., Kolathayar, S., Vipin, K.S.: Evaluation of seismic hazards for India: determin-
istic approach including local site effects. In: Indian Geotechnical Conference (2010). https://
gndec.ac.in/~igs/ldh/conf/2010/articles/047.pdf
21. Triantaphyllou, E.: Two new cases of rank reversals when the AHP and some of its additive
variants are used that do not occur with the multiplicative AHP. J. Multi-Criteria Decis. Anal.
10(1), 11–25 (2001). https://doi.org/10.1002/mcda.284
22. USGS: Earthquake Hazards (2020). https://www.usgs.gov/natural-hazards/earthquake-haz
ards/
23. Yagoub, M.M.: Spatio-temporal and hazard mapping of Earthquake in UAE (1984–2012):
Remote sensing and GIS application. Geoenviron. mental Disasters 2:13 (2015). https://doi.
org/10.1186/s40677-015-0020-y, Springer Nature, pp. 1–14
Octave Tool for Probabilistic Seismic
Hazard Assessment
D. R. Majhi and M. Shrikhande
Abstract Earthquake-resistant design of structures is aimed at withstanding a

certain level of ground shaking with repairable damage and avoid structural collapse
in more severe sharking. The design ground motion for a particular region is defined
by estimated seismic hazard which may be presented in various formats. The uncer-
tainties in seismic source characterization, lower and upper bound magnitudes, earth-
quake recurrence laws, and ground motion prediction equations play an essential
role in developing the seismic hazard curves. This study provides a simple and
effective way for estimating seismic hazard curves using the GNU-Octave plat-
form while accounting for all uncertainties. GNU-Octave is a free and open-source
numerical computing platform. For the implementation, three regions from seismic
zone IV (namely, Delhi, Patna and Darjeeling) and zone V (namely, Guwahati and
Mandi) are selected as per the IS-1893. The earthquake catalogues are collected
from various sources, mainly from the United States Geological Survey (USGS).
Before using them to estimate the seismicity using the Gutenberg–Richter (G-R)
recurrence relationship, these raw catalogue entries need to undergo some treatment
(specifically, homogenization, declustering, and completeness check). Because of
its simpler format and limited number of parameters, the ground motion prediction
equation by Boore and Atkinson is utilized for ground motion intensity measure
(IM). The seismicity of the Darjeeling region, which is categorized in zone IV of
IS-1893, is found to be miscalculated based on the hazard curves for these locations,
putting the structure in this region at greater risk.
Keywords Seismic hazard · Structural design · GNU-Octave · Seismicity
D. R. Majhi (B)
Civil Engineering Department, IIT Bombay, Mumbai, India
e-mail: diptiranjanmajhi@gmail.com
M. Shrikhande
e-mail: manish.shrikhande@eq.iitr.ac.in
https://doi.org/10.1007/978-981-99-1459-3_36
452 D. R. Majhi and M. Shrikhande
1 Introduction
Structural design guidelines ensured a certain level of safety for various structural
forms. These guidelines also take into account a range of loading and material
property uncertainties. The uncertainty around the earthquake occurrences and its
probable variations in the earthquake intensity and their energy distribution with
frequencies of earthquake ground motions make designing for earthquake effects
challenging. Thus, it is rational to predict multiple acceptable performance thresholds
(or safety margins) for different earthquake types. The dual-level design philosophy
is the basis of the current seismic design standards, which are intended to survive
moderate earthquakes with repairable damage as well as to avoid structural collapse
in case of severe earthquakes. Thus, in the event of strong earthquakes, some degrees
of damage are designed into the system.
The probability that a structural response will not exceed the specified maximum
response during the intended use time period is known as structural reliability. The
inverse of structural reliability defines its probability of failure. The limit state
functions are described in terms of global or local structural damage. Maximum
inter-story drift, roof displacement, maximum shear force and bending moment,
hysteretic energy, life cycle cost, and ductility demand are some example of the
damage levels are mostly considered for formulating the reliability framework [1–6].
Though the estimation of reliability for earthquake-resistant design is a challenging
task, however, there are few methods are well established for quantifying structural
reliability in the past literatures [7]. One of the major complications is the variability
in earthquake intensity.
Generally, seismic reliability calculations depend on probability of structural
response exceeding certain threshold and seismic hazard in a region. The primary
criterion for assessing structural reliability is seismic hazard uncertainty. Therefore,
it is crucial to carefully adhere to the seismic hazard estimation [8].
A free and open-source platform for numerical computing is GNU-Octave [9].
This platform was utilized in this study to calculate seismic hazard curves that account
for all potential uncertainties. This platform’s advantage is that it is quick, easy, and
user-friendly, allowing users to adjust their programs to suit their needs.
2 Seismic Hazard Analysis
The seismic hazard curve at a particular location is influenced by the probability of

an intensity measure (IM) exceeding the limiting value (IM*), that is conditioned on
an earthquake magnitude (MW ), source-to-site distance (R) or epicentral distance,
and the corresponding probabilities considering all related uncertainties. To account
for all earthquake ground motion uncertainties, these hazard curves are constructed
using multiple seismic source characterization, lower and upper bound magnitudes,
earthquake recurrence laws, and ground motion prediction equations (GMPEs).
Octave Tool for Probabilistic Seismic Hazard Assessment 453
2.1 Earthquake Catalogue and Treatment
In order to estimate seismicity at a particular location, it is necessary to collect a

large database of earthquake events. The database can be compiled from numerous
sources, such as United States Geological Survey (USGS) [10], International Seis-
mological Centre (ISC) [11], India Meteorological Department (IMD) [12], National
Geophysical Research Institute (NGRI) [13], Wadia Institute of Himalayan Geology
(WIHG) [14], Institute of Seismological Research (ISR) [15], and Indian Institute of
Technology (IIT), Roorkee [16]. The catalogue includes details such as the time and
date of origin, the exact location in terms of their longitude and latitude, earthquake
magnitude, epicentral distances, and some other parameters. These raw catalogue
records require some processing before being used to calculate the seismicity using
the Gutenberg–Richter (G-R) recurrence relationship (specifically, homogenization,
declustering, and completeness check).
Homogenization Treatment
The term “homogenization of earthquake catalogue” refers to the transformation of
various earthquake magnitude types into a single magnitude type utilizing appropriate
conversion relations. Considerable research has suggested relationships between
various earthquake magnitudes. Each event assigned a unique earthquake magnitude
based on these interrelationships. Since the beginning of the instrumental period in
1964, each event has produced an earthquake of a distinct magnitude. Therefore,
conversion is not necessary. The only conversion required was before the events
of 1964. Since the majority of the available attenuation relations are defined in
terms of moment magnitude, the earthquake magnitude of each event is taken into
consideration in terms of moment magnitude.
Declustering Treatment
In the declustering procedure, foreshock and aftershock sequences must be deleted
from the earthquake catalogue in order to be consistent with the assumption that
earthquake occurrences occur with a Poisson distribution (i.e., independently). The
early aftershocks are simple to distinguish since their frequency is relatively higher
than the background seismicity only immediately following a mainshock. Since there
is no physical distinction between the later aftershocks and the low magnitude main-
shocks, it becomes considerably more challenging to precisely detect the later after-
shocks. Therefore, it is impossible to perform an earthquake catalogue declustering
completely error-free. There are numerous different approaches have been proposed
for this reason over the years by various researchers. Many of these techniques are
briefly reviewed by Molchan and Dmitrieva (1992) [17].
Completeness Check of Earthquake Catalogue
The time frame for comprehensive recording (completeness check) for each earth-
quake magnitude range at any particular region must also be determined. This dura-
tion is basically based upon the consistency of each earthquake occurrence rate
with time. Because earlier recordings were unable to capture smaller earthquakes,
thus the catalogue is incomplete with regard to earthquakes of smaller magnitude
earthquakes. For accurate estimation of seismic hazard, it is thus necessary to use
the available data in different magnitude ranges only for the period for which that
magnitude range can be considered to have been recorded completely.
2.2 Gutenberg–Richter (G-R) Recurrence Law
Seismicity parameters for a particular region can be estimated using Gutenberg–

Richter (G-R) recurrence relationship [18] and given as:
log λ M = a − bMW (1)
where MW is the moment magnitude and λ M is the mean yearly rate of exceedance
of MW . The mean yearly earthquakes with a magnitude higher than or equal to 0
are given by 10a , and the ‘b’ value indicates how likely it is for major and minor
earthquakes to occur in different regions. Small earthquakes are believed to occur
more frequently when the ‘b’ value is large, whereas in case of larger earthquakes the
‘b’ value is relatively small. By applying the least square method and the maximum
likelihood method [19] to fit data points, the values of ‘a’ and ‘b’ are estimated for
each region.
The G-R recurrence relationship covers the entire range of earthquake magni-
tudes [−∞, +∞]. For engineering purpose, the smaller magnitude earthquake will
not cause significant damage to structures. Therefore, the lower bound value for
earthquake magnitude is set to be 4.0 or 5.0 while the maximum earthquake magni-
tude is specified according to the specific site. The probability density function (PDF)
and cumulative density function (CDF) for the G-R recurrence relation with upper
and lower bound can be expressed by:
βexp[−β(MW − M0 )]
f M (MW ) = (2)
1 − exp[−β(Mmax − M0 )]
1 − exp[−β(MW − M0 )]
FM (MW ) = (3)
1 − exp[−β(Mmax − M0 ]
where f M (MW ) and FM (MW ) represent the PDF and CDF of earthquake magnitudes,
respectively. Mmax and M0 are the upper and lower bound or threshold of earthquake
magnitudes. β signifies a constant with a value of 2.303b.
2.3 Identify Earthquake Distances
Estimating the distribution of earthquake distances or epicentral distances for a loca-

tion of interest is also important in order to predict ground shaking at that location.
It is typically considered that earthquakes will happen equally likely at any point
along the fault for a given earthquake source [20]. It is normally simple to determine
the distribution of source-to-site distances using the geometry of the source since
locations are equally distributed. The source zone can be considered to be a point
source, line source, and 3D volumetric source, and the corresponding distances can
be computed relatively.
2.4 Predictive Relationships
Predictive relationships are typically identified empirically by applying least-squares

regression on a specific set of strong motion parameter data. The size of an earthquake,
the distance from source to site, and some other parameters (if provided) must be
taken into account in order to quantify the uncertainty in IM. For an earthquake
of a given magnitude, MW , and occurring at a given distance, R, the chance of IM
exceeding a specific value, IM∗ , is given by:

P IM > IM∗ |MW , R = 1 − FIM IM∗ (4)
where FIM (IM∗ ) is the cumulative distribution function (CDF) of IM at MW and

R, obtained from the attenuation relations given by various researchers [21–23]
considering a set of strong motion parameters.
2.5 Seismic Hazard Curves and Mean Annual Rate

of Occurrences
The seismic hazard rate in the Poisson model is used to evaluate the probability of
exceedance of IM for specific time periods, or the exposed period of the system (also
called the structural design life) and the mathematical relationship is given by:

P IM > IM∗ |MW , R = 1 − exp(−λIM∗ t) (5)
where λIM∗ is the exceedance rate (or, probability) of IM∗ for the region of interest
and t is the exposed period of the structure.

λIM∗ = ν ∫ ∫ P I M > IM∗ |MW , R f M (MW ) f R (R)dr dMW (6)
MW R
ν = 10(a−bM0 )
where M0 is the minimum earthquake magnitude for a particular source location and
ν is the mean exceedance rate of M0 .
3 Regional Seismicity and Seismic Hazard Curves
There are four different seismicity zones on the seismic zoning map of the Indian
seismic code [24], namely zones II, III, IV, and V. In the revision of 2002, seismic
zones I and II were combined into a single zone II. These seismic zones do not even
explicitly take into account the regional seismicity because they are mostly based on
the iso-seismal contours of prior large earthquakes [25]. Only zones IV and V are
taken into account in this study due to their higher probability of severe earthquakes.
In zone IV, the regions of Delhi, Patna, and Darjeeling and Guwahati and Mandi for
zone V are selected.
These sites are spread out quite far from one another throughout the Himalayan
and sub-Himalayan region in order to cover a large geographic area.
A large database is compiled from numerous sources, and the majority of the
events are taken from the United States Geological Survey (USGS) earthquake cata-
logue. These catalogues are used to establish seismicity parameters for these areas by
taking into account all earthquake events that occurred within a 300 km radius of the
center latitude and longitude for Delhi (28.7041 °N, 77.1025 °E), Patna (25.5941 °N,
85.1376 °E), and Darjeeling (27.0360 °N, 88.2627 °E) for zone IV, and Guwahati
(26.1445 °N, 91.7362 °E), and Mandi (31.5892 °N, 76.9182 °E) for zone V. Since
USGS earthquake catalogue gives moment magnitude for each event thus no require-
ment for homogenization catalogue treatment. For declustering and completeness
check of earthquake catalogue, the window method [26] and slope method [27] are
used, respectively. The map of complete catalogue and corresponding regions for
different magnitude ranges are shown in Fig. 1. The results of the completeness study
shown in Table 1, and these values are used to estimate the earthquake occurrence
rates for G-R parameters.
Seismicity parameters (‘a,’ ‘b’) of specified regions of interest are calculated by
G-R recurrence relationship given in Eq. (1) and shown in Table 2 and Fig. 2. The
number of catalogue entries after their treatment within the radius of 300 km is also
given in Table 1.
Table 1 shows, ‘a’ and ‘b’ values for Delhi and Darjeeling (zone IV regions) are
alike to the Mandi (zone V region). The frequency and magnitude of earthquakes in
Darjeeling and Mandi are substantially different from those in Delhi. As shown in
Fig. 2, the G-R plots for the Darjeeling and Mandi regions nearly coincide, while the
G-R line for the Delhi region is rather off-centered. This is due to the significantly
lower ‘a’ value for the Delhi region, which was calculated from a very limited number
of minor earthquake events. The lower ‘a’ and ‘b’ values for Guwahati region indicate
Fig. 1 Earthquake events and different regions chosen in India. Map Source Basemap from ArcGIS
online
Table 1 Completion years for magnitude ranges

Sites Magnitude ranges
4 ≥ MW < 5 5 ≥ MW < 6 MW ≥ 6
Delhi 58 58 118
Patna 58 58 118
Darjeeling 58 68 83
Guwahati 58 68 118
Mandi 58 68 118
Table 2 Seismicity parameters for all the selected regions

Delhi Patna Darjeeling Guwahati Mandi
(85 events) (76 events) (269 events) (456 events) (296 events)
a 5.48 4.28 5.58 4.60 5.89
b 1.24 0.99 1.14 0.89 1.20
relatively smaller seismicity, but a higher likelihood of experiencing large earthquake

events in the North-eastern part of India. An earlier study that used data collected
from the seismicity catalogues of the Indian Society of Earthquake Technology and
United States Geological Survey reported similar values for the G-R parameters
for the Guwahati region (a ~ 4.60, b ~ 0.84) [28]. Therefore, the seismicity of the
Fig. 2 G-R relation curve fitting
Darjeeling region, which is categorized in IS-1893 zone IV, is miscalculated, causing

a higher risk to the structures designed and constructed in this location. With only
85 and 76 (relatively low magnitude) events in the catalogue, respectively, in Delhi
and Patna region, both indicate a lower level of seismicity than Darjeeling region in
zone IV.
The ground motion prediction equation by Boore and Atkinson [20] is used here
for ground motion intensity measure (IM). In the following, the expected peak ground
acceleration (PGA) is used as an intensity measure of an earthquake. Figure 3 esti-
mates and shows the mean yearly rate of exceeding the intensity measure (IM) for all
regions. Earthquake hazard in Darjeeling region is found to be significantly higher
than that of other locations in the same seismic zone (Zone IV). The estimated prob-
ability of exceeding IM, taking into account the exposure time of 50 years, is shown
in Fig. 4.
(a) Zone IV (b) Zone V
Fig. 3 Mean annual exceedance rate of IM for selected locations

(a) Zone IV (b) Zone V
Fig. 4 Exceedance probability of IM (PGA) for all the regions for the design life of 50 year
4 Conclusions
In this study, the GNU-Octave platform is used to perform the probabilistic seismic
hazard analysis. This platform is easy to use, quick running time, and also allowing
users to adjust their codes to meet their requirements. To estimate seismic hazard,
three regions in zone IV (Delhi, Patna and Darjeeling) and two in zone V (Guwahati
and Mandi) of IS-1893 are taken as the study area. The earthquake catalogues are
compiled from multiple sources, followed the catalogue treatments such as homog-
enization, declustering, and completeness check. The seismicity of all the regions
is evaluated using G-R recurrence relationships, and it is found that the seismicity
of the Darjeeling region, which is categorized in zone IV of IS-1893, is miscalcu-
lated, causing a higher risk to the structures that are designed and built in that region.
This study also estimates the mean rate of exceedance of intensity measure consid-
ering all the uncertainties in earthquake magnitude, source-site distance, and possible
occurrence of future earthquake. Further, it estimates the probability of earthquake
occurrence for a 50 year exposed period of structure. These estimates may also be
used in the computation of structural reliability.
References
1. Gaxiola Camacho, J.R., Azizsoltani, H., Villegas Mercado, F.J., Haldar, A.: A novel reliability
technique for implementation of performance-based seismic design of structures. Eng. Struct.
142, 137–147 (2017)
2. Ghobarah, A.: Performance-based design in earthquake engineering: state of development.
Eng. Struct. 23(8), 878–884 (2001)
3. Haukaas, T.: Unified reliability and design optimization for earthquake engineering. Probabilist.
Eng. Mech. 23(4), 471–481 (2008)
4. Collins, K.R., Wen, Y.K., Foutch, D.A.: Dual-level seismic design: a reliability-based
methodology. Earthq. Eng. Struct. Dyn. 25(12), 1433–1467 (1996)
5. Ghosh, S., Collins, K.R.: Merging energy-based design criteria and reliability-based methods:
exploring a new concept. Earthq. Eng. Struct. Dyn. 35(13), 1677–1698 (2006)
6. Lu DG, Song PY, Yu XH, Wang, G.Y.: Global seismic reliability analysis of building structures
based on system-level limit states. In: Proceedings of 14th World Conference on Earthquake
Engineering. Beijing, China (2008)
7. Ditlevsen, O., Madsen, H.O.: Structural Reliability Methods. Wiley and Sons Ltd (1996)
8. Majhi, D.R., Shrikhande, M.: Residual life of earthquake damaged structures. Soil Dyn. Earthq.
Eng. 145, 106694 (2021)
9. GNU-Octave: https://octave.org/download. Last accessed 1 Aug 2022
10. United States geological Survey (USGS): https://earthquake.usgs.gov/earthquakes/search/.
Last accessed 1 Aug 2022
11. International Seismological Centre (ISC): http://www.isc.ac.uk/. Last accessed 1 Aug 2022
12. India Meteorological Department (IMD): https://www.imdtvm.gov.in/index.php?option=com_
content&task=view&id=30&Itemid=44. Last accessed 1 Aug 2022
13. National Geophysical Research Institute (NGRI): https://www.ngri.res.in/. Last accessed 1 Aug
2022
14. Wadia Institute of Himalayan Geology (WIHG): https://waics.wihg.res.in/seismology. Last
accessed 1 Aug 2022
15. Institute of Seismological Research (ISR): https://isr.gujarat.gov.in/. Last accessed 1 Aug 2022
16. Indian Institute of Technology (IIT), Roorkee: https://pesmos.org/. Last accessed 1 Aug 2022
17. Molchan, G.M., Dmitrieva, O.E.: Aftershock identification: methods and new approaches.
Geophys. J. Int. 109, 501–516 (1992)
34(4), 185–188 (1944)
19. Weichert, D.H.: Estimation of the earthquake recurrence parameters for unequal observation
periods for different magnitudes. Bull. Seismol. Soc. Am. 70(4), 1337–1346 (1980)
20. Baker JW.: An Introduction to Probabilistic Seismic Hazard Analysis (PSHA), White
Paper, Version 1.3. https://web.stanford.edu/~bakerjw/Publications/Baker_(2008)_Intro_to_
PSHA_v1_3.pdf. Last accessed 1 Aug 2022
21. Boore, D.M., Atkinson, G.M.: Boore-Atkinson NGA Ground Motion Relations for the
Geometric Mean Horizontal Component of Peak and Spectral Ground Motion Parameters,
PEER Report 2007/01. Pacific Earthquake Engineering Research Center, Berkeley, California
(2007)
22. Campbell, K.W., Bozorgnia, Y.: A ground motion prediction equation for the horizontal compo-
nent of cumulative absolute velocity (CAV) based on the PEER-NGA strong motion database.
Earthq. Spectra 26(3), 635–650 (2010)
23. Abrahamson, N., Silva, W.: Summary of the Abrahamson & Silva NGA ground-motion
relations. Earthq. Spectra 24(1), 67–97 (2008)
24. IS 1893. Part 1: Indian Standard Criteria for Earthquake Resistant Design of Structures, part 1
General Provisions and Buildings. Bureau of Indian Standards, New Delhi (2016)
25. Agarwal, P., Shrikhande, M.: Earthquake resistant design of structures. PHI Learning Pvt. Ltd.,
New Delhi (2006)
26. Uhrhammer, R.A.: Characteristics of northern and central California seismicity. Earthq Notes
57(1), 21 (1986)
27. Mulargia, F., Tinti, S.: Seismic sample areas defined from incomplete catalogs: an application
to the Eastern Sicily. Tectonophysics 116, 335–364 (1985)
28. Das, S., Gupta, V.K., Gupta, I.D..: Codal provisions of seismic hazard in Northeast India. Curr
Sci 2004–2008 (2005)
Site Amplification Study Using Strong
Motion Data Recorded at Various
Stations in India from Far-Field
Earthquakes
Sireesha Jaladi, Babita Sharma, Himanshu Mittal, and O. P. Mishra
Abstract Given the importance of site effects, the site response characteristics in the
present work are evaluated at different locations of India using recorded data of three
earthquakes that occurred in different parts of the country. The used earthquakes
include the 2015 Afghanistan earthquake (Mw 7.5); 2016 Hindukush earthquake
(Mw 6.7) and 2016 Manipur earthquake of (Mw 6.7). The results are presented in
form of peak ground acceleration (PGA), peak ground velocity (PGV), amplification
and predominant frequency. For both Afghanistan and Hindukush earthquakes, the
highest and lowest PGA values are observed at SMLA and HYB stations, respectively.
For the Manipur earthquake, the highest PGA value of the order of 103 cm/s2 was
observed at the SHL station. In the case of the Afghanistan earthquake, the PGV
values did not show much variation, as observed in the Manipur earthquake. The
local soil effects at horizontal components of ground motion are strongly amplified
compared to the vertical components. It is found that three sites, namely SMLA,
SHL and KOHMA, are amplified in the frequency range 8–10 Hz, exhibiting less
amplification and may be treated as rock sites, or stiff-soil sites. The other sites,
namely BHPL, HYB, JORHT and LEKHA, show amplification at a lower frequency
range of 0.1–2.9 Hz with amplification a little higher than the previous three sites. The
shift of peak amplification from higher to lower frequency range indicates that these
sites lie in soft soil regions. The estimated predominant frequency and amplification
values in the present work may be very important for the valuation of seismic hazard
assessment of the region of study. The site amplification characteristics play an
important role in altering the ground motions recorded at different locations, and
because of this, the damage may vary widely.
S. Jaladi (B) · B. Sharma · H. Mittal · O. P. Mishra

National Center for Seismology, Ministry of Earth Sciences, New Delhi 110003, India
e-mail: sireesha21find@gmail.com
https://doi.org/10.1007/978-981-99-1459-3_37
462 S. Jaladi et al.
Keywords Site amplification · Seismic hazard assessment · H/V ratio
1 Introduction
The Indian subcontinent has a history of devastating earthquakes. The major reason
for the high frequency and intensity of the earthquakes is that the Indian plate is
driving into Asia at a rate of approximately 47 mm/year [1]. Geographical statistics
of India show that almost 54% of the land is vulnerable to earthquakes. The latest
version of the earthquake zoning map of India [2] divides India into 4 seismic zones
(Zones 2, 3, 4 and 5) unlike its previous version, which consisted of five or six zones
for the country. According to the present zoning map, Zone 5 expects the highest
level of seismicity whereas Zone 2 is associated with the lowest level of seismicity.
The Indian plate is continuously moving in the NE direction and colliding with
the Eurasian plate, and as a result of the collision, the state of stress in the Indian plate
is high, which in turn increases the earthquake hazard, particularly in northern India
along the Himalaya collision zone. This process has given rise to three major thrust
planes, the Main Central Thrust (MCT), the Main Boundary Thrust (MBT) and the
Main Frontal Thrust (MFT) [4]. The region has experienced several great earthquakes
in the past hundred years or so (1897 Assam,1905 Kangra; 1934 Bihar-Nepal; 1950
Assam) [1]. During the last episode of strain release, a 750-km-long segment, which
lies between the eastern edge of the 1905 rupture zone and the western edge of the
1934 earthquake, remained unbroken. This segment, called the central seismic gap,
continues to be under high strain (Fig. 1).
Fig. 1 Himalayan Arc region and the devastating past earthquakes in this zone [7]
Site Amplification Study Using Strong Motion Data Recorded … 463
Large earthquakes occurred in the seismic gap in 1803 and 1833, but the magni-
tudes of these earthquakes were less than 8, and hence, they were not gap-filling
events. Based on these considerations and a shortening rate of 20 mm/year across
the Himalayas, Khattri [7] has estimated the probability of occurrence of a great Mw
8.5 earthquake in the gap in the next 100 year to be 0.59. The north-eastern region
of India is also regarded as one of the most seismically active regions worldwide. A
seismic gap called the northeast seismic gap exists between the source zone of the
1950 Assam earthquake and the 1934 Bihar-Nepal earthquake where no major earth-
quake occurred in recent time. The past seismicity data from 1897 shows that the
north-eastern region has experienced two great earthquakes with magnitudes above
8.0 and about 20 large earthquakes with magnitudes varying between 8.0 and 7.0
[6]. Taking it into consideration the Indian subcontinent is at risk due to hazardous
earthquakes from the Indian-Eurasian collision. So, it is important to study the site
amplification using recorded earthquakes at various sites in India.
It is evident that the collision of Indian and Eurasian plates created Hindu Kush
subduction zone and it is consisting of metamorphic rocks such as schist, gneiss and
marble, as well as intrusive such as granite and diorite of different ages and sizes.
The Hindu Kush range remains geologically active and is still rising, and it is prone
to earthquakes. Afghanistan is underlain by Precambrian metamorphic rocks which
form high regions in the canter of the country and in the Hindu Kush. The swell is
connected to the Hindu Kush by Precambrian rocks. The geological, topographic and
structural characteristics have been evaluated for Manipur, a state of India. It lies in
the north-eastern part of the country. It is surrounded by Myanmar (a neighbouring
country) in the east and south-east, and by other adjacent states of India, namely
Assam in the west, Nagaland in the North and Mizoram in the south-west.
The site amplification characteristics play an important role in altering the ground
motions recorded at different locations, and because of this, the damage may vary
widely. The assessment of site effects is also needed in ground motion simulation
techniques, as the simulation is essential to know about the probable characteristics of
strong ground motion from future earthquakes, especially in a region having not many
recorded ground waveforms. The present study aims to estimate the site amplification
using recorded data of three earthquakes that occurred in the northeast and northwest
parts of the country. The used earthquakes include the 2015 Afghanistan earthquake
(Mw 7.5); the 2016 Hindukush earthquake (Mw 6.7) and the 2016 Manipur earthquake
of (Mw 6.7). The detail about the used earthquakes is given in Fig. 2 and Table 1. The
site amplification functions are important for determining the site-specific seismic
hazard assessment of a region. The locations of all stations are shown in Fig. 2.
Results of this investigation reveal the significance of site response studies.
The surface ground motion may be strongly amplified if the geological conditions
are dominated by the accumulation of sediments. The geology of all stations varies,
and so the amplitude and frequency content of the bedrock motion get modified
due to the effect of local soil conditions. The difference in predominant frequency
due to various earthquakes may be attributed to path effect, local site conditions are
considered independent of the distance between the source and the site, so that the
Fig. 2 Map showing the different parts of the India region and seismological network deployed in
the region. Three circles show major earthquakes in the region, and triangles represent the stations
predominant frequency is more in the third earthquake comparatively earthquake 1

and earthquake 2.
2 Data Processing and Methodology
India Meteorological Department (IMD) maintains a 17-station Real-Time Seismic

Monitoring Network (RTSMN). The ground motion data recorded at all 17 field
stations is transmitted in real time through VSAT communication systems to the
Central Receiving Station located at IMD, New Delhi, for processing. In the present,
the amplitude time histories and the frequency response curves were analysed using
Table 1 Geology of the recording stations

S. no. Date of earthquake Name of the earthquake Station Geology of the stations
1 26-10-2015 Afghanistan earthquake BHPL Deccan trap, alluvium,
sandstone
DDI Siwalik (boulder
conglomerate exposures),
sandstone, claystone and
mudstone
GOA The rocks of Dharwar
super group of
pre-Cambrian age
HYB Granites, granite gneisses
and alluvium
SHL Precambrian rocks of
Gneissic composition are
the oldest rocks
SMLA Tertiary and pre-tertiary
rocks
2 10-4-2016 Hindu Kush earthquake BHPL Deccan trap, alluvium,
sandstone
super group of
pre-Cambrian age
and alluvium
rocks
3 3-1-2016 Manipur earthquake BHPL Deccan trap, alluvium,
sandstone
super group of
pre-Cambrian age
and alluvium
ITANA Siwalik sediments and
unconsolidated
quaternary deposits
JORHT Precambrian gneisses,
schists and quartzites
KOHMA Alluvium and terrace
deposit, tertiary
sedimentary rocks
SHL Precambrian rocks of
Gneissic composition are
the oldest rocks
(continued)
Table 1 (continued)
S. no. Date of earthquake Name of the earthquake Station Geology of the stations
rocks
LEKHA Proterozoic gneissic
complex, Shillong group
of meso-paleo proterozoic
age, granite plutons of
neo-proterozoic-lower
paleozoic age, lower
gondwana sedimentary
rocks of
permo-carboniferous age
TADON Precambrian, tertiary
rocks
every station of all three earthquakes. The latest accelerographs are capable of
providing digital records of ground acceleration with frequency content from DC
to even up to 100 Hz with a sampling frequency of 200 sps or more. Few stations
installed in northeast region of India are also used in the present study.
For each of the case studies, the data ground motion data reported from each
station was further analysed using seismic analysis software (SEISAN) to obtain the
NS-component, EW-component and Z-component. ViewWave is the software used
for viewing strong motion records and also estimating the velocity and acceleration
response spectra. The H/V [9] spectral ratios were also derived using the Nakamura
technique to obtain the predominant frequency and related amplification. The Fourier
transform (FFT) of all the components is estimated, and then, mean of the horizontal
components is divided by the vertical components in order to calculate the H/V for
all the data considered for the present study.
The accelerograms for each station were derived from the ViewWave software. The
Fourier spectra and Pseudo response spectra have also been plotted for each station
of each case (Figs. 3, 4 and 5). The H/V [9] spectral ratios and related predominant
frequency were derived (Table 2) and plotted against the frequency for two respective
stations for each earthquake considered (Figs. 6, 7 and 8). For both Afghanistan
and Hindukush earthquake, the highest and lowest peak ground acceleration (PGA)
values were observed at SMLA and HYB. For the Manipur earthquake, the highest
value of the order of 83.9 cm/s/s was observed at SHL site. A couple of other stations
in northeast region also have very high PGA values. A sharp decrease in PGA values
is observed at sites away from the source. During Afghanistan earthquake, the peak
ground velocity (PGV) values did not show much range, as observed in the Manipur
earthquake. The highest PGV values observed did not coincide with the highest
PGA values. However, PGV is supposed to be better indicator of disaster scenario
as compared to PGA [8].
In general, nearby stations are found to experience more amplification compared
far-off stations. However, some far stations are also amplified due to ground
characteristics or local soil conditions, involving several factors such as:
• the composition of soil layers
• S-wave velocities
• soil densities (internal damping of the individual layers).
It has been observed that local soil effects are observed strongly on horizontal
components of ground motion but weakly on the vertical components. So, the H/V
Fig. 3 Fourier spectra and velocity response spectra at HYB (upper panels) and SMLA site from
Afghanistan earthquake (lower panels)
Fig. 4 Fourier spectra and velocity response spectra at HYB (upper panels) and SMLA (lower
panels) sites from Hindukush earthquake
ratio is considered to be a rough estimate of site amplification. The H/V ratio is plotted
in the frequency range 0.1 to 10 for some of the stations for Afghanistan Earthquake,
Hindukush Earthquake and Manipur Earthquake and is shown in Figs. 6, 7 and 8.
The peak PGA and PGV values at each station are depicted in Table 2.
It can be seen that site amplification for SMLA is highest at frequency range
10 Hz and for SML and KOHMA frequency range is 8–9 Hz. For BHPL, HYB,
JORHT, and LEKHA regions, the frequency range is 0.1–2.9 Hz and shows the
amplification ranging from 8 to 10. The shift to corresponding peak amplification to
lower frequency range indicates that the consistency of soil layer in BHPL, HYB,
JORHT and LEKHA region is soft as compared to BHPL, JORTH, etc.
The results obtained here can be used by structural engineers to plan their construc-
tions such that its natural frequency will not be close to the range of prominent
frequency of the site considered.
Fig. 5 Fourier spectra and velocity response spectra at HYB (upper panels) and SHL (lower panels)
sites from Manipur earthquake
4 Conclusions
In the present study, an attempt has been made to characterize the sites and to study the
seismic hazard analysis considering the amplification and predominant frequency.
The predominant frequency for soil sites is found to be less as compared to stiff
and rock sites. To testify to the exposed geology, the endeavour is also made to
classify sites in various classes based on predominant frequency. It is well known
that the soil layers at a given site influence the final surface ground motion. We
estimated amplification at various sites from three different earthquakes occurring in
different parts of the country. The results show significant changes in predominant
frequency and amplification factor from one place to another, which is caused due to
the variation in soil type or thickness of overlying soil cover. The higher amplification
Table 2 The observed PGA, PGV, amplification and predominant frequency at different stations
during three different earthquakes
S. no. Name of the Stations PGA (cm/s/s) PGV (cm/s) H/V ratio Predominant
earthquake frequency
1 Afghanistan BHPL 0.7 0.22 1.8 8.3
earthquake DDI 8.0 1.10 2.0 5.6
GOA 0.4 0.11 3.8 4.3
HYB 0.3 0.17 1.8 2.7
SHL 1.5 0.08 1.9 2.9
SMLA 8.5 1.04 2.6 7.8
2 Hindu Kush BHPL 0.6 0.05 2.3 0.1
earthquake GOA 0.3 0.01 3.8 4.4
HYB 0.1 0.03 1.9 5.7
SMLA 3.8 0.43 2.6 7.3
3 Manipur BHPL 0.2 0.04 2.2 0.1
earthquake GOA 0.1 0.02 3.6 4.5
HYB 0.1 0.02 1.8 5.3
ITANA 36.5 2.47 3.0 4.4
JORHT 10.9 1.56 1.9 2.0
KOHMA 54.1 5.01 2.5 8.1
SHL 7.4 0.35 2.1 9.9
SMLA 83.9 2.94 3.1 10.0
LEKHA 0.2 0.03 2.6 2.1
TADON 2.9 0.14 2.5 6.3
Fig. 6 H/V ratio at SMLA (left) and HYB (right) due to Afghanistan earthquake
Fig. 7 H/V ratio at HYB (left) and SMLA (right) sites due to Hindu Kush earthquake
Fig. 8 H/V ratio at HYB (left) and SHL (right) sites due to Manipur earthquake
at low predominant frequency is at some of the sites which may be treated as soil
sites, and the sites having less amplification at high frequency are regarded as stiff-soil
sites. The shift of corresponding peak amplification from higher to lower frequency
range indicates that the soil layer in BHPL, HYB, JORHT and LEKHA region is soft
as compared to the far stations. The study is conducted using only three earthquakes.
In future, the variations in site effects due to variation in distance will be attempted
using the larger dataset. This study is helpful for construction for building at particular
sites.
Acknowledgements Authors are grateful to the National Centre for Seismology (NCS) for
providing the necessary data to carry out this research work.
References
1. Bilham, R.: Earthquakes in India and the Himalaya: tectonics, geodesy and history. Ann.
Geophys. 47(2/3), 839–858 (2004)
2. BIS IS 1893 (2002) (Part 1). Indian Standard Criteria for Earthquake Resistant Design of
Structures, Part 1—General Provisions and Buildings. Bureau of Indian Standards, New Delhi
(2002)
3. Bora, D., Baruah, S.: Mapping the crustal thickness in Shillong–Mikir Hills Plateau and its
adjoining region of northeastern India using Moho reflected waves. J. Asian Earth. Sci. 48,
83–92 (2012). [Web of Science ®]
4. Chen, W., Molnar, P.: Focal depths of intracontinental and intraplate earthquakes and their
implications for the thermal and mechanical properties of the lithosphere. J. Geophys. Res. 88,
4183–4214 (1983)
5. Guha, S.K., Bhattacharya, U.: Studies on prediction of seismotectonics of northeastern India.
In: Presented at the Eighth World Conference on Earthquake Engineering, San Francisco (CA),
pp. 21–27 (1984)
6. Kayal, J.R.: Some Useful Definitions of Earthquake Seismology and Seismotectonic of
Northeast India (2008)
7. Khattri, K., Wyss, M.: Precursory variation of seismicity rate in Assam area, India. Geology 6,
685–688 (1987). [Web of Science ®]
8. Mittal, H., Yang, B.M., Tseng, T.L., Wu, Y.M.: Importance of real-time PGV in terms of lead-
time and shakemaps: results using 2018 ML 6.2 & 2019 ML 6.3 Hualien, Taiwan earthquakes.
J. Asian Earth Sci. 220, 104936 (2021)
9. Nakamura, Y.: A method for dynamic characteristics estimations of subsurface using
microtremors on the ground surface. QR RTRI 30, 25–33 (1989)
10. Studies on prediction of seismotectonics of northeastern India. Presented at the Eighth World
Conference on Earthquake Engineering, San Francisco (CA), pp. 21–27
ROSERS—A Deep Learning Framework
for Earthquake Early Warning and Its
Interpretation
Jawad Fayaz
Abstract Earthquake early warning (EEW) systems are swiftly evolving from stan-
dalone physics-inferred methods (requiring computationally expensive inversions) to
data-driven strategies for efficient earthquake hazard mitigation in real time. Besides
being speedy in prediction, data-driven approaches such as artificial neural networks
usually require minimal assumptions in the training and execution processes. This
study discusses and attempts to interpret the data-driven EEW framework: ROSERS
(Real-Time On-Site Estimation of Response Spectra) proposed by Fayaz and Galasso
(2022). ROSERS aims to utilize the early non-damaging p-waves and the recording-
site characteristics to predict the acceleration response spectrum (Sa (T )) of the antic-
ipated on-site ground motion waveform. The framework’s efficacy is analyzed using
an extensive database of ground motions, and it is observed that ROSERS leads to
exceptional prediction power when implemented in a real-time backdrop. To provide
a better interpretation of the framework, this study utilizes the concepts of explainable
artificial intelligence (i.e., Shapley additive explanation, SHAP) to obtain insights
into the decision-making process of the trained neural networks. Particularly, the
cause-effect relationship of the computed latent variables and Sa (T ) is explored.
The analyses showcase that the two latent variables of the framework complement
each other in capturing stiff short-period and flexible long-period Sa (T ) thereby
leading to excellent reconstruction power.
Keywords Variational autoencoders · Neural networks · Earthquake early

warning · Explainable artificial Intelligence · Model-agnostic Interpretability
J. Fayaz (B)
School of Computing, Engineering, and Digital Technologies, Teesside University,
Middlesbrough, UK
e-mail: j.fayaz@tees.ac.uk; j.fayaz@ucl.ac.uk
Department of Civil, Environmental, and Geomatic Engineering, University College London,
London, UK
https://doi.org/10.1007/978-981-99-1459-3_38
474 J. Fayaz
1 Introduction
The overall goal of earthquake early warning (EEW) frameworks is to alert commu-
nities during earthquake-induced ground shaking to prevent casualty and monetary
losses. Due to the constraints of available time during an earthquake event (i.e.,
time elapsed between the fault rupture and the end of ground shaking at the target
site), the efficiency of the EEW frameworks plays a vital role in making sure that
the warnings are rapid, accurate, and structurally useful. For the real-time setting,
EEW frameworks are developed to obtain characteristics of the fast-traveling p-waves
(usually possess low amplitude and do not lead to major damages) of the incoming
seismic waveforms. This information is then utilized to predict characteristics of the
incoming high-amplitude destructive transverse s-waves and appropriate alerts are
triggered [1].
Among the various ground shaking parameters of interest [2], the acceleration
response spectrum (Sa (T )) has been one of the most thoroughly used IMs that can
effectively integrate the features of the ground motion waveform (such as amplitude
and frequency content) with the dynamic characteristics of a structural system which
is represented as a single-degree-of-freedom system (SDOF) [3]. Over the years,
Sa (T ) has grown in its utility for structural and earthquake engineering purposes,
including hazard analysis [4, 5], ground motion simulation validation [2, 6], ground
motion selection and scaling [7], etc. Hence, an early and accurate estimation of
Sa (T ) of the arriving ground motion during an earthquake can assist stakeholders in
performing risk-informed decision-making [8, 9].
The study presented in this paper discusses the performance of the on-site
EEW framework: ROSERS (real-time on-site estimation of response spectra), on
Next Generation Attenuation (NGA) West 2 database [10] and attempts to inter-
pret the trained neural networks using the concepts of explainable artificial intelli-
gence (XAI) and game theory based Shapley additive explanations (SHAP) [11]. The
ROSERS framework is developed using the key components of deep neural network
(DNN) and variational autoencoder (VAE). The proposed framework utilizes a vector
of recording-site characteristics (denoted as SC) and an IM vector (representing
amplitude, energy, duration, and frequency content) of the initial three seconds of
the ground motion waveform (denoted as IM3s ) computed after detection of p-waves,
to predict a vector of PGA (also referred to Sa (T = 0)) and 95-period Sa (T ) spec-
trum of the anticipated complete on-site ground motion waveform in real time. The
VAE is utilized to conduct dimensionality reduction of the Sa (T ) spectrum into two
sufficient and efficient latent variables, which are then estimated in real-time using
the pre-trained DNN with SC and IM3s as the inputs [12]. The estimated latent vari-
ables are then used to project the complete Sa (T ) spectrum using the pre-trained VAE
decoder. Hence, ROSERS broadens the goals of EEW frameworks to be accurate,
timely, and both structurally and ground motion informed. The proposed framework
is analyzed on a comprehensive dataset and based on the findings; the proposed
framework can be highly beneficial to advancing EEW research.
ROSERS—A Deep Learning Framework for Earthquake Early Warning … 475
Due to the “black-box” nature of the neural networks, understanding the decision-
making process of the framework is imperative. This study provides insights into the
framework by interpreting the cause-effect relationships using the concepts of XAI.
In particular, SHAP is used to conduct a post hoc model interpretation of the two
neural networks, including the DNN for estimation of latent variables using SC
and IM3s , and the VAE decoder that projects the latent variables into the complete
96-point Sa (T ) spectrum (denoted as Sa ). The interpretation procedure basically
quantifies each feature’s effect on the neural networks’ output, thereby attempting
to provide necessary transparency to understand the framework. In addition, this
process allows the users and stakeholders to obtain a physical interpretation of the
two latent variables utilized by the framework.
2 ROSERS Framework
Figure 1 illustrates the proposed framework. For a given location, a digital recording
station constantly monitors the ground shaking. The site characteristics, i.e., SC of
the recording station are known apriori. As the p-waves of the incoming ground
motion are picked by the sensor [13, 14], the ROSERS framework waits to receive
three seconds of the waveform. The ground motion shaking at a site is typically
known to last for 1 to 2 min [15], and hence, using early three seconds of the arriving
ground motion still allows ample time for on-site early warning and rapid response.
Using the recorded three seconds of the waveform, a vector of seven IMs (i.e., IM3s )
is computed. The computed IM3s represent the amplitude, energy, frequency, and
significant duration of the initial three seconds of the ground motion waveform after
detecting the p-waves. IM3s and SC are used as the inputs to the pre-trained feed-
forward DNN which is trained to estimate the two statistically derived surrogate
mean latent variables (i.e., μz1 and μz2 ). The mean latent variables in conjunction
with the pre-trained VAE decoder are projected into Sa of the expected complete
waveform at the end of the shaking. The estimated Sa can then be utilized by the
stakeholders to alert a community informatively.
The mathematical basis of the framework relies on four models/algorithms,
including (1) automatic p-phase arrival-time picker (which detects the onset of p-
waves in real-time), (2) VAE (which provides regularized surrogate parameters;
whose encoder projects Sa into two sufficient and efficient latent variable space
and decoder reconstructs Sa using the latent variables), (3) DNN (which utilizes
IM3s and SC to compute the two mean latent variables μz1 and μz2 ), and 4) mixed-
effects hierarchical regression (which provides the residuals a hierarchical structure
and implicitly considers spatial correlation by dividing the residuals of the trained
VAE + DNN into between-event and within-event variabilities). Further details of
the components of the proposed framework can be obtained from Fayaz and Galasso
[12].
The IM3s include seven IMs: Arias intensity (Ia ), significant duration (D5−95 ),
mean period (Tm ), peak ground acceleration (PGA), peak ground velocity (PGV),
476 J. Fayaz
Fig. 1 Illustration of the ROSERS framework. The implementation of the ROSERS framework,
on average, takes ~ 0.75 s on a personal computer
peak ground displacement (PGD), and cumulative absolute velocity (CAV) described
in Eqs. 1–7 where a(t) represents the acceleration time history of the ground motion,
T i represents the time instance, and Ci is the Fourier amplitude spectrum at linearly
spaced frequencies, f i , ranging between 0.25 ≤ f i ≤ 20 Hz.

π
Ia = a(t)2 dt (1)
2g
D5−95 = T i (Ia@95% ) − T i (Ia@5% ) (2)

Ci2 1 f i
Tm = 2 (3)
Ci
PGA = max(|a(t)|) (4)

PGV = max a(t)dt
(5)

PGD = max a(t)dtdt (6)

CAV = | a(t)dt|dt (7)
3 Implementation on NGA-West2 Database
The framework is developed and trained using a subset of recorded ground motions
available in the comprehensive state-of-the-art database of the Next-Generation
Attenuation West version 2 (NGA-West2) project [10]. An initial database of
13,916 ground motion components recorded from 277 mainshock seismic events
was utilized. However, due to the asymmetry in the number of records available for
low-magnitude and high-magnitude events, the database was statistically altered to
obtain a conforming number of events from both ranges of magnitudes. This was
done by undersampling the events with a moment magnitude (M) less than 5.5 until
the distribution and number of the events with M ≤ 5.5 were similar to M > 5.5.
The sampling is carefully conducted by maintaining the distributions of the closest
rupture distance (Rrup ) and the site’s average shear-wave velocity of the topmost 30 m
of the soil (Vs30 ). This results in 6392 undersampled ground motions with 3 ≤ M ≤
8 and 0.1 km ≤ Rrup ≤ 90 km. The process details can be obtained from Fayaz
and Galasso [12]. Finally, to obtain the p-wave arrival times of the ground motion
components, the automated p-phase picker (PPHASE PICKER ) algorithm proposed by
Kalkan [13] is utilized.
3.1 Computation of Latent Variables Using VAE
The undersampled 6392 ground motion components are used to compute the Sa for 96
periods, and the Sa is used to train the VAE [16]. The VAE uses Sa as inputs and trains
a neural network-based encoder to reduce Sa into two probabilistic latent variable
representations while simultaneously training a neural network-based decoder that
reverses the process and projects the latent variables back to Sa . The VAE provides
a probabilistic approach to describing the Sa in their latent variable space, which
is coerced to possess smooth and continuous latent variable representations. In this
study, latent variable space of two dimensions is observed to be as a good trade-off
between accuracy and explainability. The two latent variables are assumed to be
independent and normally distributed (denoted as z 1 and z 2 with means μz1 and μz2
and variances σz21 and σz22 , , respectively) in the sampling layer. The VAE is trained
with k-fold cross-validation using randomly selected 80% of the event dataset (10%
of which is used as the validation set). The remaining 20% is used for testing, as
showcased in [12]. The M and Rr up details of the train and test datasets are provided
in Fig. 2a and b, respectively.
The μz1 and μz2 are presented in Fig. 3a and b, where the marker colors char-
acterize the M of the earthquake (Fig. 3a) and Rr up of the station site (Fig. 3b). In
general, it is observed that an increase in M leads to decrease in μz1 and μz2 and
increase in Rrup leads to increase in μz1 and μz2 . Hence, it is observed that μz1 and
μz2 have an inverted attenuation relation with the M and direct proportionality with
the Rrup . Furthermore, Table 1 presents the correlation coefficients of μz1 and μz2
478 J. Fayaz
Fig. 2 M and Rrup details of: a train, and b test, datasets
with various parameters of IM3s and SC. It is observed from Table 1 that both SC
and IM3s vectors are well correlated with μ Z 1 and μ Z 2 highlighting their signifi-
cance in real-time prediction of μ Z 1 and μ Z 2 . In nutshell, μz1 and μz2 are in positive
correlation with each other and are negatively correlated with the IM3s and SC.
The results of the training VAE are presented in Fig. 4a and b. Figure 4a presents
the coefficient of determination R2 for both train and test sets for the Sa (T ) at 96
periods. It is observed that for both train and test sets, the R2 for all periods is above
Fig. 3 Exploration of μz 1 and μz 2 against a M, b Rrup
Table 1 Correlations of latent variables with the selected SC and IM3s (NGA dataset)
Vs30 Z 2.5 Ia3s CAV3s PGA3s PGV3s PGD3s Tm3s D5−953s μZ1
μZ1 −0.27 −0.21 −0.83 −0.91 −0.82 −0.89 −0.91 −0.57 0.19 1
μZ2 −0.19 −0.07 −0.93 −0.91 −0.92 −0.92 −0.90 −0.36 0.27 0.88
Fig. 4 NGA-West2 selected ground motions: a R2 of reconstructed Sa (T ) at various periods and

b true versus reconstructed Sa (T )
0.98, thereby demonstrating excellent reconstruction power of the developed VAE

with minimal bias and variance. Furthermore, Fig. 4b shows the true vs. reconstructed
Sa (T ) for six periods of: 0.25 s, 0.75 s, 1.5 s, 3 s, 5 s, and 0 s (PGA). For all the
cases, it is observed that the true vs. reconstructed Sa (T ) closely follows the purity
(1:1) line for all ranges of values. Based on the observations made from Fig. 4a and
4b, it is clear that the trained VAE is sufficient and efficient to adequately reconstruct
μz1 and μz2 .
3.2 Estimation of Latent Variables Using DNNs
Based on the sufficiency and efficiency of the latent variables and VAE observed in the
previous section, it is vital to accurately estimate the latent variables for constructing
the Sa of the expected ground motion in a real-time setting. Hence, a DNN with
nine input nodes and two output nodes is trained to estimate the μz1 and μz2 using
the SC and IM3s vectors. The DNN is trained with cross-validation using randomly
selected 80% of the dataset, and the remaining 20% is used for testing. The mean
predictions of the trained DNN led to R2 of 0.91 and 0.93 for μz1 and μz2 , respectively,
for the test dataset. Apart from a high prediction power, the DNN provides a joint
estimation of the latent variables, thereby maintaining their internal correlation. The
predictions of μz1 and μz2 from the DNN inherit a correlation coefficient of 0.91
compared to the true correlation coefficient of 0.88 (as observed in Table 1). This
means that the trained DNN is highly successful in ensuring that the estimations are
implicitly associated, i.e., properly reflecting their internal correlation. Finally, using
mixed-effects regression [17], the residuals of Sa from the DNN + VAE predictions
are divided into inter-event and within-event variabilities to maintain the hierarchical
nature of the ground motions arising from multiple recordings from the same event
480 J. Fayaz
and recordings from different events. Hence, the corresponding inter-event T2 and
within-event 2 variance matrices are computed for the residuals of the 96 periods.
4 Interpretation of Neural Networks of ROSERS
Due to the versatility of the deep learning models, they have been widely used in
engineering applications. However, due to the “black-box” nature of these models,
there is a general reluctance in the research community to propose and utilize such
models. Hence, it is critical to provide sufficient analytics for model interpretability
and its response in terms of predictions based on variability in the input features.
With the onset of XAI, various algorithms have been developed that provide different
methods that allow interpretability of these “black-box” models and step toward
“gray-box” and even “white-box” nature (such as linear regression and decision
trees). This study uses SHAP to analyze and interpret the nature of the developed
framework for the NGA-West2 database [18]. SHAP is a post hoc model-agnostic
procedure that provides insights to explain individual predictions of the model based
on the game’s theoretically optimal Shapley values. Shapley values are a widely
used approach from cooperative game theory with desirable characteristics [11].
The Shapley value is the average marginal contribution of a feature value across
all possible coalitions where players of a coalition are acted by the feature values
of a data instance. Shapley values are the only solution that satisfies the properties
of efficiency, symmetry, dummy, and additivity, which form the basis of the most
reliable explanation methods [18].
SHAP belongs to the class of models called “additive feature attribution methods,”
where the explanation is expressed as a linear function of features, as described by
Eq. 8. In this equation, g(z) represents a local surrogate model of the original model
f (x) (in this case, f (x) is the trained DNN or the VAE decoder), K is the number
of input features of the model f (x) (i.e., Sa (T1 ) and S F), θ0 denotes the bias term,
which represents the base value of the predictions made by the model f (x), θi is the
contribution of the ith feature toward the final output (i.e., SHAP value), and z i is a
binary variable that takes a value of 1 for the feature corresponding to θi contribution
and 0 otherwise. θi evaluates the difference to the final predictions made by the
model f (x) by including the ith feature for all combinations of features other than
i. This is expressed in Eq. 9, where S represents a subset of features among all N
features except the ith feature (denoted as S ⊆ N \{i}), [ f x (S ∪ {i}) − f x (S)] is the
difference in the outputs made by the ith feature, and |S|!(M−|S|−1)!
M!
is the weighing
factor counting the number of permutations of the subset S. f x (S) in the different
part of the equation represents the expected output given the subset of features S,
which is similar to the marginal average on all other features other than the subset
S. Hence, in a nutshell, SHAP values explain the contribution of the features to the
respective outputs in a quantitative manner, thereby allowing interpretation. These
are analogically similar to the coefficients of a regression model, which provide the
impact of the corresponding feature on the target variable. Due to the computational
complexity of computing Eq. 10, SHAP values are approximated using various types
of explainers such as kernel-explainer, tree-explainer, and deep-explainer [19]. In
this study, kernel-explainer uses a special weighted linear regression to compute
the importance of each feature. The computed importance values are Shapley values
from game theory and coefficients from a local linear regression [19]. SHAP analysis
is applied to interpret the cause-effect relationship between the predictors and targets
of the trained VAE decoder and DNN in the following sections.
M
g(z) = θ0 + θi z i (8)
i=1
|S|!(K − |S| − 1)!

θi = [ f x (S ∪ {i}) − f x (S)] (9)
S⊆N \{i}
K!
f x (S) = E[ f (x)|x S ] (10)
4.1 Interpretation of the VAE Decoder
The decoder of the trained VAE is analyzed using SHAP analysis by using μz1 and μz2
of the 6392 ground motions as the inputs (features) and the corresponding predictions
of Sa vector as the outputs (targets). For each input–output combination, respective
SHAP values are computed. Hence, a total of 6392 SHAP values are computed
for each combination of the 96 outputs (Sa ) and two inputs (μz1 and μz2 ). The
SHAP values are presented for PGA, Sa (T = 0.5s), Sa (T = 1s), and Sa (T = 2.5s)
in Fig. 5, where the color of the data points represents the magnitude of the feature
values. The low refers to values close to -0.4, and the high refers to values close
to +0.4 for μz1 and μz2 (based on Fig. 3). It can be observed that, in general, with
an increase in the value of both features, their corresponding SHAP values tend to
move from positive values to negative values. This means as the values of the latent
variables increase, they tend to lower (negatively contribute) the predicted Sa (T )
values obtained from the decoder. Also, the SHAP values tend to be symmetric on
both sides of zero for both latent variables and all four targets. This means that for both
extreme values of the latent variables (i.e., −0.4 and +0.4), their impact on the four
target variables is similar (since the absolute SHAP value is similar). Furthermore, it
is observed that μz1 leads to a similar range of SHAP values as μz2 for PGA indicating
a similar level of importance of both variables to estimate PGA. However, comparing
the SHAP values across the four target variables, it is noticed that the contribution
of the μz1 increases while the contribution of μz2 diminishes with an increase in the
period of the SDOF. Thus, the contribution of μz1 to estimate the Sa is observed to
be higher than the μz1 for longer periods SDOFs. This diminishing effect of μz2 s
SHAP values is compensated by increasing variance in μz1 s SHAP values.
482 J. Fayaz
Fig. 5 SHAP of μz 1 and μz 2 for: a PGA, b Sa (T = 0.5s), c Sa (T = 1s), and d Sa (T = 2.5s)
To present the details more concisely, Fig. 6 illustrates the relative feature impor-
tance of μz1 and μz2 on the VAE decoder predictions, in terms of their mean absolute
SHAP values (| SHAP|). This is done by computing the mean | SHAP | values for the
6392 samples for both μz1 and μz2 and then dividing them by the sum of the mean |
SHAP | for each target Sa (T ). Since the SHAP values represent the contribution of
the features to the model output, computing their relative sum for the two features
in an absolute sense signifies the importance of the respective feature in predicting
Sa (T ). It can be observed from the figure that the relative importance of μz1 and μz2
is similar for Sa (T ) with shorter periods (close to PGA) and the importance of μz1
increases as the period increases with a plateau effect around Sa (T = 1). Thus, in
general, it is observed that μz2 inherits the capabilities to capture the acceleration
effects on stiffer SDOFs, and with an increase in flexibility of SDOF, μz2 is not as
effective in the prediction of Sa (T ) and μz1 leads to higher contribution. Hence, this
process of interpreting the decoder provides insights into the cause-effect relation-
ship between μz1 and μz2 and Sa , and also provides details about the latent variables
in capturing the behavior of SDOF with different periods.
Fig. 6 Relative mean

absolute SHAP values of μz 1
and μz 2
4.2 Interpretation of the DNN
Similar to the interpretation process of the VAE decoder, the trained DNN is analyzed
using SHAP analysis by using the SC and IM3s features of the 6392 ground motions
as the inputs and the corresponding mean latent variables μz1 and μz2 as the outputs.
For each case, the respective SHAP values are computed for the trained DNN. Hence,
a total of 6392 SHAP values are calculated for each combination of the nine inputs
(SC and IM3s ) and two outputs (μz1 and μz2 ). Figure 7 presents the SHAP values
for the nine features of SC, and IM3s corresponding to the two target mean latent
variables μz1 and μz2 in descending order of contribution. The color of the data points
represents the magnitude of the corresponding feature values. It can be observed from
the sub-figures that C AV , P GV , and Ia lead to the highest SHAP values for μz1
while for μz2 P GV is replaced by P G A as one of the top three contributors, and the
SHAP values of P GV get reduced significantly. This bolsters the observations made
in the previous section about μz2 containing information related to PGA and stiffer
period Sa (T ) and signifies the importance of μz2 in capturing the PGA and short-
period Sa (T ) of the ground motion waveform. On the other hand, μz1 is observed
to be connected to energy-based IMs, thereby leading to a higher contribution to
capturing the long-period Sa (T ). Furthermore, it is observed from Fig. 11 that the
DNN is less affected by the SC for the prediction of μz2 while the SC contributes
to a higher degree for the prediction of μz1 . Also, the DNN is observed to be more
impacted by the significant duration of the 3-s ground motion for the prediction of
μz2 as compared to μz1 . Unlike the previous section, the SHAP values of the features
are not observed to be symmetric around zero, thereby indicating that the value of
different features leads to different contributions to the DNN predictions. In general,
it is noticed that the lower values of the features lead to a positive contribution
to the DNN predictions. In comparison, the higher values of the features lead to
contribution in a negative sense. It should be noted here negative sense does not
mean lower contribution but rather specifies that the corresponding feature value
lowers the prediction from the average prediction value of the DNN.
484 J. Fayaz
Fig. 7 SHAP values of SC and IM3s for: a μz 1 and b μz 2
To further present the SHAP details more concisely, Fig. 8 illustrates the relative
feature importance of SC and IM3s on the DNN predictions of μz1 and μz2 , in
terms of their mean absolute SHAP values (| SHAP|). This is done by computing
the mean | SHAP | values for the 6392 samples for all features in SC and IM3s
and then dividing them by the sum of the two mean | SHAP | for the two target
latent variables. Since the SHAP values represent the contribution of the features to
the model output, computing their relative sum for the features in an absolute sense
signifies the importance of the respective feature in predicting μz1 and μz2 . It can be
observed from the figure that the DNN is affected similarly by CAV to predict μz1
and μz2 . Among the top contributors to the DNN, Ia is observed to have the highest
mean impact to predict μz2 and PGV is observed to have the highest mean impact
to predict μz1 . Also, the prediction of μz1 is relatively affected higher by SC and
prediction of μz2 is impacted more by D5−95 and Tm . In a nutshell, this process of
interpreting the DNN helps in understanding the cause-effect relationship between
the SC and IM3s and predictions of μz1 and μz2 and thereby provides insights about
the DNN and how the predictions are affected by the changes in the features (similar
to classical regression analysis).
5 Conclusions
This study discusses and interprets a deep learning-based on-site earthquake early
warning (EEW) framework called ROSERS (real-time on-site estimation of response
spectra). The framework utilizes a pre-trained deep neural network (DNN) which
uses the site characteristics and initial three seconds of the on-site ground motion
after p-wave detection to estimate two latent variables that are statistical surrogates
Fig. 8 Relative mean

absolute SHAP values of SC
and IM3s
of the Sa (T ) spectrum. The estimated two latent variables are then used in a trained
decoder of variational autoencoder (VAE) to predict the peak ground acceleration
(PGA) and 95-period spectral acceleration (Sa (T )) response spectrum of the antic-
ipated complete on-site ground motion. While many other research studies have
tried to use machine learning and deep learning frameworks for EEW purposes, they
mainly concentrate on predicting ground acceleration, which can be significantly
different from the acceleration caused in the structures (due to dynamic amplifica-
tion). The prediction of the complete Sa (T ) spectrum (while inherently maintaining
the cross-correlations) with only three seconds of the initial ground motion waveform
in a highly accurate manner coupled with an average prediction time of less than one
second makes this framework unique. However, due to the black-box nature of the
neural networks, methods like these often lack explainability and interpretability.
Hence to overcome this, a detailed explanation of the trained VAE and DNN is
provided using the XAI concept of SHAP analysis. This study is an attempt to illu-
minate the black-box nature of the neural networks used in the ROSERS framework
and try to understand the physical essense of the surrogate latent variables. Results
indicate that μz2 captures the acceleration effects on stiff short-period SDOFs, while
μz1 leads to higher contribution for estimating acceleration of flexible long-period
SDOFs.
References
1. Kramer, S.L.: Geotechnical Earthquake Engineering. Prentice Hall (1996)

2. Fayaz, J., Azar, S., Dabaghi, M., Zareian, F.: Methodology for validation of simulated ground
motions for seismic response assessment: application to cybershake source-based ground
motions. Bull. Seismol. Soc. Am. 111(1), 226–241 (2021). https://doi.org/10.1785/012020
0240
3. Bazzurro, P., Cornell, C.A., Shome, N., Carballo, J.E.: Three proposals for characterizing
MDOF nonlinear seismic response. J. Struct. Eng. 124(11), 1281–1289 (1998). https://doi.org/
10.1061/(asce)0733-9445(1998)124:11(1281)
486 J. Fayaz
4. Campbell, K.W., Bozorgnia, Y.: NGA-West2 ground motion model for the average horizontal
components of PGA, PGV, and 5% damped linear acceleration response spectra. Earthq. Spectra
30(3), 1087–1114 (2014). https://doi.org/10.1193/062913EQS175M
5. Fayaz, J., Xiang, Y., Zareian, F.: Generalized ground motion prediction model using hybrid
recurrent neural network. Earthq. Eng. Struct. Dynam. 50(6), 1539–1561 (2021). https://doi.
org/10.1002/eqe.3410
6. Fayaz, J., Rezaeian, S., Zareian, F.: Evaluation of simulated ground motions using probabilistic
seismic demand analysis: CyberShake (ver. 15.12) simulations for ordinary standard bridges.
Soil Dyn. Earthq. Eng. 141, 106533 (2021). https://doi.org/10.1016/j.soildyn.2020.106533
7. Vamvatsikos, D., Allin Cornell, C.: Incremental dynamic analysis. Earthq. Eng. Struct. Dyn.
31(3), 491–514 (2002). https://doi.org/10.1002/eqe.141
8. Cremen, G., Galasso, C.: Earthquake early warning: recent advances and perspectives. Earth
Sci. Rev. 205, 103184 (2020). https://doi.org/10.1016/j.earscirev.2020.103184
9. Cremen, G., Galasso, C.: A decision-making methodology for risk-informed earthquake early
warning. Comp. Aid. Civ. Infra. Eng. 36(6), 747–761 (2021). https://doi.org/10.1111/mice.
12670
10. Ancheta, T.D., et al.: NGA-West2 database. Earthq. Spectra 30(3), 989–1005 (2014). https://
doi.org/10.1193/070913EQS197M
11. Roth, A.E.: The Shapley Value: Essays in Honor of Lloyd S. Cambridge University Press,
Shapley (1988)
12. Fayaz, J., Galasso, C.: A deep neural network framework for real-time on-site estimation of
acceleration response spectra of seismic ground motions. Comp. Aid. Civ. Infras. Eng. (2022).
https://doi.org/10.1111/mice.12830
13. Kalkan, E.: An automatic P-phase arrival-time picker. Bull. Seismol. Soc. Am. 106(3), 971–986
(2016). https://doi.org/10.1785/0120150111
14. Akazawa, T.: A technique for automatic detection of onset time of P- and S-phases in strong
motion records. In: 13th World Conference on Earthquake Engineering, vol. 786, p. 786 (2004)
15. Fayaz, J., Medalla, M., Zareian, F.: Sensitivity of the response of Box-Girder Seat-type bridges
to the duration of ground motions arising from crustal and subduction earthquakes. Eng. Struct.
219, 110845 (2020). https://doi.org/10.1016/j.engstruct.2020.110845
16. Kingma, D.P., Welling, M.: An introduction to variational autoencoders. Found. Trends Mach.
Learn. 12(4), 307–392 (2019). https://doi.org/10.1561/2200000056
17. Demidenko, E.: Mixed Models. John Wiley & Sons, Inc., Hoboken, NJ, USA (2004). https://
doi.org/10.1002/0471728438
18. Lundberg, S.M., Lee, S.-I.: A unified approach to interpreting model predictions. Adv. Neur.
Inf. Process. Syst. 30 (2017)
19. Molnar, C.: Interpretable machine learning. Lulu.com (2020)
20. Boore, D.M., Bommer, J.J.: Processing of strong-motion accelerograms: needs, options and
consequences. Soil Dyn. Earthq. Eng. 25(2), 93–115 (2005). https://doi.org/10.1016/j.soildyn.
2004.10.007
21. Caruso, A., Colombelli, S., Elia, L., Picozzi, M., Zollo, A.: An on-site alert level early warning
system for Italy. J. Geophys. Res. Solid Earth. 122(3), 2106–2118 (2017). https://doi.org/10.
1002/2016JB013403
22. Hsu, T.-Y., Wu, R.-T., Chang, K.-C.: Two novel approaches to reduce false alarm due to non-
earthquake events for on-site earthquake early warning system. Comp. Aid. Civ. Infra. Eng.
31(7), 535–549 (2016). https://doi.org/10.1111/mice.12191
23. Borodinov, N., Neumayer, S., Kalinin, S.V., Ovchinnikova, O.S., Vasudevan, R.K., Jesse, S.:
Deep neural networks for understanding noisy data applied to physical property extraction in
scanning probe microscopy. NPJ Comput. Mater. 5(1), 1–8 (2019). https://doi.org/10.1038/s41
524-019-0148-5
24. Jumper, J., et al.: Highly accurate protein structure prediction with AlphaFold. Nature
596(7873), 583–589 (2021). https://doi.org/10.1038/s41586-021-03819-2
25. Zarrin, M., Abyani, M., Asgarian, B.: A statistical study on lognormal central tendency esti-
mation in probabilistic seismic assessments. Struct. Infra. Eng. 16(5), (2020). https://doi.org/
10.1080/15732479.2019.1668813
Early Warning System: An Efficient
Earthquake Disaster Mitigation Tool
P. Kumar, Govind Rathore, Kamal, M. L. Sharma, R. S. Jakka, Pratibha,

and A. Kumar
Abstract The earthquake risk mitigation goals are achieved by applying a number
of approaches to reduce the vulnerability of the built environment. The most widely
known long-term risk mitigation approaches are earthquake-resistant design and
construction, reviewing building use regulations and codes, and retrofitting of the
existing vulnerable infrastructure. Construction experts, earthquake, and civil engi-
neers must emphasize on structural and non-structural elements to avoid potential
damages and human loss. Accurate predictions of earthquakes would effectively
reduce the damage caused by the earthquakes. Unfortunately, seismic stress buildup
inside the earth’s crust and its release, starting from the nucleation to the end of
the rupture, is a complex phenomenon. Therefore, making accurate predictions of
earthquakes is quite tricky. Revolution in digital seismology, telecommunication,
and high-performance computing has provided strong support for risk mitigation
measures in the last four decades. The recent advancement in real-time processing
and transmission of seismic data has enabled the evolution of a real-time risk miti-
gation system like an earthquake early warning system (EEWS). The EEWS is a
real-time system for earthquake risk reduction.
Keywords Earthquake · Early warning system · Lead-time · Risk mitigation
P. Kumar (B) · G. Rathore

Centre of Excellence in Disaster Mitigation & Management, IIT Roorkee, Roorkee, India
e-mail: pkumar@dm.iitr.ac.in
Kamal
Department of Earth Sciences, IIT Roorkee, Roorkee, India
M. L. Sharma · R. S. Jakka · A. Kumar
Pratibha
Department of Mathematics, IIT Roorkee, Roorkee, India
e-mail: pratibha@ma.iitr.ac.in
https://doi.org/10.1007/978-981-99-1459-3_39
488 P. Kumar et al.
1 Introduction
Natural disasters have always been frightening and devastating to humans due to their
unpredictable nature. Earthquake is one of them, and it has always been very disas-
trous to humanity. Tectonic and volcanic activities cause earthquakes naturally, while
some human activities like an explosion, mine collapse, and reservoirs induce them
[1]. Tectonic earthquakes are the most common and can be explained by plate tectonic
theory [2]. The world seismicity map clearly shows that the seismicity distribution is
not random but follows a trend along the major plate boundary zones. Earthquakes
occur along tectonic plate boundaries and faults in the Earth’s crust, along with frac-
tures where friction has built up over time [3]. The whole fault does not move at once.
It starts at an epicenter, and the rupture moves down the fault. This release of energy
moves in two parts. Primary waves (p-waves) emit out first and fastest. They are
followed by slower secondary waves, or s-waves, which cause the ground to ripple
up and down—the shaking people experience during a quake. Earthquakes do not kill
people; buildings do. Therefore, earthquake-proof buildings need to be constructed
to mitigate earthquake hazards. Buildings made of local materials like bamboo or
timber are highly resistant to earthquake jolts. To overcome the effects of seismic
hazards in the buildings, seismic hazard study of the sites should be carried out and
the building models should be tested on the shake table to check their resilience
by applying jerks of different earthquake scenarios [4, 5]. Earthquakes cannot be
predicted as of now [6] but their effects can be mitigated through proper planning
and preparation. The municipality should be proactive in the verification of building
plans and should examine the construction compliance with building guidelines that
have been followed. Old buildings need to be retrofitted so that they can bear the
shocks during earthquakes. Public awareness is essential and crucial; people must
be aware of the consequences of earthquakes. Earthquake mitigation measures are
typically intended to reduce both casualties and damage from future earthquakes.
Special attention to early warning systems was given in the Hyogo framework
for action (HFA) 2005–2015: Building the resilience of nations and communities to
disasters. Risk assessment and early warning systems are essential investments that
protect and save lives, property and livelihoods. They contribute to the sustainability
of development and are far more cost-effective in strengthening coping mechanisms
than primary reliance on post-disaster response and recovery. Early warning systems
are an adaptive measure used to help communities to prepare for hazardous events by
taking advantage of integrated communication systems. Similarly, the Sendai Frame-
work for Disaster Risk Reduction 2015–2030 outlines seven clear targets and four
priorities for action to prevent new and reduce existing disaster risks. The seventh
target “G” clearly advocates for substantially increasing availability of and access
to multi-hazard early warning systems and disaster risk information and assess-
ments by 2030 [7]. It aims to reduce disaster risk and losses in lives and livelihoods
substantially.
The existing EEWS take advantage of the speed of the primary wave; they detect
them and, on the basis of applied approaches in the EEWS dissemination setup, issue
Early Warning System: An Efficient Earthquake Disaster Mitigation Tool 489
the warning to the public. An EEW system alerts the residents before the arrival of
the impending strong ground motion as shaking hits their dwellings and causes
damage. The lead time for EEW applications depends on the epicentral distance.
It varies from a few seconds to tens of seconds. Despite having such a short lead
time, a risk mitigation strategy and certain precautionary measures can be followed
to save the loss of lives and property. The lead time of few tens of seconds can be
enough for pre-programmed emergency measures for critical infrastructures. Several
approaches like warning sirens, TV and radio broadcasts, cell phone applications, or
dedicated alerting devices, have evolved to receive earthquake warnings. These alerts
can also be integrated into the critical facilities systems to perform specific actions
automatically upon receiving the alerts. The EEWS is one among the many effective
risk mitigation measures to prevent new and reduce existing disaster risks.
2 EEWS Set up Around the Globe
The term “earthquake early warning” (EEW) is used to describe a real-time earth-
quake information system that has the capability and potential to provide a warning
before significant ground shaking [8]. Warning time ranges from few seconds to a
little more than a minute and is primarily a function of the distance from the user to
the earthquake’s epicenter. An early warning system is mainly required to issue an
alarm to have a time margin for evacuating people, shutting down critical facilities,
and not determining exact earthquake parameters.
The first EEW system was developed for the Tohoku Shinkansen railway system
in Japan in 1982 using the front detection technique [9]. A similar approach was
also applied in Mexico’s seismic alert system in 1991 [10]. The first three seconds
of P-wave motion after P-wave detection was used in the improved front detection
techniques called UrEDAS and started operating for the Tokaido Shinkansen line in
1992 [9]. After the success of EEW system in Mexico and Japan, other countries
also started to work on the development of EEW systems for their regions, and
enormous improvements have been seen in this field in the last three-four decades.
The EEW systems are fully functional in a few countries like Japan [11, 12], Taiwan
[13–15], Mexico [16–18], and South Korea [19], to issue warnings nationwide, while
some countries issue warnings region-wise, like Anatolia-Turkey [20, 21], Southwest
Iberia [22–24], Southern Italy [25, 26], Vrancea-Romania [27–29], China [30, 31],
Chili [32], Costa Rica [33, 34], Switzerland [35], Nicaragua [36], Israel [37, 38],
and United States of America [39, 40]. While the development of the EEW system
is crucial for seismic risk reduction, increasing capacity to provide more lead time
is equally essential [41].
In India, Uttarakhand State Earthquake Early Warning System (UEEWS) has been
developed. Under this project, 167 sensors have been installed in the seismogenic
areas of Uttarakhand [42–44]. This project is being operated and maintained by
the EEWS laboratory, Centre of Excellence in Disaster Mitigation & Management,
Indian Institute of Technology Roorkee. The server has been set up in this laboratory
490 P. Kumar et al.
to fetch the data from sensors to the server using the dedicated private network. The
warning of the earthquake is issued on the mobile app and installed sirens in public
places and all thirteen district emergency operation centers of Uttarakhand.
3 Scenario Earthquake Warnings, Mock Drills,

and Awareness
As per the 2011 census, the total population of Uttarakhand was 10,096 thousands.
The projected population for the years 2021, 2025, 2030, and 2036 is expected
as 11,399 thousands, 11,993 thousands, 12,524 thousands, and 12,974 thousands,
respectively [45]. The trend shows that the population is increasing and it is our
combined responsibility to provide earthquake awareness in the state. Therefore,
mock drills of earthquakes are conducted in Uttarakhand at 12:00 PM on the first
day of every month. In this continuous process, a mock drill was conducted at 12:00
PM on 1 June, 2022 and ringed the installed sirens in Dehradun, Haldwani and District
Emergency Operation Centers. Alert messages were also received on the Uttarak-
hand Bhookamp Alert mobile app installed by the users. The received feedback from
the users was analyzed, and the system was updated accordingly. The pamphlets,
banners, posters (Fig. 1), stickers (Fig. 2), and booklets have been prepared to sensi-
tize the public about earthquakes. These ways and means are pasted and distributed
at various places by the technical teams who visit Uttarakhand during the installation
and maintenance of the sensors and sirens. The mock drills are very useful to test
the preparedness and planned response activities followed by disaster management
authority and all the line departments involved in the aftermath operations.
4 Benefit-Cost Study of Earthquake Mitigation Approaches
In India, large earthquakes are considered to be low probability but high-consequence

events. One measure of the economic impact of earthquake damage is the ratio of
total loss to insured loss (L/I). It is very high in developing nations due to the lack
of home insurance practices. In a broad sense, the economic impact of earthquakes
can be determined concerning the annual GDP of the country. Most of the urban
areas have large informal settlements and slums. The poor practices of building stan-
dards, land use planning, overcrowding, and location in significantly hazard-prone
areas make them vulnerable to seismic risks. Earthquake mitigation measures can
be applied to existing infrastructure to reduce casualties and damage in future earth-
quakes. Standard earthquake mitigation measures include improving the capacity of
a building to resist seismic forces. The retrofitting of a building improves the building
elements that hold up a building and resist lateral forces from earthquakes, including
foundations, columns, load-bearing walls, floor diaphragms, roof diaphragms, and
Fig. 1 Posters to create awareness among the students while in hostels and classrooms
Fig. 2 Pamphlet for information dissemination regarding EEW mock drill

492 P. Kumar et al.
connections between these structural elements. Replacement of an entire building

is typically more expensive than retrofitting. Still, it may be appropriate if retrofit
costs are high, especially if the existing building is in overall poor condition, needs
non-seismic repairs, is near the end of its useful life, is functionally obsolescent, or
has other deficiencies such as not being energy efficient. The facilities that may serve
as emergency shelters should be designed and constructed with the higher than the
minimum seismic standards required by building codes.
Retrofitting a building is a costly affair. Certain studies demonstrate the cost of
retrofitting school buildings and its benefit. In seismically active zones, small age
group students are most vulnerable to earthquakes as in their routine; they stay in
the schools for around eight hours on working days. In the benefit-cost analysis,
two points play an important role in retrofitting schools against earthquake damage.
First, if one considers only the physical damage to the building, the benefit-cost ratio
will be considerably less than one, and the measure will be deemed economically
unattractive. While in the second point, when one adds the reduction in fatalities by
retrofitting the structure, then this measure is likely to yield a benefit-cost ratio greater
than one. It is reasonable to consider the values of lives in determining the benefit-
cost ratio. Retrofitting of all school buildings in India would require a significant
investment, i.e., $65 billion [46] (As per the study carried out in 2012). Now, it is the
year 2022, so the cost could be hundreds of billions of dollars.
The low-cost mitigation solutions can be opted. The EEWS can play a part in
comprehensive earthquake risk reduction. The cost to set up an EEWS is not very
high, on the contrary it, is affordable, and can be implemented with low investment.
Infrastructure improvements are a selective approach while EEWS caters the service
to all members of the population equally.
India has diverse geographical aspects, consisting of densely populated cities in
natural hazard-prone areas. Investments are required to mitigate hazards and make
society disaster resilient. The disaster management act—2005, section 43, clearly
states need for the creation of the National Disaster Mitigation Fund, State Disaster
Mitigation Fund, and District Disaster Mitigation Funds to be used in hazard mitiga-
tion works in the society. In finance commission II–VIII, the money margin scheme
was created to provide funds for disaster relief operations. The IX Finance commis-
sion recommended calamity relief fund scheme and more autonomy was given to
states to use the funds in relief operations. In X finance commission recommended
that along with the calamity relief fund, a national fund for calamity relief fund should
be created. The XI finance commission recommended that the national calamity relief
fund should be abolished and create a national climate contingency fund. The XII
finance commission maintained the calamity relief fund and national climate contin-
gency fund. Finance commission XIII recommended the merger of the calamity relief
fund with the state disaster response fund. The XIV finance commission recom-
mended a national disaster response fund, state disaster response fund and district
disaster response fund. The finance commission XV has recommended setting up of
national disaster mitigation fund at the national level and state disaster mitigation
fund at state level. Instead of creating funds for them individually, these were merged
with response fund in the head of the national disaster response and mitigation fund,
and state disaster response and mitigation fund, and in the allotted funds, 20% would
be used for mitigation measures and 80% for response operations [47]. These funds
are available for the states to be used in the mitigation measure from the financial
year 2021–22 to 2025–26 [48].
5 Losses Due to Earthquakes
On 4 April 1905, Kangra earthquake of Ms 8.0 was one of the most devastating earth-
quakes in northern India. This earthquake killed around 20,000 people and around
53,000 domestic animals were also lost. The total economic cost of recovering from
the effects of this earthquake was 2.9 (1905) Indian rupees [49]. On 15 January
1934, Bihar-Nepal earthquake of Mw 8.1 caused 10,700 deaths and maximum inten-
sity X was observed in this event. On 15 August 1950, Assam–Tibet border earth-
quake of Mw 8.6 caused 1926 death and extreme loss of around $25 million and
maximum intensity IX, were observed during this earthquake [50]. On 20 October
1991, Uttarkashi earthquake of Mw 6.8, killed 768 people, injured 5,066 people,
and damaged around 42,400 houses [51]. On 29 March 1999, an earthquake of Mw
6.6 struck in Chamoli region of Uttarakhand. It killed about 100 people and injured
several hundred. The maximum intensity of VII was observed during this earthquake
[52]. On 25 April 2015, the most recent Nepal earthquake of Mw 7.6 and its after-
shocks caused a lot of losses in the Himalayan country. It killed 9000 people and
injured 23,000, rendering 785,000 people homeless. Due to this earthquake, about
2.8 million people were displaced. This earthquake caused a lot of economic loss
and its cost was around $7 billion, one-third of Nepal’s GDP. Figure 3 shows some
damaging earthquake and human and economic loss due to them. The trends in
Fig. 3b clearly show that the economic loss will be more due to large earthquakes.
Therefore, an EEWS can be an effective tool for alerting people so that they can
follow precautionary measures to save themselves and their near and dear ones.
6 Maps
A light earthquake of Mw 4.7 struck in Chamoli region of Uttarakhand, and an alert

was issued to the public by UEEW System. A peak ground acceleration map is
prepared based on the acceleration data recorded by the sensors. Figure 4 shows a
map of the variation of acceleration in the region. The acceleration maps are a very
important tool for disaster management authorities in carrying out post-earthquake
operations. A large earthquake is overdue in the Uttarakhand region, therefore having
an early warning system and its secondary products like acceleration maps. These
maps can expedite post-earthquake operations like search and rescue, identification
of severely affected areas, and rehabilitation and planning activities required to cope
with the effects of earthquake damage.
494 P. Kumar et al.
Fig. 3 a The fatalities during some disastrous earthquakes triggered in mainland India and the
Himalayas. b The economic loss is due to these earthquakes
Fig. 4 The PGA map of 11 September 2021, Mw—4.7 Chamoli earthquake

7 Tentative Lead Time
The lead time for the user is important. It is the remaining time for the S-waves
to reach at the target site. Lead time varies depending on the location of the user
from the epicenter of the earthquake. A user at a far distance would get more lead
time than the users at the nearest locations. The existing system is developed based
on the regional EEW concept therefore it takes few seconds to process and then
issue warnings. These few seconds are elapsed time to get at least 3 s of data from
at least four sensors and time needed for transmission, processing, magnitude, and
hypocenter estimation, and then sending the decision to the warning server.
The lead time can be calculated as per Eq. (1) [53], where reporting time (T r ) is
the time needed (T d ) to trigger and record a sufficient length of the waveforms and
the time (T pr ) required to process the waveforms for the hypocenter and magnitude
determination. The earthquake early warning time (T w ) is given by:
Tw = Ts − Tr (1)
and
Tr = Td + Tpr (2)
where T s is the destructive S-wave travel time. For the advanced warning, we must
have T w > 0. Clearly, this requires T s > (T d + T pr ). The locations for which T w < 0
are considered as blind zone.
Figure 5 shows a diagram that describes the expected lead time for the few major
cities of Uttarakhand and nearby states if a large earthquake hits the Uttarkashi region.
8 Conclusion
The target G of the Sendai framework clearly mentions the importance of a multi-
hazard early warning system. The retrofitting measures to strengthen the building
capacity is costly, and the EEWS is less expensive with a wider reach to all sections
of the society. A successful EEWS saves lives and jobs, land, and infrastructures and
supports long-term sustainability. It can help government officials in their planning,
saving money in the long run, and protecting the economy. The government of India
has allocated funds for the mitigation measure. Some share of it can be used to develop
pan-India EEWS. The awareness of the earthquake in mass is very important. Mock
drills can help in this to follow a set of rules which are followed in the real scenario.
The practice of drop, cover and hold on drills may reduce injuries from falling objects.
496 P. Kumar et al.
Fig. 5 The expected lead time for the major cities of Uttarakhand if a large earthquake hits the
Uttarkashi region
References
1. Baxter, S.J.: Special Publication No. 23, Earthquake Basics (2000). https://www.dgs.udel.edu/
sites/default/files/publications/sp23.pdf
2. Earle, S.: Physical Geology (2015). https://opentextbc.ca/geology/
3. Sharma, M.L.: Seismic Hazard in the Northern India Region. Seismol. Res. Lett. 74, 141–147
(2003)
4. Sharma, M.., Shanker, D.: Estimation of Seismic Hazard Parameters in the Himalayas and its
Vicinity from Mixed Data Files. http://home.iitk.ac.in/~vinaykg/Iset413.pdf (2001)
5. Sharma, M.L., Lindholm, C.: Earthquake Hazard assessment for Dehradun, Uttarakhand, India,
including a characteristic earthquake recurrence model for the Himalaya Frontal Fault (HFF).
Pure Appl. Geophys. 169, 1601–1617 (2012). https://doi.org/10.1007/s00024-011-0427-7
6. Geller, R.J., Jackson, D.D., Kagan, Y.Y., Mulargia, F.: Earthquake cannot be predicted. Science
275(80), 1616–1617 (1997). https://doi.org/10.1126/science.275.5306.1616
7. UNISDR: Sendai Framework for Disaster Risk Reduction 2015–2030 (2015)
8. Allen, R.M., Gasparini, P., Kamigaichi, O., Bose, M.: The status of earthquake early warning
around the world: an introductory overview. Seismol. Res. Lett. 80, 682–693 (2009). https://
doi.org/10.1785/gssrl.80.5.682
9. Nakamura, Y., Saita, J.: FREQL and AcCo for a quick response to earthquakes. Earthq. Early
Warn. Syst. 307–324 (2007). https://doi.org/10.1007/978-3-540-72241-0_15
10. Aranda, J.M.E., Jimenez, A., Ibarrola, G., Alcantar, F., Aguilar, A., Inostroza, M., Maldonado,
S.: Mexico City Seismic Alert System. Seismol. Res. Lett. 66, 42–53 (1995). https://doi.org/
10.1785/gssrl.66.6.42
11. Hoshiba, M., Kamigaichi, O., Saito, M., Tsukada, S., Hamada, N.: Earthquake early warning
starts nationwide in Japan. Eos (Washington, DC) 89, 73–74 (2008). https://doi.org/10.1029/
2008EO080001
12. Kamigaichi, O., Saito, M., Doi, K., Matsumori, T., Tsukada, S., Takeda, K., Shimoyama, T.,
Nakamura, K., Kiyomoto, M., Watanabe, Y.: Earthquake early warning in Japan: warning the
general public and future prospects. Seismol. Res. Lett. 80, 717–726 (2009). https://doi.org/
10.1785/gssrl.80.5.717
13. Hsiao, N.C., Wu, Y.M., Shin, T.C., Zhao, L., Teng, T.L.: Development of earthquake early
warning system in Taiwan. Geophys. Res. Lett. 36, (2009). https://doi.org/10.1029/2008GL
036596
14. Chen, D.Y., Hsiao, N.C., Wu, Y.M.: The earthworm based earthquake alarm reporting system
in Taiwan. Bull. Seismol. Soc. Am. 105, 568–579 (2015). https://doi.org/10.1785/0120140147
15. Wu, Y.M., Mittal, H., Huang, T.C., Yang, B.M., Jan, J.C., Chen, S.K.: Performance of a low-
cost earthquake early warning system (P-ALErt) and shake map production during the 2018
M w 6.4 Hualien, Taiwan, earthquake. Seismol. Res. Lett. 90, 19–29 (2019). https://doi.org/
10.1785/0220180170
16. Espinosa-Aranda, J.M., Jimenez, A., Ibarrola, G., Alcantar, F., Aguilar, A., Inostroza, M.,
Maldonado, S.: Results of the Mexico City early warning system. In: Eleventh World
Conference on Earthquake Engineering, pp. 1–8 (1996)
17. Espinosa-Aranda, J.M., Cuellar, A., Garcia, A., Ibarrola, G., Islas, R., Maldonado, S.,
Rodriguez, F.H.: Evolution of the Mexican Seismic Alert System (SASMEX). Seismol. Res.
Lett. 80, 694–706 (2009). https://doi.org/10.1785/gssrl.80.5.694
18. Espinosa-Aranda, J.M., Cuéllar, A., Ibarrola, G., Islas, R., García, A., Rodríguez, F.H.,
Frontana, B.: The Seismic Alert System of Mexico (SASMEX) and their alert signals broadcast
results. In: 15th World Conference on Earthquake Engineering, pp. 1–9 (2012)
19. Sheen, D., Park, J., Chi, H., Hwang, E., Lim, I., Seong, Y.J., Pak, J.: The first stage of an
earthquake early warning system in South Korea. Seismol. Res. Lett. 88, 1491–1498 (2017).
https://doi.org/10.1785/0220170062
20. Erdik, M., Fahjan, Y., Ozel, O., Alcik, H., Mert, A., Gul, M.: Istanbul earthquake rapid response
and early warning system. Bull. Earthq. Eng. 1, 157–163 (2003)
21. Alcik, H., Ozel, O., Apaydin, N., Erdik, M.: A study on warning algorithms for Istanbul
earthquake early warning system. Geophys. Res. Lett. 36, 3–5 (2009). https://doi.org/10.1029/
2008GL036659
22. Carranza, M., Buforn, E., Colombelli, S., Zollo, A.: Earthquake early warning for southern
Iberia: a P wave threshold-based approach. Geophys. Res. Lett. 40, 4588–4593 (2013). https://
doi.org/10.1002/grl.50903
23. Pazos, A., Romeu, N., Lozano, L., Colom, Y., Mesa, M.L., Goula, X., Jara, J.A., Cantavella,
J.V., Zollo, A., Hanka, W., Carrilho, F.: A regional approach for earthquake early warning in
South West Iberia: a feasibility study. Bull. Seismol. Soc. Am. 105, 560–567 (2015). https://
doi.org/10.1785/0120140101
24. Romeu Petit, N., Colom Puyané, Y., Jara Salvador, J.A., Goula Suriñach, X., Susagna Vidal,
T.: Development of an earthquake early warning system based on Earthworm: application to
Southwest Iberia. Bull. Seismol. Soc. Am. 106, 1–12 (2016). https://doi.org/10.1785/012015
0192
25. Zollo, A., Iannaccone, G., Lancieri, M., Cantore, L., Convertito, V., Emolo, A., Festa, G.,
Gallovič, F., Vassallo, M., Martino, C., Satriano, C., Gasparini, P.: Earthquake early warning
system in southern Italy: Methodologies and performance evaluation. Geophys. Res. Lett. 36,
(2009). https://doi.org/10.1029/2008GL036689
26. Satriano, C., Elia, L., Martino, C., Lancieri, M., Zollo, A., Iannaccone, G.: PRESTo, the earth-
quake early warning system for Southern Italy: concepts, capabilities and future perspectives.
Soil Dyn. Earthq. Eng. 31, 137–153 (2011). https://doi.org/10.1016/j.soildyn.2010.06.008
27. Wenzel, F., Oncescu, M.C., Baur, M., Fiedrich, F., Ionescu, C.: An early warning system for
Bucharest. Seismol. Res. Lett. 70, 161–169 (1999). https://doi.org/10.1785/gssrl.70.2.161
28. Böse, M., Sokolov, V., Wenzel, F.: Shake map methodology for intermediate-depth Vrancea
(Romania) earthquakes. Earthq. Spectra. 25, 497–514 (2009). https://doi.org/10.1193/1.314
8882
29. Böse, M., Ionescu, C., Wenzel, F.: Earthquake early warning for Bucharest, Romania: novel
and revised scaling relations. Geophys. Res. Lett. 34, 1–6 (2007). https://doi.org/10.1029/200
7GL029396
30. Peng, H., Wu, Z., Wu, Y.-M., Yu, S., Zhang, D., Huang, W.: Developing a prototype earthquake
early warning system in the Beijing Capital Region. Seismol. Res. Lett. 82, 394–403 (2011).
498 P. Kumar et al.
31. Wang, Y., Li, S., Song, J.: Threshold based evolutionary magnitude estimation for an earthquake
early warning system in the Sichuan—Yunnan region, China. Sci. Rep., 1–12 (2020). https://
doi.org/10.1038/s41598-020-78046-2
32. Crowell, B.W., Schmidt, D.A., Bodin, P., Vidale, J.E., Baker, B., Barrientos, S., Geng, J.: G-
FAST earthquake early warning potential for great earthquakes in Chile. Seismol. Res. Lett.
89, 542–556 (2018). https://doi.org/10.1785/0220170180
33. Brooks, B.A., Protti, M., Ericksen, T., Bunn, J., Al, B.E.T.: Robust earthquake early warning
at a fraction of the cost: ASTUTI Costa Rica. AGU Adv. 2, 1–17 (2021). https://doi.org/10.
1029/2021AV000407
34. Porras, J., Massin, F., Arroyo-solórzano, M., Arroyo, I.: Preliminary results of an earthquake
early warning system in Costa Rica. Front. Earth Sci. 9, 1–13 (2021). https://doi.org/10.3389/
feart.2021.700843
35. Massin, F., Clinton, J., Böse, M.: Status of earthquake early warning in Switzerland. Front.
Earth Sci. 9, 1–20 (2021). https://doi.org/10.3389/feart.2021.707654
36. Strauch, W., Talavera, E., Tenorio, V., Ramírez, J., Argüello, G., Herrera, M., Acosta, A.,
Morales, A.: Toward an earthquake and Tsunami monitoring and early warning system for
Nicaragua and Central America. Seismol. Res. Lett. 89, 399–406 (2018). https://doi.org/10.
1785/0220170193
37. Nof, R.N., Kurzon, I.: TRUAA—earthquake early warning system for Israel: implementation
and current status. Seism. Instr. 92, 325–341 (2021). https://doi.org/10.1785/0220200176.Int
roduction
38. Nof, R.N., Lior, I., Kurzon, I.: Earthquake early warning system in Israel—towards an
operational stage. Front. Earth Sci. 9, 1–12 (2021). https://doi.org/10.3389/feart.2021.684421
39. Cua, G., Fischer, M., Heaton, T., Wiemer, S.: Real-time performance of the virtual seismologist
earthquake early warning algorithm in Southern California. Seismol. Res. Lett. 80, 740–747
40. Allen, R.M., Melgar, D.: Earthquake early warning: advances, scientific challenges, and societal
needs. Annu. Rev. Earth Planet. Sci. 47, 361–388 (2019). https://doi.org/10.1146/annurev-
earth-053018-060457
41. Allen, R.M., Brown, H., Hellweg, M., Khainovski, O., Lombard, P., Neuhauser, D.: Real-time
earthquake detection and hazard assessment by ElarmS across California. Geophys. Res. Lett.
36, 1–6 (2009). https://doi.org/10.1029/2008GL036766
42. Kumar, P.: Earthquake Early Warning System: Site Classification, Intensity Map and Attributes
(2020)
43. Chamoli, B.P., Kumar, A., Chen, D.Y., Gairola, A., Jakka, R.S., Pandey, B., Kumar, P., Rathore,
G.: A prototype earthquake early warning system for Northern India. J. Earthq. Eng. 25, 2455–
2473 (2019). https://doi.org/10.1080/13632469.2019.1625828
44. Kumar, A., Chamoli, B.P., Pandey, B., Kumar, P., Rathore, G.: Earthquake early warning
system: its relevance for India. In: Sitharam, T.G., Kolathayar, S., Jakka, R.S., and Matsagar,
V. (eds.) Theory and Practice in Earthquake Engineering and Technology. Springer Tracts in
Civil Engineering, Springer Tracts in Civil Engineering, Springer, Singapore, pp. 233–260
(2022)
45. National Commission on Population: Population projections for India and states 2011–2036.
Report of the technical group on population projections (2019)
46. Kunreuther, H., Michel-Kerjan, E.: Challenge paper: natural disasters. Policy Options for
Reducing Losses from Natural Disasters: Allocating $75 billion (2012)
47. Sharma, S.: Exploring disaster mitigation in India: a financial viewpoint. Emerg. Econ. Stud.
7, 7–22 (2021). https://doi.org/10.1177/23949015211057915
48. Government of India: Report of fifteenth finance commission (2021–2026) (2020)
49. Ambraseys, N., Bilham, R.: A note on the Kangra Ms = 7.8 earthquake of 4 April 1905. Curr.
Sci. 79, 45–50 (2000)
50. USGS National Earthquake Information Center: M 8.6—1950 Assam-Tibet Earthquake,
https://earthquake.usgs.gov/earthquakes/eventpage/official19500815140934_30/impact
51. Jain, S.K., Das, S.: Analysis of strong motion from Uttarkashi earthquake for assessment of
code provisions for different Seismic Zones. Earthq. Spectra. 9, 739–754 (1993)
52. Jain, S.K., Murty, C.V.R., Arlekar, J.N.: Chamoli (Himalaya, India) Earthquake of 29 March
1999, (1999)
53. Satriano, C., Wu, Y.M., Zollo, A., Kanamori, H.: Earthquake early warning: concepts, methods
and physical grounds. Soil Dyn. Earthq. Eng. 31, 106–118 (2011). https://doi.org/10.1016/j.
soildyn.2010.07.007
Application of Regression Techniques
for Preparing a Homogeneous
Earthquake Catalog—An Overview
Ranjit Das, H. R. Wason, and Claudio Meneses
Abstract Standard least squares regression (SLR) and general orthogonal regression
(GOR) address different questions and make different assumptions about measure-
ment errors in one or both of the variables. SLR minimizes the sum of squares of the
vertical deviations and provides estimation of the dependent variable (Y ). It assumes
that the independent variable (X) is an observed value which is known without error,
and only the dependent variable (Y ) suffers from measurement error. GOR, on the
other hand, yields a linear relationship (Yt = β0 + β1 X t ) between the dependent (Y t )
and the independent (X t ) variables based on observed data (X, Y ) having measurement
errors in both the variables involved. Therefore, it is mathematically incorrect to use
observed value X in place of X t in the equation Yt = β0 + β1 X t, and thus, if done so
(as in the conventional GOR method), this procedure will produce biased estimates.
The present study is an overview of different methodologies used for preparing a
homogeneous earthquake catalog for different seismic environments. The error vari-
ance ratio (η) used in GOR has not been addressed in seismological literature, and
this overview will address this critical issue as well. This study will also suggest a
guideline for the use of regression methods for preparing a homogeneous earthquake
database which is an important input to obtain improved seismic hazard assessment.
1 Introduction
Seismic catalogs generally encompass occurrence time, hypocentral coordinates,

earthquake size, and various others seismic parameters related to earthquake. Earth-
quake catalogs give useful information regarding seismicity study, seismotectonics,
R. Das (B) · C. Meneses

Computer Science and Engineering Department, Universidad Catolica del Norte, Antofagasta,
Chile
e-mail: ranjit.das@ucn.cl
H. R. Wason
Indian Institute of Technology Roorkee, Roorkee, India
https://doi.org/10.1007/978-981-99-1459-3_40
502 R. Das et al.
and earthquake hazard and risk of an earthquake-prone area. A homogeneous earth-

quake catalog is of crucial importance for understanding earthquake patterns in space
and time, land use planning, seismic risk, and other earthquake-related studies. All
these studies need well defined and consistent earthquake magnitudes. Historical
earthquakes mainly before 1976 don’t fulfill this criterion.
The first earthquake size measurement scale was proposed by [1] considering
seismic body wave recordings from standard Wood–Anderson torsion seismometers
using Southern Californian seismicity. Seismologists had felt a need for expansion of
the Richter magnitude scale to produce data in different environments using different
seismic wave types. These efforts culminated in formulation of additional magni-
tude scales, namely body wave magnitude (mb /mB ), surface wave magnitude (M s ),
moment magnitude (M w ), Das magnitude (M wg ). Most of these earthquake size esti-
mates use earthquake source process and, therefore, are contained with observation
errors. In seismic applications dealing with earthquake catalogs including seismic
hazard and risk assessment of a seismic region, it is required to define different
magnitude types into a preferred magnitude type.
Standard linear regression (SLR) is the most used regression methodology for
homogeneous earthquake catalog preparation considering one of the variables is
either error-free, or the order of its error is negligibly small compared to the measure-
ment error of the other variable. However, SLR is not appropriate considering the
measurement error inherent in the independent variable. As an alternative, it is better
to use general orthogonal regression (GOR) relation, which considers the errors on
both the dependent and the independent variables ([2–5]). More recent work for earth-
quake magnitude conversion, however, considers the fact that both the earthquake
magnitudes involved are affected by errors. Related research papers in statistical
seismology, listed chronologically, have been published by [4–14]. In these publi-
cations, the approach used is called GOR. In statistical literature, this approach is
also called orthogonal regression ([15]). Many statisticians (e.g., [15]) asserted that
GOR is applied incorrectly in statistics and hence provide overestimated slope. [15]
suggested to modify the error variance ratio (ŋ) by incorporating the equation error.
Recently, Das and co-workers pointed out the limitations of GOR and suggested a
procedure to remove the limitations of GOR.
GOR yields a linear relationship between the dependent (Y t ) and the independent
variable (X t ) based on observed data (X obs , Y obs ) in the form of Y t = β0 + β1 X t which
is incorrectly used in GOR2 application as Y t = β0 + β1 X obs . Inappropriateness of
using equation Y t = β0 + β1 X obs instead of equation Y t = β0 + β1 X t produces bias in
GOR2 (conventional GOR). An exception is the paper of [4] who use the theoretical
true points of GOR line given by [16].
Application of Regression Techniques … 503
2 Regression Methodologies
2.1 SLR
As SLR is the most used regression technique, it details are skipped here for brevity.
This well-known linear regression procedure applies when the variance of indepen-
dent variable tends to zero and only error is in the dependent variable. Under these
assumptions, the best fitting line is achieved by minimizing the square sum of the
vertical distances between the true points/experimental (X i , yi ) on the line and their
equivalents observed points (X i ,Y i ), that is,

n
(Yi − βslr X i − β1slr )2 (1)
i=1
where β0 slr andβ1 slr represent the slope and intercept of SLR line.
2.2 GOR
The general orthogonal regression (GOR) methodology is very popular among

researchers in different areas since the work of [16]. Earlier studies focused on GOR
estimation, often in different ways. The contributors included [16–20], and many
others. Methodology of GOR has been defined and re-defined by many investigators
(e.g., [4, 5, 10, 12, 15, 21]). GOR takes place when (1) two procedures measure the
same quantity, or (2) when dependent and independent variables are related to each
other by following the same physical laws ([15]). The basic methodology of GOR is
discussed in different studies (e.g., [8, 13, 15, 16, 20]).
General orthogonal regression (GOR) is based on the minimization of the squares
of the statistical Euclidean distances ([22]), and minimization distance is given below
2

n
Yi − β0 − β1 xi (X i − xi )2
+ (2)
i=1
σu2 σe2
In Eq. (2), (X i , Y i ) and (x i , yi ) denote the observed and the true (on GOR line)
values, respectively, of the n data pairs of the independent and the dependent vari-
ables. The slope and intercept of the GOR line are β0 and β 1 , respectively. σe2 and σu2
represent the errors of X i and Y i , respectively. GOR relationships among different
earthquake size estimates have been discussed in many studies (e.g., [3, 4, 7, 8, 10,
12, 13]). It has been observed in several publications of Das and co-authors (e.g.,
[5]) that Eq. (2) is inappropriately presented in some research papers (e.g., Eq. (8)
of [7]). In fact, a GOR relation for magnitude pairs (Xt , Yt ) provides the following
form:
504 R. Das et al.
yt = β̂1 xt + β̂0 (3)
where x t (unlike X t ) is the abscissa of a theoretical true point (x t , yt ) on the GOR line
and β̂1 , β̂0 represent the slope and the intercept of the GOR line, respectively. The
theoretical true point does not indicate the natural true value; it indicates theoretically
achieving the best unbiased value (x t , yt ) on the GOR line. Users cannot use Eq. (3)
for estimation because users have only the observed value (X t ), not the theoretical
true value (x t ) ([5]). Note that in the estimation stage, the measurement error in X t
is ignored and the following equation
yt = β̂1 X t + β̂0 (4)
with X t with the same slope and intercept as in Eq. (3) above is incorrectly used. This
form of GOR is referred to as GOR2 (as discussed above). This incorrect way of
using GOR introduces serious bias in the dependent variable estimations that lead to
biased estimations of seismicity parameters and eventually to seismic hazard results
(e.g., [5]).
2.3 A Graphical Representation of GOR
The graphical representation of GOR is provided below for easy understanding of the
limitations involved in the GOR method. It is observed that the problem of GOR is
not well addressed in the existing literature; therefore, to provide a clear and complete
view of GOR, two different cases are discussed below ([5]).
Case I:
A SLR line is obtained considering three observed data pairs (X 1 , Y 1 ), (X 2 , Y 2 ), and
(X 3 , Y 3 ), and the equivalent true points/experimental points on the SLR line are (X t1
= X 1 , Y t1 ), (X t2 = X 2 , Y t2 ), and (X t3 = X 3 , Y t3 ), respectively. These experimental/true
points on the line are used in deriving the best fitting SLR line by minimizing the
square sum of the vertical residuals. After substituting the independent variables X 1 ,
X 2 , X 3 in SLR line, the true points/experimental points can be obtained that had been
used in the preparing the best fitting SLR line (see Fig. 1a).
Case II:
A GOR line is obtained considering three observed data pairs (X 1 , Y 1 ), (X 2 , Y 2 ),
and (X 3 , Y 3 ) having dependent and independent variable errors. The theoretical
true points/experimental points of these observed data pairs have been obtained
considering statistical Euclidean distance and are (X t1 , Y t1 ), (X t2 , Y t2 ), and (X t3 ,
Y t3 ). It is not possible to achieve theoretical true points/experimental points on
substitution of observed dependent variables (i.e., X 1 , X 2 , X 3 ) in the GOR line just
like case I; hence, conventional GOR procedure doesn’t follow the basic norms of
mathematics and produces biased results. Rather than achieving the theoretical true
Fig. 1 Schematic diagram showing true points/experimental points on the fitted regression line for
a set of three observed points (X 1 , Y 1 ), (X 2 , Y 2 ), and (X 3 , Y 3 ), a SLR line, b GOR line
points/experimental point, a totally different points is achieved in the conventional

GOR (see Fig. 1b).
2.4 Chi-Squared Regression
The chi-square regression methodology considered in seismic study mainly origi-

nated from [6], and as per authors, the chi-square regression procedure is given as
below.
n
(Msi − am bi − b)2
Chi-square = i=1 (5)
σ (Msi ) + a 2 σ 2 (m bi )
2
In Eq. (5), the terms M Si represents the dependent variable and mbi represents
the independent variable. The constant terms ‘a’ represents slope and ‘b’ represents
intercept. The denominator of Eq. (5) is the equivalent of Var (M si − ambi − b) from
Gaussian error propagation law ([23]):

δ(Msi − am bi − b)2
Var (MSi − am bi − b) ≈ Var (Msi )
δ(Msi )

δ(Msi − am bi − b)2
+ Var (Mbi ) (6)
δ(Mbi )
= σ 2 (Msi ) + a 2 σ 2 (m bi ) (7)
GOR given by [16] and other statisticians are the same with the chi-squared regres-
sion of [6] for the linear case. Stromeyer et al. [6] used the case of uncorrelated errors
506 R. Das et al.
in variables. Therefore, there is no difference between [6] and [16]; the only differ-
ence is the name. Some literature incorrectly stating that chi-square methodology of
[6] and GOR methods are two different methodologies, but in fact, they are the same.
3 Homogenization of Earthquake Catalogs in Seismic

Literature
Homogenized earthquake catalogs have been prepared all over the world using regres-
sion techniques. Reliable past earthquake catalog for the Indian region exits for about
past 100 years. A general compilation of the earthquake catalog for the Indian region
has been made by [24], a detailed catalog for earthquakes occurring in India from
historical times to 1869 has been made by [25]. Later several researchers made signif-
icant compilations of earthquake catalog for the Indian region [26–31]. Srivastava
and Ramachandran [32] has prepared earthquake catalog using events data from 1594
to 1975 covering latitudes 5° N–28° N and longitudes 67.5° E–90° E. The Indian
Society of Earthquake Technology prepared an earthquake catalog [30] for India and
its adjoining region up to 1979. Bapat et al. [33] prepared a catalog for Northeast India
and neighborhood area. Several other researchers who complied earthquake catalogs
for India and neighboring region include [34–44]. All these seismic records use
different earthquake size and are heterogeneous in nature. Recently, unified homo-
geneous earthquake catalogs have been made by several researchers for the Indian
region or specific areas like Northeast India region (e.g., [2, 13, 42, 45–48]).
To understand the effect of magnitude conversion in seismicity parameters, we
used homogenized earthquake catalog of CIGIDEN (Centro National de investiga-
tion Para la Gestion Intergrade de desastres Naturale) of 39,976 events for Northern
Chile and Southern Peru for the period 151,302,016 using GOR1, GOR2, and SLR
approaches. The unified catalog has been declustered and estimate magnitude of
completeness Mc and Gutenberg–Richter parameters (‘b’, ‘a’) [5]. Our analysis
finds a biased in the Gutenberg–Richter parameter b around 5–42%. The differ-
ence between magnitude of completeness Mc given by GOR1 and SLR vary in the
range 0–9%. The biased introduced in the total activity rate ‘a’ due to incorrect use
of regression vary in the range 4–33%
4 Conclusions
GOR is the best method for encountering errors of both the variables involved only
when it appropriately considers the equation error. Many statisticians reported that
due to incorrect use of error variance ratio (ŋ), GOR gives overestimated slope. It is
not correct to use ŋ = σσue [16] because ŋ value does not consider the equation error.
The suggested method for estimation of ŋ by [15] is not so simple but important to
address the equation error for GOR to get the better accuracy. A correct form of GOR
has been suggested by [5] and their earlier publications. Improved GOR technique
of [5] is the best GOR technique because (1) It is correctly expressed in terms of
observed variables; (2) It incorporates equation error; (3) It provides improved values
of uncertainties in terms of slope and intercept; (4) It removes the overestimation
problem of conventional GOR, and (5) It provides improved correlation coefficient
Rxy and root mean squared error (RMSE) values.
Significant variations in the seismicity parameters have been observed due to
incorrect use of regression analysis; therefore, it is important to use the proper
regression methodology for seismicity and seismic hazard assessment study.
Chi-square regression of [6] and GOR in statistical literature are the same. There
is no added advantage of using Chi-square regression in place of GOR method.
Homogeneous earthquake catalog using GOR technique helps us to address the
biased in seismicity parameters. GOR technique further improved the seismic hazard
results by encountering the error in magnitude conversion.
Acknowledgement The article has benefitted from FONDECYT Grant 11200618.
References
1. Richter, C. F.: An instrumental earthquake magnitude scale. Bull. Seismol. Soc. Am. 25, 1–32
(1935). ISSN 0037-1106
Seismicity in North–East India and Its Adjoining Region. J. Seismolog. 12, 107–123 (2008)
3. Ristau, J.: Comparison of magnitude estimates for New Zealand earthquakes: moment magni-
tude, local magnitude, and teleseismic body-wave magnitude. Bull. Seism. Soc Am. 99,
1841–1852 (2009)
4. Das, R., Wason, H.R., Sharma, M.L.: Unbiased estimation of moment magnitude from body
and surface wave magnitudes. Bull. Seism. Soc. Am. 104, 1802–1811 (2014)
5. Das, R., Wason, H.R., Gonzalez, G., Sharma, M.L., Chodhury, D., Roy, N., Salazar, P.:
Earthquake magnitude conversion problem. Bull. Seismol. Soc. Am. 108(4), 1995–2007 (2018)
6. Stromeyer, D., Gruünthal, G., Wahlström, R.: Chi-square regression for seismic strength 216
parameter relations, and their uncertainties, with applications to an Mw based earthquake
catalog 217 for central, northern and north-western Europe. J. Seismol. 8(1), 143–153 (2004).
218 https://doi.org/10.1023/B:JOSE.0000009503.80673.51
7. Castellaro, S., Mulargia, F., Kagan, Y.Y.: Regression problems for magnitudes. Geophys. J.
Int. 165, 913–930 (2006)
8. Das, R., Wason, H.R., Sharma, M.L.: Global regression relations for conversion of surface
wave and body wave magnitudes to moment magnitude. Nat. Hazard 59, 801–810 (2011)
9. Gutdeutsch, R., Castellaro, S., Kaiser, D.: The magnitude conversion problem: further insights.
Bull. Seis. Soc. Am. 101, 379–384 (2011)
10. Wason, H.R., Das, R., Sharma, M.L.: Magnitude conversion problem using general orthogonal
regression. Geophys. J. Int. 190(2), 1091–1096 (2012)
11. Lolli, B., Gasperini, P.: A comparison among general orthogonal regression methods applied
to earthquake magnitude conversions. Geophys. J. Int. 1135–1151, 190 (2012)
12. Das, R., Wason, H.R., Sharma, M.L.: Magnitude conversion to unified moment magnitude
using orthogonal regression relation. J. Asian Earth Sci. 50, 44–51 (2012)
508 R. Das et al.
13. Das, R., Wason, H.R., Sharma, M.L.: General orthogonal regression relations between body
wave and moment magnitudes. Seismol. Res. Lett. 84, 219–224 (2013)
14. Gasperini, P., Lolli, B., Castellaro, S.: Comparative analysis of regression methods used 198
for seismic magnitude conversions. Bull. Seism. Soc. Am. 105(3), 1787–1791 (2015). https://
doi.org/10.1785/0120150018
15. Carroll, R.I., Ruppert, D.: The use and misuse of orthogonal regression in linear errors-in-
variables models. Am. Stat. 50(1), 1–6 (1996)
16. Fuller, W.A.: Measurement Error Models. Wiley, New York (1987)
17. Adcock, R.J.: A problem in least squares. Analyst 5, 53–54 (1978)
18. Kummell, C.: Reduction of observation equations which contain more than one observed
quantity. Analyst 6, 97–105 (1879)
19. Pearson, K.: On lines and planes of closest fit to systems of points in space. Philos. Mag. 2,
559–572 (1901)
20. Lindley, D.: Regression lines and the linear functional relationship. J. Roy. Stat. Soc. Supp. 9,
218–229 (1947)
21. Madansky, A.: The fitting of straight lines when both variables are subject to error. Am. Stat.
Assoc. J. 54, 173–205 (1959)
22. Anderson, T.W.: Estimating linear statistical relationships. Ann. Stat. 12, 1–45 (1984)
23. Williamson, J.: Least-square fitting of a straight line. Can. J. Phys. 46, 1845–1847 (1968)
24. Press, W.H., Teukolsky, S.A., Vetterling, W.T. and F.B.P.: Numerical Recipes in C The Art of
Scientific Computing, University Press, Cambridge (1992)
25. Baird-Smith, R.: Memoirs of Indian earthquakes. J. Asiatic Soc. Bengal 13(2), 964–983 (1844)
26. Oldham, T.: A catalogue of Indian earthquakes from the earliest times to the end of AD 1869.
Mem. Geol. Surv. India 19(3), 1–53 (1883)
27. Gubin, I.E.: Seismic zoning of the Indian Peninsula. Bull. Int. Inst. Seism. Earthq. Eng. 5,
109–139 (1968)
28. Kelkar, Y.N.: Earthquakes in Maharashtra in last 300 years. Kesari Daily (Marathi newspaper),
Pune, January 7 (1968)
29. Guha, S.K., Gosavi, P.D., Nand, K., Padale, J.G. and Marwadi, S.C.: Koyna earthquakes
(October 1963–December 1973), Central Water and Power Research Station (CWPRS), Pune,
India (1974)
30. Chandra, U.: Earthquakes of Peninsular India: a seismotectonic study. Bull. Seism. Soc. Am.
67, 1387–1413 (1977)
31. Rao, B.R., Rao, P.S.: Historical seismicity of peninsular India. Bull. Seism. Soc. Am. 74(6),
2519–2533 (1984)
32. Srivastava, H.N., Ramachandran, K.: New catalogue of earthquake for peninsular India during
1839–1900. Mausam 36(3), 351–358 (1985)
33. Bapat, A., Kulkarni, R., Guha, S.: Catalogue of Earthquakes in India and Neighbourhood from
Historical Period up to 1979. Indian Society of Earthquake Technology, Roorkee (1983)
34. Gupta, H.K., Rajendran, K., Singh, H.N.: Seismicity of the North–East India region, Part I: the
database. Jour. Geol. Soc. India 28, 345–365 (1986)
35. Pendse, C.G.: Earthquakes in India and neighborhood: scientific notes. Ind. Met. Dept. 10(129),
177–220 (1949)
36. Banerji, S.K.: Earthquakes in the Himalayan Region. Indian Asoc. Cultiv. Sci., 64 (1957)
37. Tandon, A.N. and Srivastava, H.N.: Earthquake Occurrence in India. In: Arya, A.S., et al. (eds.),
Earthquake Engineering (Jai Krishna Volume), Sarita Prakashan, Meerut, pp. 1–48 (1974)
38. Srivastava, H.N., Das, S.K.: Historical seismicity and earthquake catalogues for the Indian
region. In: Lee, W.H.K., et al. (eds.) Historical Seismograms and Earthquakes of the World.
Academic Press, pp. 335–348 (1988)
39. Ramachandran, K., Srivastava, H.N.: New catalogue of felt Indian earthquakes during 1901–
1971. Mausam 42(2), 171–182 (1991)
40. Iyengar, R.N.: Earthquakes in Ancient India. Curr. Sci. 77, 827–829 (1999)
41. Bhattacharya, S.N.: A perspective of historical earthquakes in India and its neighbored up to
1900. Mausam 49, 375–382 (1998)
42. Ambraseys, N., Jackson, D.: A note on early earthquakes in northern India and southern Tibet.
Curr. Sci. India 84(4), 570–582 (2003)
43. Jaiswal, K., Sinha, R.: Web portal on earthquake disaster awareness in India (2004). www.ear
thquakeinfo.org
44. Rao, B.R.: Monograph on History of Indian earthquakes from earliest to 2005. http://seisinfo-
india.org/seismocity.html
45. NEIST.: Earthquake catalogue in and around Northeastern region of India (2013). http://www.
neist.res.in/geo_science/database/EarthquakeCatalogue_NEIST_ASDMAproject_First.pdf
46. Raghu Kanth, S.T.G.: Estimation of Seismic hazard parameters for India. Seismol. Res.
Lett. 81(2), pp. 207–217 (2010)
47. Raghu Kanth, S.T.G.: Seismicity parameters for important urban agglomerations in India. Bull.
Earthq. Eng. 9(5), 1361–1386 (2011)
48. Yadav, R.B.S., Bormann, P., Rastogi, B.K., Das, M.C., Chopra, S.: A homogeneous and
complete earthquake catalog for northeast India and the adjoining region. Seismol. Res. Lett.
80, 609–627 (2009)
49. Das, R., Menenses, C.: A unified moment magnitude earthquake catalog for Northeast India.
Geom. Nat. Hazard. Risk, 167 (2021). ISSN 1947-5705 (Print)
A Cluster-Based Seismic Risk
Assessment: Economic Loss Using GIS
for Jaipur Sub-Urban Area
V. Anand , M. Mahatab, A. Sharma, D. Raj , M. K. Jat, R. Sarkar,

and S. Pal
Abstract Economic risk is an integral part of seismic risk assessment, which is

eventually required in seismic risk mitigation, response planning, and better resource
allocation for adaptation and mitigation. Geospatial techniques like remote sensing-
based image clustering and Geographical Information systems (GIS) can be used
for seismic risk assessment at city level. In this paper, a GIS-based framework is
proposed to assess the seismic risk of a part of the city in terms of economic loss using
homogeneous clusters of different built forms, which are proxies to different building
typologies. A random sampling strategy is adopted for geo-tagged proportionate
building surveys as a function of homogenous built-form clusters and built-up density
for Jaipur Sub-urban Area. An extensive survey was conducted using GNSS receiver
to collect the geo-tagged information of different building typologies, and the GIS
database was created in real-time. Further, an analysis is carried out using HAZUS
methodology for the estimation of seismic risk in terms of economic loss spatially,
which can also help policymakers in planning and mitigation.
Keywords Geographic information system · Economic loss · Seismic hazard ·

Earthquake · Damage
1 Introduction
Earthquakes are extremely sudden occurring natural disasters which can wreak havoc
in cities [1]. The devastating effects of an earthquake may cause the city to an
V. Anand · M. Mahatab · A. Sharma · D. Raj (B) · M. K. Jat

Deptartment of Civil Engineering, MNIT Jaipur, Jaipur, Rajasthan 302017, India
e-mail: dhiraj.ce@mnit.ac.in
R. Sarkar
Deptartment of Civil Engineering, IIT ISM, Dhanbad, Jharkhand 826004, India
S. Pal
Department of Civil Engineering, DTU, New Delhi, Delhi 110042, India
https://doi.org/10.1007/978-981-99-1459-3_41
512 V. Anand et al.
instantaneous halt and disrupt the day-to-day activities related to economic, social,
recreational, etc. Past studies revealed the consequences of potential earthquakes on
various cities, where the damaging effects of destruction were a significant number
of casualties and moderate to severe damage to infrastructure including building
stocks, roads, railways, bridges, water supply systems, power distribution systems,
and telecommunication systems [2, 3]. Sometimes, the destruction from the earth-
quake was so monumental that it was regarded as an ‘Act of God’, which is inevitable
[4]. After the earthquake, the rescue or relief team could only assist the victims of the
disaster by providing medical, shelter, and food facilities and help in resuming their
day-to-day activities as soon as possible [5]. In the past 50 years, due to technological
advancement in multiple disciplines related to engineering seismology, monitoring
and collection of earthquake data, and development of new building materials and
construction techniques, it is now possible that by using the proper earthquake-
resistant method to build infrastructure, and buildings and technical tools to predict
expected earthquakes at any site, the damages and casualties can significantly be
reduced [4, 6, 7]. However, various newly developed techniques are still not applied
in practice due to a lack of the necessary financial or human resources and also a
lack of awareness at the managerial or bureaucratic level [8–14].
In past research, it has been found that the increase in seismic vulnerability of
building stock, infrastructure, etc., and the associated risk was a major concern
in metropolitan areas. Due to lacuna in the efforts for seismic disaster reduction
procedure implementation and immediately available budget, scenario generation is
essential to find out what will happen if an earthquake hits a particular city. Seismic
damage estimation can work as an initial attempt for an effective Seismic risk reduc-
tion program [15]. It is well known that these procedures call for an assessment of the
potential damages due to seismic events before making suggestions for mitigation,
preparation, and response. Risk assessment is the main component of the estima-
tion of probable disaster damages, among other things. As far as decision-making
and emergency management purposes are concerned, seismic risk can be character-
ized in terms of possible economic, social, and environmental losses from a specific
earthquake occurrence [16, 17].
The first substantial attempts to develop a seismic risk index were made in the
United States of America (USA), which can compare the relative risk of several
cities but not the risk of the urban fabric inside a metropolis. In addition to phys-
ical danger, the impact of social fragility and societal resilience were also taken into
account while defining overall risk [18]. The disaster risk index has been defined glob-
ally using a variety of approaches that incorporate hazard and vulnerability analyses
at various levels. Federal Emergency Management Agency (FEMA) has included
the socio-economic factors related to urban earthquake risk in the development of
HAZUS software [19]. The HAZUS technique is sophisticated for urban seismic
risk assessment [20]. Generally, the output of the majority of the research conducted
in Europe was in the form of software packages for estimating seismic risk and
estimating earthquake losses as monetary and social damages [21, 22]. Meanwhile,
there are several projects happening to provide approaches for estimating earthquake
A Cluster-Based Seismic Risk Assessment: Economic Loss Using GIS … 513
losses and assessing earthquake risk. Further, the fragility function-based vulnera-
bility assessment method was incorporated in the HAZUS methodology, which was
a probabilistic and more physically sound and exhaustive method as compared to
qualitative methods [23–25]. The fragility functions give the probability of reaching
or exceeding the particular damage states (like No Damage, Slight, Moderate, Exten-
sive and Complete Damage) subjected to a particular level of seismic hazard defined
in terms of Peak Ground Acceleration (PGA), Peak Ground Velocity (PGV), Spectral
Acceleration (S a ), average spectral acceleration (S avg ) or Spectral Displacement (S d )
[26–28].
To implement the HAZUS methodology in the Indian scenario, an intensive liter-
ature review has been conducted to collect the fragility function for different building
typologies existing in India [29]. These past studies have developed the fragility func-
tions for a particular building typology, such as masonry buildings [29], RC Frame
buildings [30, 31], RC buildings with infill walls [32] and steel buildings [33, 34].
Using these fragility and vulnerability functions, the economic loss for a particular
building can be estimated. However, to estimate the loss at the city level, several
authors have used the GIS framework in the past [19, 22, 35–37]. In this study, an
attempt has been made to estimate the economic loss due to earthquakes considering
the seismic hazard and vulnerability of building stock present in the Jaipur Sub-urban
Area, using HAZUS methodology within the GIS framework.
2 Study Area
Jaipur, also known as the pink city, is the capital city of the largest state Rajasthan (by
area), located at the latitude of 26° 55, 19.4520,, N and a longitude of 75° 46, 43,, E
having an area of 472 km2 [38]. The pink city enjoys heavy footfall of tourists
throughout the year due to the fascinating monuments, markets offering culture and
heritage-rich products, and wonderfully laid-out gardens [39]. It also makes the
pink-walled city of Jaipur a UNESCO world heritage site.
The present study focuses on the different types of clusters (formed by similar
building typologies) surrounding the Malviya National Institute of Technology,
Jaipur. Four different types of clusters have been identified, namely Malviya Nagar
Commercial, Malviya Nagar Residential, Malviya Nagar Industrial Area, and Jhalana
Gram. These four clusters comprise residential buildings, commercial buildings, and
slums. The satellite image for this particular region has been used for the identification
of different clusters. To capture the structural and architectural features of buildings
present in the identified clusters, an intensive field survey has been conducted. For the
field survey, the representative buildings have been selected using stratified random
sampling to cover all the existing typologies. Global Navigation Satellite System
(GNSS) data logger has been used to store the building information in the form of
attributes which can be directly imported to the GIS platform.
514 V. Anand et al.
3 Methodology
In this paper, the HAZUS methodology has been modified based on the fragility func-
tions available for different building typologies present in Indian cities and adopted
for Indian scenarios using GIS. Further, the economic loss has been estimated based
on these vulnerability functions for the considered study area, as explained in the
following section.
3.1 Study of City Urbanization Process Through Identifying

the Homogenous Built-Form Regions Using Satellite
Image
Segregation of different homogenous built form-based clusters provides an aid in the

prior planning of field survey and a further improvement in the identified clusters
of the entire built-up area. These identified clusters have been used further to utilize
for the selection of representative buildings of different typologies using stratified
random sampling and also to estimate the cluster-wise economic loss. In order to
form the clusters, different variables (e.g., texture, association, pattern, etc.) have
been identified as the proxy for various built forms. Using these proxy variables, the
study area has been categorized into different clusters, as shown in Fig. 1.
3.2 Field Survey
The identification of building typology and its architectural and structural features
are required for the selection of appropriate vulnerability functions used for the
estimation of seismic risk of a particular building and further extended to the city
level. GNSS-based survey automates the process of data collection, transfer of data
to the GIS platform, and further processing along with the photographs. This process
also reduces the possibility of any error which may occur during manual data entry.
In this study, the survey forms used for data collection are in line with NDMA having
some additional fields. As GNSS-based survey (shown in Fig. 2) was used, the entire
collected building inventory data was geo-tagged, including the spatial locations with
other attributes in digital format. Figure 2 shows a sample data collection screen in
GNSS datalogger Trimble Juno T41/5. All the fields are filled in the GNSS datalogger,
which generates the building information along with their spatial coordinates as ESRI
point Shapefile format.
Fig. 1 a Satellite image from Sentinel 2B platform of Jaipur City (sensed on 22-04-2022) and b
segmented satellite image
Fig. 2 a GNSS receiver Trimble Juno T41/5, b sample screenshots during the field survey data
collection through Juno
516 V. Anand et al.
Fig. 3 A sample of the GIS-based dataset prepared through the GNSS-based data collected from
the field survey
3.3 Creation of GIS-Based Dataset
The collected field survey data have been exported as ESRI shapefile format, and
the corresponding attributes are shown in Fig. 3. The assessment of economic loss
due to seismic hazard has been performed for four different visually homogeneous-
looking clusters based on building typologies present in the study area, as shown
in Fig. 4. These clusters have been classified using a satellite image of sentinel
2B with object-based image analysis. Cluster A, i.e., Malviya Nagar, is dominated
by commercial areas adjacent to JLN Marg and residential buildings in a planned
manner usually comprised of RC-framed structures. Cluster B is also similar in
nature, with a relatively higher number of residential buildings as compared to cluster
A. Cluster C is the industrial area of Malviya Nagar, which comprises relatively larger
size buildings to incorporate industries. Steel-framed and RC-frame structures are
the dominant building typologies in Cluster C. Cluster D, i.e., Jhalana Gram, is an
unplanned settlement which has a drastic difference in terms of building typologies
as compared to all the other clusters. It comprises Stonemasonry and Brick Masonry,
with only a few houses having RC-frame structures. Figure 4 also shows the cluster
along with the sample data point used for the study area. Further, the GIS-based
dataset has been prepared directly through GNSS-based field survey data and used
for the further processing of vulnerability and exposure estimation.
3.4 Hazard
In the present study, the seismic hazard (expressed in terms of PGA) has been adopted
from the NDMA Probabilistic Seismic Hazard Map of India database. This gridded
data contains the PGA values with a 2% probability of exceedance in 50 years (Return
Fig. 4 GIS map showing the clusters and points surveyed in the study area
Period ~ 2500 years) on A-type Sites. Further, the PGA value has been modified to
include the effect of overlaying soil presented at the site. The gridded data of PGA
has been converted into a raster map using geostatistical interpolation with a spatial
resolution of 500 m. The hazard map generated in terms of PGA data on each cell
has been used to estimate the spatial distribution of risk within the study area after
the inclusion of exposure and vulnerability.
3.5 Vulnerability Estimation
Vulnerability describes the degree of damageability of exposed entities of a city when

subjected to a specific level of hazard. The vulnerability has been estimated from the
fragility functions defined for the different building typologies. The fragility function
for the building typologies present in the study area has been taken from past studies.
518 V. Anand et al.
Table 1 Cost of
Sr. no. Building typology Cost per square meter
repair/replacement rates
adopted in this study 1 RC building Rs. 13550
2 Brick masonry Rs. 9540
3 Stone masonry Rs. 8500
3.6 Economic Loss Estimation
Now, the economic loss has been calculated using the following relationship:
Economic Loss = (PN × RCSN + PS × RCSS + PM

× RCSM + PE × RCSE + PC × RCSM ) × Cri × Ns
where the structure repair cost ratio for no damage state is RCSN , slight damage
state is RCSS , moderate damage state is RCSM , extensive damage state is RCSE , and
complete damage state is RCSC ; C ri is the location-based cost of repair/replacement
per unit area, and N s is the number of stories. The location-based cost of
repair/replacement per unit area has been taken from standing order No. X-3/2015
Public Work Department, Govt. of Rajasthan. The cost of repair/replacement rates
adopted in this study are shown in Table 1.
4 Result and Discussion
Through a selection of the adequate number of points for the field survey using
GNSS data logger, a GIS database containing building structural and architectural
features has been generated. Using this GIS database, all the required inputs for the
estimation of economic loss have been calculated at each surveyed building loca-
tion and then interpolated for the whole cluster. The probabilities of slight damage,
moderate damage, extensive damage, and complete damage according to building
typologies and at a given hazard level have been calculated and interpolated for the
whole region. In the same manner, the repair cost ratio for slight damage, moderate
damage, extensive damage, and complete damage according to occupancy class has
been calculated and further converted into a raster layer. Further, the spatial inputs for
economic loss have been prepared. After the generation of input layers, as discussed
above, the economic loss has been fed into the raster calculator in ArcGIS software
to delineate the Economic loss Map of the study region. The economic loss map of
the study is depicted in Fig. 5.
The economic loss map generated through this study highlights the regions with
improper planning and building practices with relatively higher value of economic
loss in those regions. In our current study, Jhalana Gram is showing the highest value
Fig. 5 Economic loss map of the study area
of economic loss due to improper building practices when subjected to considered

seismic hazard levels.
5 Conclusion
In this study, HAZUS methodology based on the GIS framework has been used to
assess seismic risk for Jaipur Sub-urban Area, in terms of economic loss. Various
homogeneous clusters of similar built forms within the study area have been identified
based on remotely sensed images. For the number of geo-tagged buildings to be
surveyed within the identified clusters, a random sampling strategy has been adopted
as a function of the homogenous built form and built-up density. An extensive survey
has been conducted using GNSS receivers to collect the geo-tagged information
of different building typologies and to create the GIS database. Further, spatially
distributed seismic risk in terms of economic loss has been estimated using the
HAZUS methodology in the GIS framework.
520 V. Anand et al.
The outcomes of this study provide a useful framework for quickly finding out
the economic loss for every location of a city on a cell basis, which can further be
updated according to the different levels of seismic hazard. The spatial distribution
of economic loss can also help policymakers in planning and mitigation through
funds allocation optimally using the micro-zonation city map based on the severity
of the seismic risk. Thus, this methodology can be used for the whole city for the
identification of seismically prone areas having more amount of economic losses,
and by following the guidelines given in standards/codes, this probable economic
loss can be reduced.
Acknowledgements The work presented in this article is funded by the National Disaster Manage-
ment Authority, Government of India. The support received from the funding agency and the
institutes are gratefully acknowledged.
References
1. Torrence, R., Grattan J.: The archaeology of disasters: past and future trends. In: Natural
Disasters and Cultural Change. Routledge, pp. 17–34 (2003)
2. Kates, R.W.: Experiencing the environment as hazard. In: Experiencing the Environment,
pp. 133–156 (1976)
3. Helsloot, I., Ruitenberg, A.: Citizen response to disasters: a survey of literature and some
practical implications. J. Contingencies Cris. Manage. 12(3), 98–111 (2004)
4. d’Albe, E.M.F.: An approach to earthquake risk management. Eng. Struct. 4(3), 147–152 (1982)
5. Wijkman, A., Timberlake, L.: Natural Disasters: Acts of God or Acts of Man? (2021)
6. Coburn, A., Spence, R.: Earthquake protection. John Wiley and Sons, Chichester, UK (2003)
7. Kenny, C.: Disaster risk reduction in developing countries: costs, benefits and institutions.
Disasters 36(4), 559–588 (2012)
8. Hendriks, E., Opdyke, A.: Adoption of seismic-resistant techniques in reconstructed housing
in the aftermath of Nepal’s 2015 Gorkha earthquake. Earthq. Spectra 37(4), 2662–2686 (2021)
9. Mahin, S.A.: Lessons from damage to steel buildings during the Northridge earthquake. Eng.
Struct. 20(4–6), 261–270 (1998)
10. Xie, L.-L.: Some challenges to earthquake engineering in a new century. In: Earthquake
Engineering Frontiers in the New Millennium, pp. 7–13 (2017)
11. Yates, D., Paquette, S.: Emergency knowledge management and social media technologies: a
case study of the 2010 Haitian earthquake. Int. J. Inf. Manage. 31(1), 6–13 (2011)
12. Kramer, S.L.: Geotechnical Earthquake Engineering. Prentice Hall, Upper Saddle River, New
Jersey, 1996, United States of America (1996)
13. Voigt, S., Giulio-Tonolo, F., Lyons, J., Kučera, J., Jones, B., Schneiderhan, T., Platzeck, G.,
Kaku, K., Hazarika, M.K., Czaran, L.: Global trends in satellite-based emergency mapping.
Science 353(6296), 247–252 (2016)
14. Sorensen, J.H.: Hazard warning systems: review of 20 years of progress. Nat. Hazard. Rev.
1(2), 119–125 (2000)
15. Alam, N., Alam, M.S., Tesfamariam, S.: Buildings’ seismic vulnerability assessment methods:
a comparative study. Nat. Hazard. 62(2), 405–424 (2012)
16. Carreño, M.L., Cardona, O.D., Barbat, A.H.: New methodology for urban seismic risk
assessment from a holistic perspective. Bull. Earthq. Eng. 10(2), 547–565 (2012)
17. Tesfamariam S., Eeri M., Saatcioglu M.: Risk-Based Seismic Evaluation of Reinforced
Concrete Buildings. Earthquake Spectra—EARTHQ SPECTRA 24 (2008)
18. Davidson, R.: EERI annual student paper award a multidisciplinary urban earthquake disaster
risk index. Earthq. Spectra 13(2), 211–223 (1997)
19. Alam, M.N., Tesfamariam, S., Alam, M.S.: GIS-based seismic damage estimation: case study
for the City of Kelowna. BC. 14(1), 66–78 (2013)
20. Risk assessment and GIS in natural hazards: issues in the application of HAZUS. Int. J. Risk
Assess. Manage. 6(4–6), 408–422 (2006)
21. Anagnostopoulos, S., Providakis, C., Salvaneschi, P., Athanasopoulos, G., Bonacina, G.: SEIS-
MOCARE: an efficient GIS tool for scenario-type investigations of seismic risk of existing
cities. Soil Dyn. Earthq. Eng. 28(2), 73–84 (2008)
22. Whitman, R.V., Anagnos, T., Kircher, C.A., Lagorio, H.J., Lawson, R.S., Schneider, P.: Devel-
opment of a national earthquake loss estimation methodology. Earthq. Spectra 13(4), 643–661
(1997)
23. Shibin, L., Lili, X., Maosheng, G., Ming, L.: Performance-based methodology for assessing
seismic vulnerability and capacity of buildings. Earthq. Eng. Eng. Vib. 9(2), 157–165 (2010)
24. Nofal, O.M., van de Lindt, J.W., Do, T.Q.: Multi-variate and single-variable flood fragility and
loss approaches for buildings. Reliab. Eng. Syst. Safety 202, 106971 (2020)
25. Patel, N.K., Vasanwala, S.A.: Influence of geometric variation on seismic fragility assessment.
In: Sustainable Building Materials and Construction, pp. 233–240 (2022)
26. Porter, K.A.: An overview of PEER’s performance-based earthquake engineering methodology.
In: Proceedings of 9th International Conference on Applications of Statistics and Probability
in Civil Engineering, pp. 1–8 (2003)
27. Pang, Y., Zhou, X., He, W., Zhong, J., Hui, O.: Uniform design-based Gaussian process regres-
sion for data-driven rapid fragility assessment of bridges. J. Struct. Eng. 147(4), 04021008
(2021)
28. Tan, Z., Zhao, Q., Zhao, Y., Yu, C.: Probabilistic seismic assessment of CoSPSW structures
using fragility functions. Metals 12(6), 1045 (2022)
29. Halder, L., Paul, S.: Seismic damage evaluation of gravity load designed low rise RC building
using non-linear static method. Proc. Eng. 144, 1373–1380 (2016)
30. Kumar, P., Samanta, A.: Seismic fragility assessment of existing reinforced concrete buildings
in Patna, India. Structures 27, 54–69 (2020)
31. Apte, T., Maru, S.: Fragility analysis of regular multi-storey RC building. IJSRSET 4(4),
891–895 (2018)
32. Agarwal, P., Navya, G.: Seismic retrofitting of structures by steel bracings. Dept. Earthq. Eng.
Indian Inst. Technol. Roorkee 144, 1364–1372 (2015)
33. Kumar, R., Sahoo, D.R.: Seismic fragility of steel special truss moment frames with multiple
ductile vierendeel panels. Soil Dyn. Earthq. Eng. 143, 106603 (2021)
34. Sharma, V., Shrimali, M.K., Bharti, S.D., Datta, T.K.J.S.: Behavior of semi-rigid steel frames
under near-and far-field earthquakes. Compos. Struct. Int. J. 34(5), 625–641 (2020)
35. Zhang, S., Yang, K., Cao, Y.: GIS-based rapid disaster loss assessment for earthquakes. IEEE
Access 7, 6129–6139 (2019)
36. Karimzadeh, S., Miyajima, M., Hassanzadeh, R., Amiraslanzadeh, R., Kamel, B.: A GIS-based
seismic hazard, building vulnerability and human loss assessment for the earthquake scenario
in Tabriz. Soil Dyn. Earthq. Eng. 66, 263–280 (2014)
37. Zhang, Y., Zheng, S., Sun, L., Long, L., Yang, W., Li, L.: Developing GIS-based earthquake
loss model: a case study of Baqiao District, China. Bull Earthq. Eng. 19(5), 2045–2079 (2021)
38. Chandra, S., Dubey, S.K., Sharma, D., Mitra, B.K., Dasgupta, R.: Investigation of spatio-
temporal changes in land use and heat stress indices over Jaipur City Using Geospatial
Techniques. Sustainability 14(15), 9095 (2022)
39. Maitra, R., Bansal, T., Joseph, A.I.: Dimensions of culinary tourism and hospitality in golden
triangle of India. Studia Universitatis Babes-Bolyai, Negotia 66(3), (2021)
Modelling of Empirical Accelerograms
of 1999 Chamoli Earthquake (Himalaya)
Using a Modified Hybrid Approach
A. Sharma, D. Kumar, and A. Paul
Abstract In the present study, the empirical accelerograms of the 1999 Chamoli
earthquake (M s 6.6) have been modelled using a modified hybrid approach. The
earthquake occurred in the Central Seismic Region of the Himalayan region and
was recorded at the seismic network installed and maintained by the Department
of Earthquake Engineering, I.I.T.—Roorkee. The modified hybrid technique [1]
has been used to model the empirical accelerograms of the 1999 Chamoli earth-
quake at 9 recording sites. The epicentral distance of various observing stations
lies in a range of 10–110 km. The empirical accelerograms of the earthquakes are
available at the sampling rate of 50 Hz. The hybrid technique includes the gener-
ation of envelope functions by summing the envelope functions of the randomly
distributed subevents on the fault plane. In the present modified hybrid technique,
the site response high-frequency decay parameter, i.e. kappa factor (κ), has also been
evaluated and incorporated into the existing technique. The site response functions
have been estimated using the HVSR technique. The high-frequency decay param-
eter ‘κ’ has been estimated to be in the range of 0.04–0.07. The simulated accelero-
grams have been compared with those of recorded ones in terms of Peak Ground
Acceleration (PGA), duration, Response and Fourier spectra. The modelled value of
PGA is 347 cm/s2 estimated at Gopeshwar is found to be close to the observed one
(352 cm/s2 ). The essential parameters of simulated accelerograms, including PGA
values, duration, Response and Fourier spectra, are well matched with those of the
recorded accelerograms for most sites.
Keywords Seismic Hazard · Simulation · Earthquake · Kappa · Site response

function
A. Sharma (B) · D. Kumar

e-mail: anjali.11587@gmail.com
A. Paul
Wadia Institute of Himalayan Geology, Dehradun, India
A. Sharma
Ministry of Earth Sciences, Borehole Geophysics Research Laboratory, Karad 415105, India
https://doi.org/10.1007/978-981-99-1459-3_42
524 A. Sharma et al.
1 Introduction
50 Mya ago, the collision of the Indian-Eurasian plates led to the rise of the
Himalayas. In the last two centuries, more than eight earthquakes have occurred
in the Himalayas and no earthquakes of magnitude greater than or equal to 8 have hit
since 1950 [2, 3], although a large part of the destructive boundary remains undis-
turbed. Based on the regional earthquake pattern, three earthquake zones near the belt
of the Himalayas have been described by Khattri and Tyagi [4] and Khattri [5]. From
the 1905 Kangra to the 1934 Bihar earthquake, the Central seismic gap (CSG) is the
longest stretch. They have proposed that CSG may expect next future earthquakes.
The Kumaun-Garhwal region of Western Himalaya was hit by an earthquake of
magnitude (6.6) on 29 March 1999 at 00:35 h IST. The epicentre location and focal
depth reported were 79.416 °E, 30.408 °N and 21 km, respectively. Earthquake-
affected region is not the same topographically. Population growth is found to be
quite low in the area. The quake shook the twin cities of Gopeshwar and Chamoli
along with the neighbouring region of Rudraprayag. The maximum affected districts
are mainly the river valley and Alaknanda. The affected region is in the V seismic
zone of India, which is the highest risk zone and needs to establish the earthquake-
resistant structures. The earthquake studies give us the chance to learn about the
nature and the earthquake processes.
Many researchers have studied 1999, Chamoli earthquake such as Shrikhande et al.
[6], Kumar et al. [7], Joshi [8], and Chopra et al. [9] for various research purposes.
In the present study, the empirical accelerograms of 1999, Chamoli earthquake have
been simulated using the modified Hybrid technique. In the present modified tech-
nique, the estimated site response and the high-frequency decay parameter, i.e. kappa
(κ) factor, have been incorporated into the already existing technique.
2 Seismotectonics
The Kumaun-Garhwal section is situated in the central seismic gap (CSG), which
is a rupture zone between the historic 1905, Kangra and the famous 1934, Bihar-
Nepal earthquakes. The continuous movement of the Indian plate is moving towards
the Tibetan segment of the Eurasian plate with a convergence rate of 0.05–0.06 m
annually (Molnar and Tapponnir [10]. This movement not only distorts the rocks
to developing the higher Himalayas but making the Garhwal region of Himalaya
seismically active.
In Garhwal and the adjacent province of the Himalayas, the three most important
tectonic divisions are parted from one another by Main Central Thrust (MCT) and
Main Boundary Thrust (MBT). Molnar [11] has given various models regarding
the Himalaya’s evolution and the effect of the MBT and MCT in these mountain
buildings. Most of the earthquakes in this region are situated near the north and south
of MCT. The tertiary belt of sub-Himalayas is overthrust by the Proterozoic-Eocene
Modelling of Empirical Accelerograms of 1999 Chamoli Earthquake … 525
Fig. 1 The location of the 1999 Chamoli earthquakes along with the recording sites of an installed
seismic network
sequence of the lesser Himalayas. In the southern direction of it, the MFT, i.e. Main
Frontal Thrust, separates the above belt from the Indo-Gangetic plains (IGP). The
metamorphosed sediments (low grade) of lesser Himalayas are overridden sharply
by high to medium-grade gneiss, migmatite and schist which belong to crystalline
Higher Himalayas [12]. The location of the 1999 Chamoli earthquakes along with
the recording sites of an installed seismic network has been shown in Fig. 1.
3 Methodology
The Renowned hybrid technique suggested by [13] is an addition of two famous

methods of simulation, i.e. composite source model (CSM) [14, 15], and envelope
function (EF) [16, 17] methods. Firstly, the EF of the main event is estimated by the
addition of EF’s corresponding to small size events. Small-size events or subevents
are distributed non-uniformly in terms of size and their location on the fault plane.
Multiplication of obtained EF for the main event with the band-limited white noise
provides the final simulated accelerograms. Yadav [1] has modified the existing
hybrid technique by introducing site amplification functions and κ factor.
The EF, eij (t), calculated for individual (i, j)th small event on fault using Eq. (1)
[18]:
ei j (t) = (ai j t/di j ) ∗ exp((1 − t/di j )) (1)
where a and d are the PGA value and the duration parameter of the EF. The EF of
the main event e(t) is estimated by Eq. (2) [17]:
⎡ ⎤1/
2
ΣΣ
e(t) = ⎣ ei j t − ti j
2 ⎦ (2)
i j
where tij is the delay in time caused due to travel and the rupture produced by the
seismic waves.
The number of randomly distributed subevents has been evaluated using Eq. (3)
suggested by Zeng et al. [15]
−D
−D /
N (R) = P R − Rmax D (3)
where D, N and Rmax define the fractal dimension, number of sub/small events having
radii R, and largest sub-event’s radius, respectively, and parameter ‘p’ is calculated
using Eqs. (4) and (5) by Zeng et al. [15]:
/
p = 7M O (3 − D) [16Δσ (Rmax − Rmin )] when D /= 3
E
(4)
E/
p = 7M O [16Δσ ln(Rmax − Rmin )] when D = 3 (5)
The sub-events size is estimated using Eq. (6):
/ −D
−1/D
Ri = D Ni p + Rmax (6)
where N i and Ri is the ith random number and ith subevent’s radius. Equation (7)
helped in realizing the actual seismic moment by adjusting stress drop [15]:
Σ / Σ
M OR = M O = 16 7 Δσ Ri3 (7)
Numerous fault models may be possible for the random distribution of these
random-size subevents. To choose the best model for simulating the accelerograms
renowned, the optimization technique, i.e. Genetic algorithm (GA) technique, has
been used. This algorithm is based on the basic fittest existence standards.
It begins with a generation of probable models. The artificial data produced from
these probable models are then compared with the recorded ones, and a fitness score
is estimated for every model of generation. The three operators of genetic algorithm
(GA) are selection, crossover and mutation. These three of them are applied to the
current generation to generate the models of the next generation. This procedure
repeats it again till a constant fitness score is obtained or fitness score reaches a
maximum saturation level.
In the present study, we have used the PGA-attenuation relation [Eq. (8)] suggested
by Abrahamson and Litehiser [19]:

log(PGA) = −0.62 + 0.177 MS − 0.982 log R + e0.248MS
+ 0.132F−0.0008 ER (8)
In this expression, PGA (in g), M s is surface (S) wave magnitude and R denotes
the hypocentral distance (in km). F values are either 1 (reverse fault) or 0 otherwise.
The value of E is 0 or 1 for intraplate and interplate events, respectively. Midorikawa
[17] has established relation [Eq. (9)] below to estimate the parameter of duration:
d = 0.015 × 100.05m + 0.12R 0.75 (5 ≤ m ≥ 6) (9)
where
d = duration parameter
m = earthquake magnitude.
The site amplification functions and kappa factor have been calculated using the
famous horizontal to vertical ration method (HVSR) using Eq. (10):
/
abs H ( f )2ew + abs H ( f )2ns /2
H/ = (10)
V abs V ( f )z
Kappa factor values have been estimated using the method suggested by Anderson
and Hough [20]. According to this method, the decay of spectra A(f ), can be written
as Eq. (11):
A( f ) = AO e−πκ f ; f > f E (11)
A0 depends on the distance and source, etc., f is frequency and f E is that frequency
below which the shape of spectral shape is not linear on log-lin scale as Eq. (12):
log[A( f )] = log(AO ) − π κ f. log(e) (12)
This indicates that κ can be calculated with a linear least-square fit to observed
log-lin spectra. The parameter kappa ‘κ’ is estimated using Eq. (13):
Fig. 2 Step-by-step methodology of modified hybrid technique used in this study
/
κ = φ π. log(e) (13)
φ is the slope of the linear fit. Figure 2 shows the step-by-step methodology of
Modified Hybrid technique used in this study.
Earthquakes have proven themselves very damaging for society but they also give the
wealth of information that helps us to understand the earthquake process better. This
understanding eventually helps us to better understanding of earthquake phenomenon
and helps to diminish the hazard caused by any earthquake in a particular area. In
the current study, the conformity of a modified hybrid method has been presented
to model the empirical accelerograms of 1999 Chamoli (M 6.6) earthquake. The
earthquake has been observed at strong motion array equipped with three component
SMA-I instruments. The recording stations of 1999 Chamoli earthquake are located
in epicentral distance range is 10–110 km. The digital empirical accelerograms of the
earthquake are available at the sampling rate of 50 Hz. The earthquake parameters
are given in Table 1.
Recorded accelerograms of 1999 Chamoli earthquake at 9 stations have been
modelled with the help of modified hybrid technique. The site response functions
have been estimated using the famous horizontal to vertical ration method for each
Table 1 Parameters used for modelling of empirical accelerograms of 1999 Chamoli earthquake
Sr. no. Parameter Value References
1 Fault orientation Strike: 280° Dip: 7° CMT
2 Focal depth (km) 15 CMT
3 Epicentre 30.38 N, 79.21 E CMT
4 Seismic moment (dyne-cm) 7.8 × 1025 CMT
5 Fault dimension (km) 25 × 12 Wells and Coopersmith
[21]
6 No. of subevents 11 This study
7 κ value 0.03–0.07 This study
8 Fractal dimension 2
9 Shear wave velocity (km/s) 3.0
10 Rupture velocity (km/s) 2.7
11 Dynamic stress drop (bars) 100
12 Q-relation 126f 0.95 Gupta et al. [22]
13 Duration parameter d = 0.015 × 100.05 M ‘ + Modified by Kumar et al.
0.12R0.75 [23]
14 PGA-attenuation relation Log(PGA) = −0.62 + Abrahamson and Litehiser
0.177Ms −0.982log(R + [19]
exp(0.248Ms)) + 0
132 F −0.0008 ER
station. The smooth site response curve has been used in the study. The least-square
fitted line for the log-lin curves of Fourier spectra helped us to estimate the kappa
factor. The kappa range estimated for this earthquake shows a range of kappa value
from 0.04 to 0.07. Total 11 number of subevents having different magnitudes have
been found on the fault plane. Their best location on the same has been decided using
the renowned genetic algorithm. In this event, 30 generations of 200 models have
been created and best model of the 30th generation has been chosen for further study.
Table 2 shows the simulated and the observed PGA at various stations. The PGA
values of modelled waveforms are found in good agreement with those of recorded
PGA values for the 6 stations Almora, Barkot, Ghansali, Gopeshwar, Tehri and
Uttarkashi. For rest of the 3 stations—Chinyalisaur, Joshimath and Ukhimath—the
matching is not good. The radiation pattern or/and non-linear site effects along with
the directivity effect may be the cause of such difference. The decay rate of both
simulated and observed PGA values are found similar.
Figure 3a–d depicts the comparison of simulated and recorded accelerograms
with their corresponding Fourier and Response spectra at Gopeshwar, Ukhimath,
Tehri and Ghansali recording stations. We have observed that the duration of simu-
lated accelerograms is matching quite well with those of recorded ones at most of
the stations. The matching of duration is satisfactory at Joshimath. The duration
of simulated accelerograms has found longer as compared to recorded ones at the
stations Almora and Uttarkashi.
530
Table 2 Comparison of observed and synthetic PGA values for the 1999, Chamoli earthquake
Sr. no. Station name Station code Latitude Longitude Epicentral Hypocentral PGA-L-Comp PGA-T-Comp PGA-simulated
distance distance (km) (cm/s2 ) (cm/s2 ) (cm/s2 )
1 Gopeshwar GOPE 30.4 79.33 11.7024 19.0249 195 353 347
2 Ukhimath UKHI 30.5 79.1 16.9776 22.6548 89 95 141
3 Joshimath JOSH 30.55 79.56 38.4371 41.2603 66 62 107
4 Ghansali GHAN 30.4 78.65 53.6652 55.7221 71 82 82
5 Tehri TEHR 30.3 78.58 61.0023 62.8194 53 61 63
6 Uttarkashi UTTA 30.73 78.45 82.3814 83.7359 52 62 67
7 Chinyalisaur CHIN 30.55 78.33 86.2826 87.5768 51 44 28
8 Almora ALMO 29.58 79.65 98.3621 99.4993 26 27 21
9 Barkot BARK 30.8 78.2 107.1775 108.2221 17 22 14
A. Sharma et al.
Fig. 3 Simulated accelerograms, observed accelerograms at various stations of 1999, Chamoli

earthquake along with Fourier and Response spectra
Fig. 3 (continued)
The Fourier spectra of simulated accelerograms are matching well with that of
recorded ones at Chinyalisaur, Gopeshwar and Uttarkashi. For the rest 6 stations,
the matching is in well agreement for higher frequencies as compared to the low
frequencies.
Fig. 4 Hypocentral distance versus PGA graph of observed and simulated accelerograms of 1999,
Chamoli earthquake
The matching of Response spectra of recorded accelerogram with that of simulated

spectra is well matched for the Uttarkashi station. At Almora station, the matching
is good at lower and intermediate periods, and for station Barkot, it is good for the
periods 0.3–0.5 s. The response spectrum of simulated accelerogram is good up to
0.9 s for the station Tehri. For Gopeshwar, it is good for lower and longer periods.
The matching of Response spectra is satisfactory for the stations Ghansali for lower
periods. The Response spectra of observed accelerogram is weak as compared to that
of simulated one at the station Ukhimath. Figure 4 shows the simulated and observed
PGA value versus hypocentral distance graph for different stations.
5 Conclusion
The fidelity of the modified hybrid technique has been demonstrated successfully
for the modelling of empirical accelerograms of 1999 Chamoli earthquake. Both
frequency and time domain parameters of modelled accelerograms are found to be
well matched with those of the recorded accelerograms for most of the sites. The lower
frequency contents of modelled accelerograms are overestimated at some sites. This
may be due to the limitation of the modified hybrid technique in its present form
as envelop waveform of the large/target earthquake (Eq. 2) is only valid for high
frequencies. The procedure of technique involves the product of envelope function
with white Gaussian noise in time domain which has same effect as convolution
in frequency domain. Overall the modified hybrid technique has been found to be
successful for the modelling of empirical accelerograms of 1999 Chamoli earthquake.
This technique can be further used to find out the seismic Hazard in the region also.
References
1. Yadav, R: Modelling of Empirical Accelerograms of 2011 Sikkim Earthquake (M 6.9), Kappa

(κ)- Model For NE Region (India) and Scenerio Seismic Hazard Maps for Sikkim Region of
Himalaya. Ph.D. Thesis, Kurukshetra University, Kurukshetra, Haryana, India (2019)
2. Bilham, R.: Location and magnitude of the 1833 Nepal earthquake and its relation to the rupture
zones of contiguous great Himalayan earthquakes. Curr. Sci. 69, 101–128 (1995)
3. Khattri, K. N: An evaluation of earthquakes hazard and risk in northern India, Himalayan Geol.
20, 1–46 (1999)
4. Khattri, K.N., Tyagi, A.K.: Seismicity patterns in the Himalayan plate boundary and
identification of the areas of high seismic potential. Tectonophysics 96, 281–297 (1983)
5. Khattri, K.N.: Great earthquakes, seismicity gaps, and potential for earthquake disaster along
the Himalaya plate boundary. Tectonophysics 138, 79–92 (1987)
6. Shrikhande, M., Rai, D.C., Naryan, J., Das, J.: The March 29, 1999 earthquake at Chamoli,
India. In: Proceedings of the Twelfth World Conference on Earthquake Engineering, Upper
Hutt, NZ: New Zealand Society for Earthquake Engineering, p. 8 (2000)
7. Kumar, D., Ram, V.S., Khattri, K.N.: A study of source parameters, site amplification functions
and average effective shear wave quality factor Q seff from analysis of accelerograms of the
1999 Chamoli earthquake, Himalaya. Pure Appl. Geophys. 163(7), 1369–1398 (2006)
8. Joshi, A.: Analysis of strong motion data of the Uttarkashi earthquake of 20th October 1991
and the Chamoli earthquake of 28th March 1999 for determining the mid crustal Q value and
source parameters. ISET J. Earthq. Technol. 468(43), 1–2 (2006)
9. Chopra, S., Kumar, D., Rastogi, B.K., Choudhury, P., Yadav, R.B.S.: Estimation of site
amplification functions in Gujarat region, India. Nat. Hazard. 65(2), 1135–1155 (2013)
10. Molnar, P., Tapponnier, P.: Cenozoic tectonics of Asia: effect of a continental collision. Science
489, 419–426 (1975)
11. Molnar, P.: The structure of the mountain ranges. Sci. Am. 254, 70–79 (1986)
12. Valdiya, K.S.: Tectonics and evolution of the central sector of the Himalaya. Philosophical
transactions of the Royal Society of London. Ser. A Math. Phys. Sci. 326(1589), 151–175
(1988)
13. Kumar, D., Teotia, S.S., Sriram, V.: Modelling of strong ground motions from 1991 Uttarkashi,
India, earthquake using a hybrid technique. Pure Appl. Geophys. 168, 1621–1643 (2011).
https://doi.org/10.1007/s00024-010-0236-4
14. Frankel, A.: High-frequency spectral fall off earthquakes, fractal dimension of complex rupture,
b-value, and the scaling of strength on faults. J. Geophy. Res. 96, 6291–6302 (1991)
15. Zeng, Y., Anderson, J.G., Yu, G.: A composite source model for computing realistic synthetic
strong ground motions. Geophys. Res. Lett. 21, 725–728 (1994)
16. Irikura, K.: Prediction of strong acceleration motion using empirical green’s function. In:
Proceedings of the 7th Japan Earthquake Engineering Symposium, vol. 151, pp. 151–156
(1986)
17. Midorikawa, S.: Semi empirical estimation of peak ground acceleration from large earthquakes.
Tectonophysics 218, 287–295 (1993)
18. Kameda, H., Sugito, M.: Prediction of strong earthquake motions by evolutionary process
model. In: Proceedings of the Sixth Japan Earthquake Engineering Symposium, pp. 41–48
(1978)
19. Abrahamson, N.A., Litehiser, J.J.: Attenuation of vertical peak acceleration. Bull. Seism. Soc.
Am. 79, 549–580 (1989)
20. Anderson, J.G., Hough, S.E.: A model for the shape of the Fourier amplitude spectrum of
acceleration at high frequencies. Bull. Seismol. Soc. Am. 74(5), 1969–1993 (1984)
21. Wells, D., Coppersmith, K.: New empirical relationships among magnitude, rupture length,
rupture width, rupture area, and surface displacement. Bull. Seismol. Soc. Am. 84, 974–1002
(1994)
22. Gupta, S.C., Singh, V.N., Kumar, A.: Attenuation of Coda Waves in the Garhwal Himalaya,
India. Phys. Earth Planet. Inter. 87(3), 247–253 (1995)
23. Kumar, D., Khattri, K.N., Teotia, S.S., Rai, S.S.: Modelling of accelerograms of two Himalayan
earthquakes using a novel semi-empirical method and estimation of accelerogram for a
hypothetical great earthquake in the Himalaya. Curr. Sci., 819–830 (1999)
24. Yadav, R.: Modelling of Empirical Accelerograms of 2011 Sikkim Earthquake (M 6.9), Kappa
(κ)- Model For NE Region (India) and Scenerio Seismic Hazard Maps for Sikkim Region of
Himalaya. Ph.D. Thesis, Kurukshetra University, Kurukshetra, Haryana, India (2019)
Development and Implementation
of a Regional Earthquake Early Warning
System in Northern India
Govind Rathore, Pankaj Kumar, Mukat Lal Sharma, Kamal,

Ravi Sankar Jakka, and Ashok Kumar
Abstract The Himalayan region has experienced many devastating earthquakes,

due to the collision of the Indian Plate with the Eurasian plate. The movement of
the Indian Plate is continuous but for a long time, this region has not released the
accumulated energy. This accumulated energy makes this region more vulnerable
to major to great earthquakes in near future. The urbanization and initialization in
this region have magnified the effect of earthquakes by many folds. Earthquake
Early Warning (EEW) systems are being used to mitigate the effects of earthquakes
in many countries, by sending alerts of upcoming damaging waves to the users.
The development of an EEW system in Uttarakhand was started in 2014 as a pilot
project after validating its feasibility, which is now operational for the public. The
operation EEW system includes (i) a central processing server at Roorkee; (ii) a
warning server at Google cloud platform; (iii) around 160 sensors in Kumaun and
Garhwal region of Uttarakhand; (iv) 80 public sirens at schools, hospitals, district
emergency centers, etc., and (v) a smartphone for the public. This paper elaborates on
the different components and architecture of the implemented EEW system. Further,
this paper also includes information about the tasks carried out to aware the public
about earthquake safety tips and the EEW system to utilize its potential at the user
end.
Keywords Earthquake early warning system · Warning dissemination system ·

Disaster risk reduction · EEW smartphone app
G. Rathore (B)
Regional Remote Sensing Centre North, National Remote Sensing Centre, Indian Space Research
Organisation, Bengaluru, India
e-mail: govind.eqd@gmail.com
P. Kumar
EEW System Laboratory, IIT Roorkee, CoEDMM, Roorkee, India
e-mail: pkumar@dm.iitr.ac.in
M. L. Sharma · R. S. Jakka · A. Kumar
Kamal
Department of Earth Sciences, IIT Roorkee, Roorkee, India
https://doi.org/10.1007/978-981-99-1459-3_43
538 G. Rathore et al.
1 Introduction
The Himalayan region has an area of around 2500 sq km and covers 13 Indian
states Uttarakhand, Himachal Pradesh, Jammu and Kashmir, Ladakh, Arunachal
Pradesh, Manipur, Meghalaya, Mizoram, Nagaland, Sikkim, Tripura, Assam, and
West Bengal, which is vulnerable to large earthquakes as this region has not seen
any major earthquake for a long time and storing the energy to generate one [21].
Earthquake Early Warning (EEW) systems are used to alert their end-users about
the upcoming destructive seismic waves by detecting the primary seismic waves
traveling at twice the speed of secondary waves. These systems have successfully
shown their potential in saving lives and economic losses in many developed countries
like Japan [4, 11, 12, 15–18, 20], Taiwan [6, 13, 22, 24, 25], Mexico [8–10], Turkey
[1, 7, 26], and Romania [3, 14, 23]. So, the EEW system could be also used in the
Himalayan region as a disaster risk reduction tool but it requires huge instrumentation
to cover all vulnerable areas. An initiative for developing an EEW system in Northern
India was started by the Ministry of Earth Sciences (MoES) and the Indian Institute
of Technology Roorkee (IITR) in 2014 [5], which is further funded and extended
by the Uttarakhand State Disaster Management Authority (USDMA), Uttarakhand
Government.
2 Regional Eew System in Uttarakhand
A regional EEW system is also being developed in Uttarakhand and has around 165
strong ground motion sensors that stream ground acceleration data over broadband
leased lines to a central server at IIT Roorkee. The central server has been equipped
with well-known software Earthworm, which is developed by USGS. The server has
various modules attached to Earthworm for data receiving, storing, processing, and
displaying. The central server is the main component of the developed system, as
it is responsible for earthquake detections and parameter calculations. The central
server uses Allen’s modified STA/LTA algorithm [2] for P-onset of the earthquake.
It uses the Pd parameter of 3 s for the magnitude determination [25]. The central
server sends all information about the detected earthquakes to the warning server,
which is disseminated to the required devices based on the earthquake parameters.
3 Sensor Array
The Uttarakhand EEW system uses low-cost accelerometer sensors, which are being
imported from Taiwan and the locations of deployed sensors in Uttarakhand. These
sensors stream real-time ground acceleration data to a central server situated at IIT
Roorkee over TCP/IP protocol. Further, we have also developed a sensor similar to
Development and Implementation of a Regional Earthquake Early … 539
Fig. 1 Deployed sensors Palert (left) and Palert + (right) in the fields
the existing one to reduce the cost of the EEW system, as well as maintenance of
these sensors, could be easily done. Further, a reduced cost of the sensors helps us
to extend the sensor network for covering a large portion of the Himalayan region.
The images of the deployed sensors are being shown in Fig. 1.
4 Warning Dissemination System
The Warning Dissemination System (WDS) is a complicated system that includes

dedicated warning devices, smartphone apps, desktop software, and a warning server
for receiving the earthquake information from the central server and pushing it to all
warning devices and software/apps. The warning server has an MQTT broker, which
is used to distribute the warning message to all sirens and desktop/laptop apps. A
database has been created to store all information related to public sirens, home sirens,
desktop, and smartphone app. A program named Siren Manager has been developed
for logging the status of all dedicated warning devices such as public and home
sirens into the database. The database also keeps the information about smartphone
app users and logs their conditions during the earthquake. A dedicated GIS-based
web portal has been developed for providing assistance to the maintenance team. The
maintenance team can log in to the web portal for viewing the real-time status of the
sirens on the maps. The web portal could be accessed from anywhere to make field
installations and maintenance easy. All newly installed sirens are being automatically
added to the database with the help of a program named Siren Manager. The early
warnings to the smartphone app are being disseminated through Firebase Cloud
Messaging (FCM) service by the warning server as running background processes
or sockets are not allowed in Android and iOS.
Fig. 2 EEW public sirens installed in the fields
5 EEW Warning Dissemination Devices
The dedicated EEW public sirens were developed for warning dissemination with the
help of loudspeakers [19], and image of the developed siren is shown in Fig. 2. These
sirens have three LEDs for giving information about power, system, and warning
server connectivity and three pushbuttons for validating connections at the time of
installation for easy testing of the siren. Further, these sirens have been equipped
with different panel power connectors for making installation easy and avoiding
misconnection. These sirens are being installed in the fields for more than 3 years
and successfully tested and validated. Currently, these sirens have been installed in
two major cities Dehradun and Haldwani of Uttarakhand and locations are shown in
Fig. 2.
6 Warning Dissemination App: Uttarakhand Bhookamp

Alert
The installation and maintenance of public sirens are time-consuming and labor-
intensive works, which creates a need for the development of a smartphone app for
warning dissemination to the public at large. So, a smartphone app for the public has
been developed for warning dissemination and officially launched for the public on
4 August 2021. The app could be downloaded from Play Store or App Store. The
app needs to be installed on the smartphone, and notification permission to the app
should be granted for receiving early warnings for the earthquakes. The developed
app could be used in two language options Hindi and English. The home screen of the
app has three tabs named Home, Instructions, and Recent Earthquakes. The home tab
is being used to provide instructional videos for users about safeguarding themselves
during earthquakes. The instruction tab is being used to give textual instructions for
safety in brief. The recent earthquakes tab gives details about the past earthquake
in the region. The warning message is sent through notification with a loud human
voice to alert users so that users should get alerted even the smartphone resides in the
pocket or at some distance. On tapping the alert notification, the user gets information
about the origin, magnitude, and time left for the impinging damaging earthquake
waves. 1 min after when the timer stops, users are being asked for their conditions.
If user presses “I need Help,” then this information is being shared with the disaster
management authority, and if user presses “I am Safe,” then the user is marked as
safe.
7 Mock Drills and Success Stories
Apart from the development of the EEW system, it is equally important to spread
awareness about the EEW system and dos/don’ts during the early warning in the
public. For this reason, we are constantly conducting monthly mock drills, which
encourage the public to install the app for their safety. We are sticking the poster
about steps that need to be followed during the earthquake as well as distributing
the pamphlets to the public about the app. Further, we are also spreading aware-
ness through conferences and stalls. We are also taking the help of news media and
radio channels. Some pictures related to these activities are being shown in Fig. 3.
The developed EEW system was successfully validated for 3 real earthquakes in
Uttarakhand. Both of these earthquakes were less than 5.0 magnitude, so users were
informed only through the smartphone apps with an early notification without any
alerting sound. The public sirens were also kept silent as earthquakes with a magni-
tude less than 5.0 are not damaging. The estimated magnitudes of the earthquakes
were not accurate, but close to the actual ones. These types of notifications build
confidence in the public and encourage people to install the app for their safety.
5 mock drills are also successfully conducted in public through public sirens and
smartphone app.
8 Conclusion
The first EEW system in Uttarakhand State has been successfully implemented with
around 165 sensors in Kumaun and Garhwal region with a central data processing
server at Roorkee for earthquake detection. Further, a warning dissemination server
has been established that stores all information about the smartphone app users and
public sirens and disseminates the early warnings to the desired users based on the
Fig. 3 Activities performed for spreading awareness about EEWS in India
earthquake parameters received from the central processing server. The implemented
EEW system has successfully delivered three earthquake early notifications to the
public after the app launches. Further, the complete warning dissemination system
is being tested publicly each month during the mock drills. The smartphone apps are
being developed for IOS and Android platforms for alerting the public, which is also
spreading the awareness about EEW system in public. The paucity of strong-motion
data for the development of GMPEs for this region makes it difficult to predict the
strong ground motion close to the accurate one. Therefore, the low-cost strong ground
motion sensors and sirens with low-cost MEMS-based sensors have been developed
to collect the data for such works in the near future.
9 Future Scope
The strong ground motion records for the Himalayan region are less in number
to get the regression analysis for magnitude estimation with various well-known
EEW parameters, due to the lack of instrumentation, which could be increased by
installing the low-cost sensors in this region. Further, a denser network of strong
ground motion sensors will help in detecting earthquakes faster and avoiding the
consequence of sensor/network failure. An array of low-cost sensors should be estab-
lished for covering all state 13 states for earthquake early warning systems. The data
collected by these sensors should be used for developing AI-based algorithms for
earthquake detection and size estimation. The public should be trained and there
should be some compulsory routine exercises and mock drills in the schools about
safeguarding themselves during the various regional disasters. The density of the
public siren should be increased to cover a large area for the safety of the public as
well as CAP protocol should be used to alert the public.
Acknowledgements The authors would like to acknowledge the Uttarakhand State Disaster
Management Authority, Uttarakhand Government, and Ministry of Earth science, Government
of India, for supporting and funding this project. This project is very important for the safety of
people, and the authors are also thankful to USDMA for their support in conducting mock drills
and spreading awareness.
References
1. Alcik, H., Ozel, O., Apaydin, N., Erdik, M.: A study on warning algorithms for Istanbul
earthquake early warning system. Geophys. Res. Lett. 36, (2009). https://doi.org/10.1029/200
8GL036659
2. Allen, R.V.: Automatic phase pickers: their present use and future prospeCTS. Bull. Seismol.
Soc. Am. 72, 225–242 (1982)
3. Böse, M., Ionescu, C., Wenzel, F.: Earthquake early warning for Bucharest, Romania: Novel
and revised scaling relations. Geophys. Res. Lett. 34 (2007). https://doi.org/10.1029/2007GL
029396
4. Brown, H.M., Allen, R.M., Grasso, V.F.: Testing elarms in Japan. Seismol. Res. Lett. 80,
727–739 (2009). https://doi.org/10.1785/gssrl.80.5.727
5. Chamoli, B.P., Kumar, A., Chen, D.-Y., et al.: A prototype earthquake early warning system
for Northern India. J. Earthq. Eng. (2019). https://doi.org/10.1080/13632469.2019.1625828
6. Chen, D.Y., Hsiao, N.C., Wu, Y.M.: The earthworm based earthquake alarm reporting system
in Taiwan. Bull. Seismol. Soc. Am. 105, 568–579 (2015). https://doi.org/10.1785/0120140147
7. Erdik, M., Fahjan, Y., Ozel, O., et al.: Istanbul earthquake rapid response and the early warning
system. Bull. Earthq. Eng. 1, 157–163 (2003). https://doi.org/10.1023/A:1024813612271
8. Espinosa-Aranda, J.M., Cuellar, A., Garcia, A., et al.: Evolution of the Mexican seismic alert
system (SASMEX). Seismol. Res. Lett. 80, 694–706 (2009). https://doi.org/10.1785/gssrl.80.
5.694
9. Espinosa-Aranda, J.M., Cuéllar, A., Ibarrola, G., et al.: The seismic alert system of Mexico
(SASMEX) and their alert signals broadcast results. 15th World. Conf. Earthq. Eng. (2012)
10. Espinosa-Aranda, J.M., Jimenez, A., Ibarrola, G., et al.: The seismic alert system in Mexico
City and the school prevention program. Early Warn. Syst. Nat. Disas. Reduct., 441–446 (2003).
https://doi.org/10.1007/978-3-642-55903-7_56
11. Horiuchi, S., Negishi, H., Abe, K., et al.: An automatic processing system for broadcasting
earthquake alarms. Bull. Seismol. Soc. Am. 95, 708–718 (2005). https://doi.org/10.1785/012
0030133
12. Hoshiba, M., Kamigaichi, O., Saito, M., et al.: Earthquake early warning starts nationwide in
Japan. Eos (Washington DC) 89, 73–74 (2008). https://doi.org/10.1029/2008EO080001
13. Hsiao, N.C., Wu, Y.M., Shin, T.C., et al.: Development of earthquake early warning system in
Taiwan. Geophys. Res. Lett. 36, 3–7 (2009). https://doi.org/10.1029/2008GL036596
14. Ionescu, C., Böse, M., Wenzel, F., et al.: An early warning system for Deep Vrancea (Romania)
earthquakes BT—earthquake early warning systems. In: Gasparini, P., Manfredi, G., Zschau,
J. (eds.) Springer, pp. 343–349. Berlin Heidelberg, Berlin, Heidelberg (2007)
15. Kamigaichi, O., Saito, M., Doi, K., et al.: Earthquake early warning in Japan: warning the
general public and future prospects. Seismol. Res. Lett. 80, 717–726 (2009). https://doi.org/
10.1785/gssrl.80.5.717
16. Nakamura, Y.: UrEDAS, urgent earthquake detection and alarm system, now and future. 13th
World Conf. Earthq. Eng. (2004). ISBN 4-89580-010-5
17. Nakamura, Y., Saita, J.: UrEDAS, the earthquake warning system: today and tomorrow. In:
Gasparini, P., Manfredi, G., Zschau, J. (eds.) Earthquake Early Warning Systems, pp. 249–281.
Springer, Berlin Heidelberg, Berlin, Heidelberg (2007)
18. Odaka, T., Ashiya, K., Tsukada, S., et al.: A new method of quickly estimating epicentral
distance and magnitude from a single seismic record. Bull. Seismol. Soc. Am. 93, 526–532
(2003). https://doi.org/10.1785/0120020008
19. Rathore, G., Kumar, A., Jakka, R.S., Chamoli, B.P.: Development of earthquake early warning
Siren for Regional earthquake early warning system in India. In: 16th Symposium on
Earthquake Engineering. Roorkee, pp. 154 (2018)
20. Saita, J., Nakamura, Y.: UrEDAS: the early warning system for mitigation of disasters caused
by earthquakes and Tsunamis. Early Warn Syst. Nat. Disas. Reduct., 453–460 (2003). https://
doi.org/10.1007/978-3-642-55903-7_58
21. Sreejith, K.M., Sunil, P.S., Agrawal, R., et al.: Audit of stored strain energy and extent of future
earthquake rupture in central Himalaya. Sci. Rep. 8, 1–9 (2018). https://doi.org/10.1038/s41
598-018-35025-y
22. Wen, K.-L., Shin, T.-C., Wu, Y.-M., et al.: Earthquake early warning technology progress in
Taiwan. J. Disas. Res. 4, 530–538 (2009). https://doi.org/10.20965/jdr.2009.p0202
23. Wenzel, F., Oncescu, M.C., Baur, M., et al.: An early warning system for Bucharest. Seismol
Res Lett 70, 161–169 (1999). https://doi.org/10.1785/gssrl.70.2.161
24. Wu, Y.M., Teng, T.L.: A virtual subnetwork approach to earthquake early warning. Bull.
Seismol. Soc. Am. 92, 2008–2018 (2002). https://doi.org/10.1785/0120010217
25. Wu, Y.M., Zhao, L.: Magnitude estimation using the first three seconds P-wave amplitude in
earthquake early warning. Geophys. Res. Lett. 33, 4–7 (2006). https://doi.org/10.1029/2006GL
026871
26. Zschau, J., Isikara, M., Ergünay, O., et al.: Towards an earthquake early warning system for
the Megacity of Istanbul. Early Warn. Syst. Nat. Disas. Reduct., 433–440 (2003). https://doi.
org/10.1007/978-3-642-55903-7_55
A Critical Review of Existing Building
Regulations and Bye-Laws in Hilly
Regions of India
Abstract In this article, an attempt has been made to understand and review the
existing Building Regulations and Bye-laws available in the hilly region of India,
primarily focusing on the structural configurations and construction provisions, like
the number of stories and building height in seismically active hilly areas. This
article also examines and compares the customization of individual-specific regula-
tions, framed after considering guidance from Model Building Bye-Laws 2016. A
comprehensive summary is presented herein, including most of the available Bye-
Laws containing the construction and architectural regulations in hilly regions, which
will help classify different building typologies along with their associated constraints.
Keywords Bye-laws · Seismicity · Hilly regions · Topography · Building

regulation
1 Introduction
Rapid urbanization, increasing population, tourist influx, and scarcity of flat land
led to extensive building constructions in hilly regions. Consequently, these factors
drove the enormous developmental activities beyond the region’s land capacity. The
studies on the impact of past strong earthquakes, specifically on the structures in the
seismically active hilly areas, affirmed the detrimental performance of the existing
buildings. The combined action of potential seismic activity, traditional design,
and non-engineered construction practices on sloping land has caused human lives
and property devastation after strong seismic events like Sikkim Earthquake 2011
[1] and Nepal Earthquake 2015 [2]. To cater these issues related to buildings, central
authorities have formulated specific standards called Bye-laws, at the municipal level.
P. Das Choudhury · D. Raj (B)

Department of Civil Engineering, MNIT Jaipur, Jaipur, Rajasthan 302017, India
e-mail: dhiraj.ce@mnit.ac.in
P. Das Choudhury
Department of Civil Engineering, UEM Jaipur, Jaipur, Rajasthan 303807, India
https://doi.org/10.1007/978-981-99-1459-3_44
546 P. Das Choudhury and D. Raj
Building bye-laws are general guidelines, which is applicable to a particular

municipal area; used to control the building’s coverage, height, design, and construc-
tion; and to execute resilient and orderly development of that area. These laws
protect buildings against structural failures, fire, earthquake, noise, and other hazards.
However, haphazard coverage, development, and encroachment are witnessed in
various small to medium-sized towns, which do not fall under the jurisdiction
of building bye-laws. The second revision of model building Bye-laws, “Model
Building Bye-Laws 2016,” [3] was released by the Ministry of Urban Development,
India, to incorporate modern and advanced provisions for construction in a hazard-
prone area and with many other conditions (like, regulation for, conservation of
heritage sites including heritage buildings, Rooftop Solar PV Installation, etc.) and
integrate the same to the existing Bye-laws of different regions.
Many researchers studied existing building regulations and their application to
specific major hilly towns/cities to assess the extent of unprecedented development
[4, 5]. Along with controlling haphazard development, building regulations play a
critical role in deciding the general building typologies witnessed in a city or town,
concerning structural configuration based on foundation level and slope retaining
system [6].
In this paper, an attempt is made to catalog and inspect the various critical parame-
ters concerning hilly regions of India in existing Building Regulations and Bye-laws
available in the hilly area of India. The selection of parameters depended on the struc-
tural and construction provisions, like the maximum height of buildings, the number
of stories, slope in hilly areas, and others. Based on this review, the conclusions have
been drawn and presented herein.
2 Overview
This paper concisely articulates Building regulations and Bye-laws for 13 states
and Union Territories encompassing north, northeast, east, and south India, based
on the altitude and severity of seismic zonation. The states covered in this study
are Himachal Pradesh, Uttarakhand, Jammu and Kashmir, Sikkim, Meghalaya,
Arunachal Pradesh, Manipur, Mizoram, Nagaland, Assam, West Bengal, Tamil Nadu,
and Kerala. The Building regulations, selected for this review, are taken from the latest
existing building bye-laws and development plans available on the corresponding
websites of local/state governing authorities. The general information of the region,
like the population, seismic zone, range, total geographical area, and proportionate
hill area, is listed in Table 1.
A Critical Review of Existing Building Regulations and Bye-Laws … 547
Table 1 Hilly regions/states with their characteristics

State Population* Seismic zone Ranges/region GA Geographical
geographical area in hill
area (km2 ) [7] districts (km2 )
[7]
Himachal 6,864,602 IV, V Western 55,673 55,673
Pradesh Himalayas
Uttarakhand 10,086,292 IV, V Western 53,483 53,483
Himalayas
J&K 12,541,302 IV, V Western 222,236 222,236
Himalayas
Sikkim 610,577 IV Eastern 7096 7096
Himalayas
Meghalaya 2,966,889 V Khasi and 22,429 22,429
Jaintia Hills
Arunachal 1,383,727 V Eastern 83,743 83,743
Pradesh Highlands
Manipur 2,855,794 V Purvanchal 22,327 22,327
Mizoram 1,097,206 V Purvanchal 21,081 21,081
Nagaland 1,978,502 V Purvanchal 16,579 16,579
Assam 31,205,576 V Barail Range 78,438 19,153
West Bengal 91,276,115 III, IV Eastern 88,752 3149
Himalayas
Tamil Nadu 72,147,030 II Eastern Ghats 130,058 22,789
Kerala 33,406,061 III Western Ghats 38,863 29,572
* According to 2011 Census of India
3 Building Regulations for Residential Buildings
Several critical parameters have been identified in this study, and based on these
parameters, a detailed review of the building regulations or bye-laws is presented
here and summarized in Tables 2 and 3:
1. Plot Size: Depending on the housing types, different hilly regions have different
plot size limits, which is necessary for orderly development of municipal area.
2. Setbacks: Setbacks are the vacant area around a building to ensure optimum
daylight and ventilation, which later can be modified into landscaping, as shown
in Fig. 1a. Setbacks also depend on building typology. The plot size’s dimension
decides the setbacks’ size, as smaller plots demand lesser setback requirements
than bigger plots.
3. Maximum Ground Coverage: Ground coverage represents the percentage of
the total plot area used by the building footprint and varies with the size of the
plot area. In northern hilly states, the ground coverage ranges from 30 to 80%.
Table 2 Summary of existing building regulations/bye-laws in hill states (part 1)
548
State Housing type1 /plot Setbacks (m) Max. ground Max. F.A.R.3 Max. height of building (m) No. of Stories
size (m2 ) F/L/R/Re2 coverage
(HP) R 90–120 3.0/-/-/2.0 63 (Bilaspur), 1.50–2.00 Amb Gagret, 21.00 4.0 to 5.0,
Himachal 65 Dharamshala, except
Pradesh (Dalhousie, Ghumarwin, Chamba and
[8–18] Hamirpur, Kandaghat, Nahan (3.0)
Manali, Kullu Bhuntar,
Mandi, Una), Nadaun, Paonta
75 (Chamba, Sahib,
Kasauli, Parwanoo,
Nadaun, Rampur
Nahan, Bushahr,
Parwanoo, Reckong Peo,
Solan) Rohru,
Sundernagar,
Theog,
Trilokpur
SD 121–150 2.0–3.0/2.0/2.0/2.0 60 (Bilaspur, Bilaspur, 18.80
Dalhousie, Manali,
Hamirpur, Manikaran,
Mandi, Palampur
Parwanoo),
75 (Chamba,
Nahan, Solan)
(continued)
Table 2 (continued)
121–250 50 (Nadaun), Bir-Billing 18.00

60 (Kasauli,
Manali, Una)
D 121–250 3.0/2.0/2.0/2.0 55 (Mandi), Chamba, Nahan 10.70
60 (Manali)
151–250 2.0–3.0/2.0/2.0/2.0 50 (Bilaspur, Hamirpur 15.50
Kasauli,
Nadaun), 55
(Dalhousie,
Hamirpur,
Parwanoo),
60 (Chamba,
Nahan, Solan,
Una)
251–500 3.0/2.0–3.0/2.0–3.0 45 (Kasauli, Kasauli 20.00
Nadaun), 50
(Bilaspur,
Chamba,
A Critical Review of Existing Building Regulations and Bye-Laws …
Dalhousie,
Hamirpur,
Mandi,
Nahan,
Parwanoo,
Solan, Una),
55 (Manali)
(continued)
549
Table 2 (continued)
550
501 and 3.0–5.0/3.0/3.0, 40 (except for Mandi, Shimla 13.50

above 2.0–3.0 Manali and
Shimla, which
is 50)
(UK) RS 30–45 1.20/-/-/- 80 1.80 Hilly regions 12.00 4.0
Uttarakhand (general)
[19] 46–75 1.50/-/-/- 80 1.80 Mussoorie 11.00 4.0
76–150 1.50/-/-1.20 75 1.60 Nainital 7.50 2.0
Municipal Area
151–300 2.00/-/-/1.50 70–65 1.50 Char Dham 6.50 2.0
towns
301–500 3.0/1.50/-/2.00 65–55 1.40
501–750 4.0/2.5/2.0/2.5 55–50 1.30
751–1000 5.0/3.5/3.5/3.5 50 1.20
1001 and 6.0/4.0/4.0/4.0 45 1.10 NA NA NA
above
(J) Jammu PR Up to 75 1.50/-/-/- 75 12.00 3.0
region of J&K 76–125 2.50/-/-/1.50 75
[20, 21]
126–275 3.00/-/-/2.00 65
PSD 276–350 4.0/2.0/-/2.0 55
(continued)
Table 2 (continued)
351–450 5.0/2.0/-/2.0 50
PD 451–600 6.0/3.0/2.0/3.0 45
601–1000 7.0/3.0/2.0/3.0 40
1001 and 10.0/3.0/2.0/3.0 35
above
G 4000–8000 1/3rd of height of 30 1.75 40.00 NA
8001 & building or 7.5 m or 30 2.25 40.00 NA
above building line of the
abutting road
whichever is more
(K) Kashmir P(Cat. NA 10/10/10/10 60 2.40 16.76 NA
region of J&K A)
[22, 23] P(Cat. 10/10/10/10 55 2.20 15.24
B)
P(Cat. 10/10/10/10 50 2.00 15.24
C)
P(Cat. 10/10/10/10 50 15.24

D)
P(Cat. 10/10/10/10 50 2.00 15.24
D)
(continued)
551
Table 2 (continued)
552
P(Cat. 10/10/10/10 50 1.00 9.14

E)
F(Cat. 1/4th of height of 40 2.80 To be determined on the basis of maximum
A) building or 4.6 m ground coverage and maximum FAR/FSI
F(Cat. whichever is more 40 2.40
B)
F(Cat. 40 2.40
C)
F(Cat. 40 2.00
D)
F(Cat. Not permitted in this category of towns/areas
E)
Note The towns/cities are categorized into A, B, C, D, and E on the basis of Physiography/Terrain of this region. The elevation
progressively increases from A to E
(SK) Sikkim RS 251–500 3.0/1.5/1.5/1.5 70 NA Accordingly 16.76, 4.57, 5.5, 1.5, 3.5,
[24] 501–930 50 restricted as per 10.67, 7.62, 2.5, 1.5, and
stability of the 4.57, and no no
Above 920 40 area as construction is construction
identified by the allowed for allowed for
mines and categories 1, categories 1,
geology 2, 3, 4, 5, and 2, 3, 4, 5, and
department and 6, respectively 6, respectively
the land profile.
The stability of
area is divided
into categories
1, 2, 3, 4, 5, and
6
(continued)
Table 2 (continued)
(ML) RS Less than 3.0/1.0/1.0/1.8 50 (for 2.00 Residential 13.00 4.0

Meghalaya 185 residential Bungalows
[25] 186–464 3.0/1.2/1.2/1.8 bungalows Residential 19.00 6.0
and Apartments
apartments)
465–697 3.0/2.1/1.8/2.4
698–1858 4.0/3.0/2.4/3.0
1859–3251 5.0/4.5/3.0/3.0
3252–4645 6.0/5.0/5.0/6.0
Above 6.0/6.0/6.0/6.0
4645
(AR) RS 48 2.0/1.0/1.0/1.0 75 1.50 8.40 NA
Arunachal 49–60 2.0/1.0/1.0/1.0 75 1.50 8.40
Pradesh [26]
61–100 3.0/1.2/1.2/1.2 65 1.80 14.40
101–250 3.0/1.2/1.2/1.2 65 1.80 14.40
251–500 3.0/1.2/1.2/1.2 60 2.00 17.40
501–1000 3.0/1.2/1.2/1.2 50 2.50 17.40

1000–1500 3.0/1.2/1.2/1.2 50 2.50 17.40
1501–3000 3.0/1.2/1.2/1.2 45 2.25 17.40
(continued)
553
Table 2 (continued)
554
(MN) Manipur RS 90–150 1.50/0.9/0.9/0.9 NA 1.80 11.00 3.0

[27] 150–300 2.0/1.2/1.2/1.2 1.80 11.00 3.0
300–500 3.0/1.8/1.2/2.0 1.80 13.00 4.0
500–750 4.0/2.4/1.2/3.0 2.00 13.00 4.0
750–1000 5.00/4.2/2.5/4.5 2.00 16.00 5.0
1000–1500 6.0/4.5/3.0/4.50 2.50 19.00 6.0
1500–2000 7.0/4.5/3.5/4.5 2.50 22.00 7.0
2000–2500 8.0/4.5/4.0/4.5 2.50 25.00 8.0
(MZ) RS 46.45–93 1.20/0.6/0.6/0.6 71 3.00 for residential A 12.85 4.0
Mizoram [28, 94–130 1.5/0.9/0.9/0.9 68 (sleeping accommodation 16.00 5.0
29] < 150m2), 2.50 for
Above 130 2.0/1.2/1.2/1.2 68 residential B (sleeping 19.00 6.0
accommodation > 150m2)
(NL) RS 125 2.0/-/-/1.8 NA NA Depend on building plans accompanied by
Nagaland [29] 125–500 2.0/-/-/2.0 relevant structural designs and drawings
depending on soil conditions, adoption of
501–1000 3.0/-/-/3.0 foundations duly certified by the structural
engineer/geo-technical expert
(continued)
Table 2 (continued)
Above 3.0/-/-/5.0
1000
(AS) Assam RS 134 3.0/1.5/1.5/3.0 50 1.00 11.50/shall not exceed 1.5 times 3.0
[30] 200 3.0/1.8/1.8/3.0 50 (For both 1.25 of the width of the road plus
RCC and front setback
Assam type)
402 3.0/1.8/1.8/3.0
(WB) West RS 100–400 1.5/1.5/1.5/1.5 70 For 3, 3–5, 5–7.5, and For 3, 3–5, 5–7.5, and above 7.5, NA
Bengal [31, 400–1200 60 above 7.5, width (means width (means for access + Open
32] for access + Open space), space), heights are 4.50, 6.50,
Over 1200 50 FAR are 1.00, 2.00, 2.50 11.50, and 13.50
and 2.75
(TN) Tamil RS NA 1.50/–/–/– NA NA 7.00 2.0
Nadu [33–35]
(KR) Kerala RS NA 3.0/1.0/1.0/1.5 65 3.00 Shall not exceed twice the width NA
[36, 37] of the street abutting the plot
plus twice the setback from the
abutting road. If the building is

abutted by streets having varying
width, consider the street having
greater width
1R row, SD semi-detached, D detached, PR plotted row, PSD plotted semi-detached, PD plotted detached, G group housing, P plotted, F flatted, RS residential
2 F/L/R/Re front/left/right/rear,
3 F.A.R floor area ratio
555
Table 3 Summary of existing building regulations/bye-laws in hill states (part 2)

State Min. floor height Max. floor height Plinth level (m) Max. acceptable
(m) (m) slope (deg.)
HP [8–18] 2.70–3.00 3.5 0.45–4.00 45 (30 for Mandi)
UK [19] NA NA Not less than 0.45 30 (in case of
Nainital Municipal
Area, it shall not be
more than 26.5)
J [20, 21] 2.75 3 Not less than 0.45 30
K [22, 23] 2.75 3 0.45–1.5 30
SK [24] 2.75 (above NA NA NA
1372 m altitude),
3.00 (below
1372 m)
ML [25] 3.00 3.25 Not less than 0.45 NA
AR [26] NA NA Not less than 0.3 30
MN [27] 2.75 NA NA NA
MZ [28, 29] 2.40 NA Not less than 0.45 NA
NL [29] 2.75 NA NA NA
AS [30] 2.40 3.0 (4.8 for ground 0.5–0.75 45
story)
WB [31, 32] 2.75 3.6 NA 30 (in case of land
vulnerable to
debris slide)
TN [33–35] NA NA Not less than 0.45 33
KR [36, 37] 3 NA NA 45
Whereas, in Northeastern hilly states and the hilly region of West Bengal, it
ranges from 40 to 75% and from 50 to 70%, respectively.
4. Maximum FAR: FAR is the ratio of total covered area on all floors to the plot
area. It represents the extent of development (combined covered area/plinth area
of all floors) permissible on a plot area. Hence, FAR also varies according to plot
sizes, as shown in Fig. 1b. In hilly towns, smaller plots have lesser FAR than
larger plots, as the built-up area increases for smaller plot sizes. In northern hilly
states, especially in Himachal Pradesh and Uttarakhand, FAR ranges from 1.1 to
2.0, whereas in Jammu and Kashmir, it ranges from 1.0 to 2.8. In northeastern
hilly states, the range is from 1.0 to 3.0. In West Bengal, an eastern hilly state FAR
ranges from 1.00 to 2.75. In southern states, where hills regions are prominent,
like Kerala, FAR is limited to 3.0.
5. Maximum height of the building: The maximum height of the building and
number of stories depend on the elevation profile and seismic zone. In case of
building on slope, the maximum height is measured from the uppermost founda-
tion level to the roof of the building, as per Fig. 1c. In the North, the maximum
Fig. 1 a Setbacks b (a)

variation of FAR with plot
size c building on a sloping
ground
(b)
(c)
building height ranges from 10.7 to 21.0 m. In the Northeast, having a compara-
tively lesser peak and abundance of flat lands, the height of building varies from
8.4 to 25.0 m. In hilly region of West Bengal, the height of a building varies from
4.5 to 13.5 m, whereas in hilly region of Tamil Nadu, the height of a building is
limited to 7.0 m.
6. Number of stories: In Himachal Pradesh, the number of story ranges from 4

to 5, except for Chamba and Nahan which lie in seismic zone V. Rest of the
considered northern hilly states, the number of story ranges from 2 to 5, whereas
in the Northeast, this parameter ranges from 1.5 to 6. The variation of no. of
stories with respect to states is shown in Fig. 2a.
7. Floor height: The minimum and maximum floor height generally vary from 2.4
to 3.00 m and 3.60 m throughout the hilly states of India.
8. Plinth level: The plinth level remains in the same range from 0.30 m to 0.45 m
except in Assam, which has 0.70 m.
9. Maximum acceptable slope angle: Across the hilly regions, the maximum
acceptable slope angle is regulated and maintained between 30° and 45°, based
on topographical features and soil types, which also influence seismic activity.
The variation of this parameter with respect to states is shown in Fig. 2b.
5
No. of Stories
1
NA
NA
NA
NA
0 NA
HP UK J K SK ML AR MN MZ NL AS WB TN KL
States
(a)
50
Max. acceptable slope (deg.)
40
30
20
10
NA
NA
NA
NA
NA
HP UK J K SK ML AR MN MZ NL AS WB TN KL
States
(b)
Fig. 2 a Variation of no. of stories (average) with the states considered. b Variation of
max/acceptable slope with the states considered
In this paper, a summary of the existing Building Regulations and Bye-Laws available
for various states has been presented. The building regulations and bye-laws in hilly
regions are found significantly different from the regulations in flat terrain, regarding
some critical parameters, like structural configurations.
The most critical and differing parameter is the capping on maximum height of
buildings constructed in hilly regions compared to their flat-terrain counterparts. The
capping on maximum height of buildings in hilly terrain holds for both conventionally
designed and earthquake-resistant structures. The intensity of a region’s seismicity
also influences the building’s height limitation. For example, in Himachal Pradesh,
Chamba, and Nahan, being in seismic zone V, have a height limitation of 10.3 m or two
stories. Whereas, Manali being in seismic zone IV has a height limitation of 18.8 m
or five stories. Due to the scarcity of flat terrains in hilly states and increasing demand
of rapid population growth and urbanization, the buildings are often constructed to fit
the natural slope of the ground. This reality encouraged constituting the acceptable
limit of ground slope angle as another building regulation parameter. The limitation
of the ground slope angle of a particular region depends on the seismicity and type
of soil.
The available studies based on the after-effects of past earthquakes have exposed
the vulnerability of existing buildings in seismically active hilly terrain. These studies
have also shown that even though governing bodies made concrete stipulations of
regulation as the developmental rule in the hilly areas, disparities between the provi-
sion and enforcement of these mandatory regulations are prevalent. Frequent and
non-engineered construction/development activity on slopes beyond the permissible
slope angle and building height lead to several consequential hazards like landslides,
slope instability, and more seismically vulnerable building.
Considering the complex nature of hilly terrain, seismicity, increasing population,
and haphazard constructional practices, a critical review of the existing building
regulations and bye-laws of the significant Indian hilly states is presented in this
study. This study will aid in developing building typologies for seismic vulnerability
analysis and risk management of buildings specific to hilly regions.
Acknowledgements The authors are grateful to the Department of Civil Engineering, MNIT Jaipur,
for providing all the necessary facilities for this investigation.
References
1. Sharma M.L., Sinvhal A., Singh Y., Maheshwari B.K.: Damage Survey Report for Sikkim
Earthquake of 18 September 2011. Seismol. Res. Lett. 84(1), 49–56 (2013)
2. Lizundia B., Davidson R.A., Hashash Y.M.A., Olshansky R.: Overview of the 2015 Gorkha,
Nepal, Earthquake and the Earthquake Spectra Special Issue. Earthq. Spectra. 33(1_suppl),
1–20 (2017)
3. Model Building Bye-Laws, 2016. Town and Country Planning Organization, Ministry of Urban
Development (2016)
4. Kumar, A.: Pushplata: building regulations for hill towns of India. HBRC J. 11(2), 275–284
(2015)
5. Kumar, A.: Impact of building regulations on Indian hill towns. HBRC J. 12(3), 316–326 (2016)
6. Surana, M., Singh, Y., Lang, D.H.: Seismic characterization and vulnerability of building stock
in hilly regions. Nat. Haz. Rev. ASCE 19(1), 04017024 (2018)
7. The Committee to Study Development in Hill States Arising From Management of Forest
Lands with Special Focus on Creation of Infrastructure, Livelihood and Human Development.
Planning Commission, New Delhi (2013)
8. Draft Development Plan Shimla Planning Area 2041. Town and Country Planning Department,
Himachal Pradesh (2022)
9. Development Plan for Amb-Gagret Planning Area 2035. Town and Country Planning
Department, Himachal Pradesh (2016)
10. Development Plan Baijnath-Paprola Planning Area. Town and Country Planning Department,
11. Development Plan for Bilaspur Planning Area. Town and Country Planning Department,
12. Development Plan for Chamba Planning Area. Town and Country Planning Department,
13. Development Plan for Dalhousie Planning Area. Town and Country Planning Department,
14. Development Plan Dharamshala Planning Area 2035. Town and Country Planning Department,
15. Development Plan for Kasauli Planning Area. Town and Country Planning Department,
16. Draft Development Plan for Manali Planning area 2021. Town and Country Planning
Department, Himachal Pradesh (2005)
17. Development Plan Manikaran Special Area 2035. Town and Country Planning Department,
18. Development Plan Nadaun Planning Area 2035. Town and Country Planning Department,
19. The Uttarakhand Building Construction and Development Bye Laws/Regulations, 2011
(Amendment 2017). Mussoorie Dehradun Development Authority, Uttarakhand (2011)
20. Jammu and Kashmir Unified Building Bye-Laws. Housing and Urban Development Depart-
ment, Jammu and Kashmir (2021)
21. The Jammu Municipal Corporation (Building) bye-laws 2011. Jammu Municipal Corporation,
Jammu and Kashmir (2011)
22. Pre-Draft Uniform Building Code-2019 for Kashmir Region. Town Planning Organisation,
Kashmir, J&K (2020)
23. Building Regulations and Bye-Laws (Kashmir Division). Srinagar Municipal Corporation,
Jammu and Kashmir (2010)
24. The Sikkim Building Construction (Amendment) Regulations. Urban Development and
Housing Department Gangtok, Sikkim (2001)
25. Meghalaya Building Bye-Laws, 2021. Urban Affairs Department, Meghalaya (2021)
26. Arunachal Pradesh Building Bye-Laws—2019. Department of Town Planning and ULBs,
Arunchal Pradesh (2019)
27. The Imphal Municipal Council Building Bye-laws (First Amendment), 2019. Imphal Municipal
Corporation, Manipur (2019)
28. The Aizawl Municipal Council Building Regulations, 2012. Urban Development and Poverty
Alleviation Department, Mizoram (2013)
29. Aizawl Municipal Council Building Regulations (Amendment) 2019. Urban Development and
Poverty Alleviation Department, Mizoram (2019)
30. The Nagaland Building Bye-Laws 2012. Urban Development Department (2012)
31. Assam Notified Urban Areas (Other than Guwahati) Building Rules, 2014. Urban Development
(T) Department, Assam (2014)
32. The West Bengal Municipal (Building) Rules, 2007. Department of Municipal Affairs, West
Bengal (2007)
33. The West Bengal Municipal (Building) Rules, 2007 (Amendment). Department of Municipal
Affairs, West Bengal (2016)
34. Tamil Nadu Combined Development and Building Rules, 2019. Municipal Administration and
Water supply Department (2019)
35. Tamil Nadu District Municipalities (Hill Stations) Building Rules, 1993. Municipal Adminis-
tration and Water Supply Department, Tamil Nadu (1993)
36. The Tamil Nadu Hilly Areas Special Building Rules, 1981. Rural Development and Local
Administration, Tamil Nadu (1982)
37. Kerala Municipality Building Rules, 2019. Local Self Government (RD) Department, Kerala
(2019)
38. Kerala Municipality Building (Amendment) Rules, 2020. Local Self Government (RD)
Department, (2020)
Microseismic Analysis Using Event
Count and Potency Displacement
for Stability Evaluation
of an Underground Cavern
Vikalp Kumar , V. R. Balasubramaniam , and K. S. Divyalakshmi
Abstract The long-term stability of an underground hydroelectric powerhouse

cavern, which depends on the rockmass deformation rate, is paramount for tunnel
engineers. The underground powerhouse cavern of the Tala Hydroelectric Project
(THP), Bhutan, had faced several strata instabilities issues during its construction
and even post-construction. So, a Microseismic Monitoring System, which is contin-
uous real-time monitoring, was installed at this underground powerhouse cavern to
assess the rockmass strata status. Seismic source parameter event count and potency
displacement revealed the zone of maximum inelastic damage zone in both the major
tunnels, namely machine hall and transformer hall, while Gutenberg-Richter relation-
ship forecasted the maximum magnitude in and around the powerhouse cavern. This
helped to augment timely support measures to enhance the life of this underground
structure.
Keywords Microseismic monitoring · Underground cavern · Hydropower · THP ·

Seismic potency
1 Introduction
The life of a mega hydroelectric project may extend even beyond 50 years [1], and it
is required to keep the underground powerhouse safe and stable for this long lifespan,
which is a vital component of the hydroelectric project. The underground powerhouse
has many vital installations and 365 * 7 * 24 working personnel. Thus, it requires
continuous monitoring to save the lives and machinery from rockmass collapses that
may occur as violent rockbolt failure, rockburst, wall heaving, etc., and due to deep
micro-cracks, fault, high stress, etc. [2, 3]
To arrest the rockmass damage/collapse timely, support measures are needed.
Conventional instruments are useful to monitor the rockmass movements, but they
V. Kumar (B) · V. R. Balasubramaniam · K. S. Divyalakshmi

National Institute of Rock Mechanics, Bengaluru, Karnataka, India
e-mail: vikalpk@gmail.com
https://doi.org/10.1007/978-981-99-1459-3_45
564 V. Kumar et al.
Fig. 1 Microseismic monitoring system layout
are of limited use to monitor the deep micro-cracks in the rock mass before a macro-
scopic rockmass failure. The formation of micro-cracks deep inside the rockmass
and their build-up with time before a macroscopic failure releases energy in the form
of stress waves, i.e. seismic signals [4, 5]. These energies are basically Microseismic
signals that can be recorded with a network of suitable sensors installed in the cavern
rockmass (Fig. 1). Data from the sensor arrive at the recording workstation server
through the data acquisition unit and communication equipment via armoured and
network cable. It is a passive method that indicates the inelastic damage zones of the
underground structure by using spatiotemporal analysis of the microseismic events.
The seismic signals can be recorded to gain insight into the origin time, micro-
crack location, event count, seismic potency, longitudinal and transverse wave energy,
dynamic and static stress drop, apparent stress, corner frequency and various other
seismological parameters that are derived after processing the seismogram. Seismic
event clustering may help to provide information about the rockmass stability status
in and around a particular zone. Seismic potency P represents the rockmass volume,
of whatever shape, related to co-seismic inelastic deformation at the seismic source.
The seismic potency will further help for the identification of zones of likely rockmass
deformation.
The underground powerhouse of THP located in Chukka Bhutan has encoun-
tered several rockmass engineering issues since its construction to today. Rockbolts
have failed and ejected irregularly, which increases the threat to the life of working
personnel and damage to the machinery at this underground powerhouse cavern.
The wall of the machine hall cavern is converging at a slow rate [6, 7]. So, THP
underground cavern has a continuous induced threat. Therefore, National Institute
of Rock Mechanics (NIRM) has installed a Microseismic Monitoring System of the
Institute of Mines Seismology (IMS), Tasmania, Australia, at the THP powerhouse
cavern to monitor and evaluate the rockmass stability status.
This Microseismic Monitoring method in recent years has been developed as
an early warning method for underground rockmass safety monitoring in several
underground structures (during and post-excavation) and mines. Various countries are
using this method as an engineering practice to timely assess the rockmass stability
issues during and post-construction [8–10].
Microseismic Analysis Using Event Count and Potency Displacement … 565
2 Microseismic Monitoring System at the Underground

Powerhouse of Tala Hydropower Plant
The Microseismic Monitoring System at THP mainly consists of the following

components: 15 triaxial geophones, 15 uniaxial geophones, eight data acquisition
units (DAQ), data transmission equipment and a workstation (Figs. 2 and 3). All
the geophones were installed in the boreholes of various depths and are connected
to the seismic DAQ having DSL modem, UPS, netSP and ADC. The underground
laboratory at this powerhouse mainly consists of a workstation (for data recording,
processing and analysis) and data transmission equipment. The seismic DAQ is
further connected to the data transmission equipment in the underground laboratory
using an armoured shield cable. IMS data recording, processing and analysis software
are installed on the workstation which receives data as a seismic signal is generated
due to the micro-crack that occurred in and around the underground powerhouse at
THP.
Using the web, the recorded data at the underground laboratory can be transferred
anywhere in the world. Using the reconnaissance method, longitudinal (P) and trans-
verse (S) wave velocities were obtained [11] and other parameters were set up after
three months of the calibration of the entire microseismic monitoring system at THP.
Fig. 2 THP underground Powerhouse showing major tunnel and thirty installed geophones
566 V. Kumar et al.
Fig. 3 Microseismic monitoring system at THP
3 Four Major Wall Stability Status Using Microseismic

Data Processing and Analysis
For the period of one year from 01 January to 31 December 2020, seismograms were
recorded on the workstation of the underground laboratory. During microseismic
data processing, signals (an example Fig. 4) and noises (Fig. 5) were separated.
During this one-year time period, 20,007 seismograms were recorded and in those
only 109 sets of seismograms were considered microseismic events and the rest were
rejected as noise over a monitoring volume of 410 * 380 * 132 m3 . Microseismic
events were further analysed to understand the stability status of the walls (upstream
and downstream) of the machine and transformer hall using event count and potency
displacement contours (Fig. 6).
3.1 Upstream Wall of the Machine Hall
In the upstream wall of the machine hall, the maximum event count contour between
RD 70–100 m and elevation 509–517 m with a maximum value at (10,772,514)
(Fig. 7). But the maximum seismic potency displacement observed was not in the
same zone as the maximum number of events (Fig. 8). The cluster of events is more
in and around 514 m elevation and RD 70–100 m. As events having co-seismic
Fig. 4 Microseismic signal recorded by six sensors on 08 April 2020 at 17:49:49 h
Fig. 5 Noise recorded by microseismic monitoring system
inelastic deformation more in and around 490 m, this induces maximum potency
displacement at this elevation.
568 V. Kumar et al.
Fig. 6 Microseismic event in and around the THP underground powerhouse over a monitoring
volume (410 m * 380 m * 132 m)
Fig. 7 Event count (upstream wall: machine hall)

Fig. 8 Potency displacement (upstream wall: machine hall)
3.2 Downstream Wall of the Machine Hall
The event count and displacement contour of the downstream wall of the machine
hall are shown in Figs. 9 and 10, respectively. Most of the events occurred below
bus duct-1 but due to the occurrences of high local magnitudes at a lower elevation,
maximum potency displacement has occurred below 491 m elevation.
Fig. 9 Event count (downstream wall: machine hall)

570 V. Kumar et al.
Fig. 10 Displacement (downstream wall: machine hall)
3.3 Upstream Wall of Transformer Hall
Maximum event count in the upstream wall of the transformer hall occurred in and
around the location: X-10736 m, Z-503 m and RD: 73 m (Fig. 11). Maximum potency
displacement is around the location having co-ordinate X-10758 m, Z-493 m and
RD 110 m (Fig. 12). The difference associated with the location in event count and
potency displacement contour is due to the occurrences of high-magnitude events in
and around X-10758 m and Z-493 m.
3.4 Downstream Wall of Transformer Hall
If one considers only the downstream wall of the transformer hall, more event count
below EL 449 m and RD 78 m appear in the zone opposite bus duct-1 (Fig. 13).
Maximum displacement is also more in the below EL 493 m and RD 103 m. (Fig. 14).
The maximum deformation at EL 493 m and RD 103 m is because of high-magnitude
events in and around this location.
The stability of the powerhouse cavern, especially the four major walls, was
studied separately. Microseismic event count contour in these four major walls
requires future monitoring as the addition of more microseismic events in the zone
of maximum event count contour may increase the threat to the wall in those zones.
Fig. 11 Event count (upstream wall: transformer hall)
Fig. 12 Potency displacement (upstream wall: transformer hall)

572 V. Kumar et al.
Fig. 13 Event count (downstream wall: transformer hall)
Fig. 14 Potency displacement (downstream wall: transformer hall)

Maximum potency displacement was in and around the locations (RD: 10,797, EL:
492), (RD: 10,783, EL: 491) in the respective upstream and downstream walls of the
machine hall, while (RD: 10758, EL: 493) and (RD: 10,751, EL: 493) in the respec-
tive upstream and downstream walls of the transformer hall of the underground
powerhouse cavern. These four locations need immediate ground support systems
for the long-term stability of the underground cavern.
4 Statistical Hazard Assessment of THP Underground

Powerhouse Cavern
The occurrences of seismic events are not strictly an arbitrary process. A statistical
approach to the seismic event analysis in an underground cavern provides a reason-
able basis for seismic hazard assessment. The seismic events in and around the
underground cavern follow the same laws as those followed by natural earthquakes.
A prominent analysis of both types of seismicity obeys the Gutenberg-Richter law
(Number of events N-magnitude M relation) that on a logarithmic scale takes the
following equation:
log10 N = a − bM (1)
In this relationship, “a” and “b” are constant that depend on space and time. The
b-value is a measure of the seismic environment for a given region and defines the
possible seismicity pattern of that region like stress build-up pattern and material
conditions in the focal region [12]. The value of parameter “b” normally varies from
0.8 to 1.1 with a notional value of 1. A higher b-value characterizes more occurrences
of microseismic events of low magnitude, whereas a lower b-value shows the domi-
nance of higher-magnitude microseismic events. For this THP powerhouse during the
period of one year of monitoring, the values of parameters “a” and “b” are 1.02 and
0.49, respectively. As local magnitude drops by 1.02, there is an increase of approxi-
mately ten times in the number of microseismic events. As the value of parameter “b”
is low, it indicates there is a dominance of higher-magnitude microseismic events.
One important parameter that can be derived from this semi-log plot is the mean
recurrence time (Table 1). It appears that internal constant driving forces due to the
composition of rockmass are acting continuously on the walls of the powerhouse.
The values of the parameters a and b are 1.02 and 0.49, so the local magnitude
of the largest seismic event mmax to be expected in one year is about 2.08. So,
the occurrence of such high seismic magnitude in and around the powerhouse may
damage the structure. So, advanced safety precautions are required to handle such
high-magnitude seismic events at THP powerhouse cavern (Fig. 15).
As local magnitude increases, the mean recurrence time increases, and the prob-
ability of the corresponding events decreases. In this THP underground powerhouse
cavern, a microseismic event of local magnitude of 0.6 takes 68 days to recur and
574 V. Kumar et al.
Table 1 Probability table and recurrence times for various local magnitude
Local magnitude −0.6 −0.4 −0.2 0 0.2 0.4 0.6
Mean recurrence days 18 22 28 35 43 54 68
Pr (2 weeks) 0.54 0.46 0.39 0.32 0.27 0.22 0.18
Pr (1 month) 0.81 0.73 0.65 0.57 0.49 0.41 0.35
Pr (3 months) 0.99 0.98 0.96 0.92 0.86 0.80 0.72
Pr (6 months) 1.00 1.00 0.99 0.99 0.98 0.96 0.92
Pr (1 year) 1.00 1.00 1.00 1.00 1.00 0.99 0.99
Fig. 15 Gutenberg-Richter relationship for THP powerhouse
its probability to occur in two weeks is 0.18, while an event having a local magni-
tude of −0.6 will return in 18 days and its probability to occur in two weeks’ time
period is 0.54. It is evident that for a microseismic event of definite local magnitude,
higher is the probability as time period increases. For example, the probability of
occurrence of a seismic event of local magnitude −0.2 increases from 0.39 in two
weeks to 0.99 in six months. Therefore, microseismic events of lower magnitude
recur more frequently than the higher local magnitude events and their probability
also increases for events of the same local magnitude, and the probability of the
occurrence increases as time duration increases. The occurrence probability of a
seismic event of maximum magnitude 0.6 is 0.994 in one year.
5 Conclusions
Rockmass performance assessment of an underground cavern to ensure its long-term

stability is a challenging task. Damage in the form of micro-cracks and its growth with
time is a dynamic process. The development of micro-crack linkage with real-time
monitoring helps to plan the appropriate restorative steps in a timely manner.
Microseismic Monitoring analysis provides information in real time about the
stability status of the underground powerhouse rockmass cavern of THP, Bhutan.
Microseismic monitoring at THP provides a 3-D picture of micro-cracking in the
spatiotemporal domain over a defined monitoring volume. Of all the recorded wave-
forms, only 109 waveforms are genuine microseismic events. The walls of the
machine and transformer hall of this underground powerhouse were investigated
using seismic events and potency displacement contours. The seismic event count
contour indicates the probable failure zone in the future, while the seismic potency
demarcates the zone that requires immediate attention with ground support systems.
The parameter b of the Gutenberg-Richter law indicates that there are dominances
of higher-magnitude seismic events and forecasts the maximum local magnitude,
which will help to take the necessary steps in advance to minimize the damage due
to future higher-magnitude events.
This methodology could be extended to assess the condition of rockmass not only
in underground hydropower projects, but also in underground hydrocarbons, LPG
storage caverns and a variety of other underground rockmass structures.
Acknowledgements The authors thankfully acknowledge the facility and support extended by Dr.
H. S. Venkatesh, Director, National Institute of Rock Mechanics, to publish this research work.
We are thankful to the management of THP, DGPC Ltd., for providing the required funds for this
research work.
576 V. Kumar et al.
References
1. Dincer, I., Ishaq, H.: Hydro energy-based hydrogen production. In: Renewable Hydrogen
Production. Elsevier, pp. 191–218 (2022)
2. Wang, M., Zhou, J.W., Shi, A.C., Han, J.Q., Li, H.B.: Key factors affecting the deformation and
failure of surrounding rock masses in large-scale underground powerhouses. Adv. Civ. Eng.
(2020). https://doi.org/10.1155/2020/8843466
3. Backstram, A.: Rock Damage Caused by Underground Excavation and Meteorite Impacts
(2008)
4. Yuan-hui, L., Gang, L., Shi-da, X., Da-wei, W.: The spatial-temporal evolution law of micro-
seismic activities in the failure process of deep rock masses. J. Appl. Geophys. 154(2017),
1–10 (2018). https://doi.org/10.1016/j.jappgeo.2018.04.024
5. Lai, X., Jia, C., Cui, F., Chen, J., Zhou, Y., Feng, G.: Microseismic energy distribution and
impact risk analysis of complex heterogeneous spatial evolution of extra—thick layered strata.
Nat. Publ. Group UK (2022). https://doi.org/10.1038/s41598-022-14538-7
6. Bhasin, R., Pabst, T.: Case study of a large underground cavern in the Himalayas. J. Rock
Mech. Tunn. Technol. 20(1), 5–20 (2014)
7. Bhasin, R., Pabst, T.: Finite element and distinct element analysis of the stability of a large
underground hydropower machine hall in the Himalayas. KSCE J. Civ. Eng. 19(3), 725–732
(2015). https://doi.org/10.1007/s12205-013-1351-4
8. Guanglia, F., Ma, Q., Zhang, X., Qu, D., Wang, G., Liu, J.: Developments of microseismic moni-
toring technology in deep tunnels in China. In: Proceedings of the 8th International Conference
on Civil Engineering, pp. 541–548 (2022)
9. Hudyma, M., Brown, L., Cortolezzis, D.: Seismic risk in Canadian Mines. In: CIM MEMO—
Maintenance Engineering and Mine Operations (2016)
10. Alexander, J., Trifu, C.: Monitoring mine seismicity in Canada. In: Proceedings of the Sixth
International Symposium on Rockburst and Seismicity in Mines Proceedings, pp. 353–358
(2005). https://doi.org/10.36487/acg_repo/574_34
11. Kumar, V., Jha, P.C., Singh, N.P., Cherukuri, S.: Dynamic stability evaluation of underground
powerhouse cavern using microseismic monitoring. Geotech. Geol. Eng. (2020). https://doi.
org/10.1007/s10706-020-01588-9
12. Kulhanek, O.: Seminar on b-value (2005) [Online]. Available: https://citeseerx.ist.psu.edu/vie
wdoc/download?doi=10.1.1.627.514&rep=rep1&type=pdf
Characteristics of Strong Ground
Motions for Delhi National Capital
Region (NCR) Using Small to Moderate
Size Earthquakes
Abhishek, Manisha Sandhu, and Babita Sharma
Abstract The characteristics of strong ground motions such as spectral acceleration

and their behaviour on the structural design code are rigorously studied using small
to moderate earthquakes occurred in the Delhi NCR region and recorded at local
seismic network installed by IIT, Roorkee under the sponsorship of the Ministry of
Earth Sciences. Seven earthquakes with magnitude range 3.3 to 4.9 occurred at depths
of 5 to 22 km have been used in the present study. Total 159 earthquake waveforms
from 27 seismic stations are processed for this purpose. The observed peak ground
acceleration (PGA) associated with 5 March 2012 (M 4.9) lies in range of 2.4 to
30.8 cm/s2 and the peak ground velocity (PGV) lies in between 0.07 and 0.88 cm/s.
The maximum PGA (30.8 cm/s2 ) is observed at Jaffarpur station and minimum
PGA (2.6 cm/s2 ) is observed at ridge observatory of New Delhi. The maximum
horizontal spectral amplification occurred at 0.054 s in Holocene age formation,
0.109 s in Pleistocene age formation and 0.068 s in Proterozoic age formation. The
maximum vertical spectral amplification is at 0.045 s in Holocene age formation,
0.084 s in Pleistocene age formation and 0.060 s in Proterozoic age formation. The
normalized acceleration response spectra for Holocene, middle to late Pleistocene
and Proterozoic age group is overestimated when compared with that of the Bureau of
Indian standard code particularly in short period range, whereas it is underestimated
for the longer periods associated with the available data. The present analysis is
beneficial for the seismic hazard studies in the Delhi NCR and may be utilized to
improve the design code by characterizing the strong motion scenarios in the region.
Keywords Strong ground motion · Amplification · Normalized response spectra
Abhishek (B) · B. Sharma

National Centre for Seismology, Ministry of Earth Science, New Delhi, India
e-mail: abhishekthakur2407@kuk.ac.in
M. Sandhu
https://doi.org/10.1007/978-981-99-1459-3_46
578 Abhishek et al.
1 Introduction
National capital region (NCR) is well known seismically active region of India
which experienced small to moderate size earthquakes in the past. This region lies
in the Indo-Yamuna plain situated at the foot of the Himalayas. Many seismotec-
tonic features like faults, sub-surface faults, lineaments pass through this region.
According to the Bureau of Indian Standards [1], Delhi NCR region falls in seismic
zone IV which is associated with the seismic intensity scale of VIII (MMI scale). On
5 March 2012 at 07:41:05 IST, an earthquake of magnitude 4.9 occurred in Delhi
NCR region (Delhi-Haryana Boarder) at the depth of 14 km, which created panic
among peoples in the early morning. This earthquake was well recorded at 20 seismic
stations installed in this region. There was another similar earthquake six months ago
on 9 September 2011 with magnitude of 4.3 hit this region near Sonipat, Haryana.
Due to its proximity to the central Himalayan seismic gap, its own intense seismicity,
many fold increase of population in metropolitan city Delhi, and lower standard of
multi-storey buildings, there is an urgent need for the seismic hazard assessment
for this region in an appropriate manner. The importance of response spectra is well
known among the seismologist and civil engineers. It is used to design the engineering
structures that can bear the forces generated by earthquakes as civil structures expe-
rienced three-dimensional earthquake ground motion [2]. It is very simplest method
to estimate the response of engineering structures to the ground motion produced by
earthquakes [3]. The concept of response spectra was first coined by Biot [4, 5] and
Housner [6]. The response spectrum is the graph between maximum response value
of single degree of freedom system within the period range of interest and collection
of them at various damping level known as response spectra. Housner [7] estimate
the response spectra using number of earthquakes and developed an average acceler-
ation response spectrum using different level of damping. Hayashi et al. [8] studied
the response spectra at various soil conditions in Japan. Chopra and Choudhary
[3] studied the response spectra at different geological age units in Gujarat region.
Sharma et al. [9] studied the characteristic of strong ground motion in Nepal India
boarder region and calculated the response spectra according to geological exposure
of the region. In the present study, we are trying to characterize the strong ground
motion for national capital region using small to moderate size earthquakes. Different
seismic hazard attributes like peak ground acceleration, peak ground velocity, peak
ground displacement and acceleration response spectra are estimated from accelero-
grams. Peak ground acceleration is very important ground motion parameter which
is used to build earthquake resistant structures. This kind of study is needed for the
Delhi NCR region as this region associated with seismic activity not only in the local
regime but also from the adjacent Himalayan region.
Characteristics of Strong Ground Motions for Delhi National Capital … 579
2 Geology and Seismotectonic
The Delhi NCR is basically flat region except central Delhi and northern part of
Rajasthan due to the presence of Delhi-Haridwar ridge and Aravalli Mountain belt
[10]. The Yamuna River passes through the north-eastern side of the region which
deposits sediments derived from the Himalayas into the study region. The upper layer
is mostly composed of alluvium deposits. These alluvium deposits are divided into
younger and older alluvium deposits. The younger alluvium composed of slit and
clay, whereas sand silt clay kankar are present in older alluvium. The important rock
formation is Delhi super group which consists of quartzite of Alwar series. The whole
region is divided into three geological age groups, namely Proterozoic (Older), Pleis-
tocene and Holocene (Younger) as shown in Fig. 1. Proterozoic age group mainly
composed of phyllite, slate, limestone, quartzite and schist. Pleistocene age group
composed of oxide slit-clay with kankar and micaceous sand, whereas Holocene age
group consist of grey micaceous sand, slit and clay. Tectonically, this region is rich
in form of various faults, ridges and lineaments. The Mahendragarh-Dehradun sub-
surface fault (MDF), Delhi-Haridwar Ridge (DHR), Delhi-Sagodha Ridge (DSR),
Sohna fault and Mathura fault are major tectonic features. The Rajasthan Great
Boundary Fault (RGBF) zone is another well-defined fault, which runs for about
400 km NNE–SSW to NE–SW as a major dislocation zone in Rajasthan. The inter-
section of these faults with Delhi axis of folding is found to be the reason for the
associated seismicity of this region as shown in Fig. 1.
Fig. 1 Geological set-up and seismotectonic set-up of the study area. Red stars are the earthquake
events and blue triangle are seismic stations used in present analysis
580 Abhishek et al.
3 Strong Ground Motion Data and Analysis
The strong ground motion data are available from accelerographs installed by Indian
Institute of Technology, Roorkee IIT-R and sponsored by Ministry of Earth Science,
New Delhi. The data is downloaded from PESMOS website (https://pesmos.org/dow
nload-ground-motion-records-2005-2017). The data from accelerometers installed in
Delhi NCR region are used for the present analysis. These accelerometers (strong
ground motion sensor) are Kinemetrics K-2 (K-2S) type with model episensor
(Internal accelerometer) having digitizer of 18 bit [11]. All the instruments within
Delhi NCR region are set at sampling frequency of 200 samples per second. The
threshold level 0.001 was set for all the instruments. The recording is done on a
256 MB GeoSIG or 1 GB Kinemetrics compact flash card [11]. The earthquakes of
magnitude range 3.3 to 4.9 occurring at the depth of 5.0 to 22.0 km are used in present
analysis. Figure 1 shows the analysed earthquakes, seismic stations along with seis-
motectonic present in the region. Total 159 accelerographs recorded at 27 seismic
stations are used to generated the acceleration response spectra and strong ground
motion characteristics. Out of total 27 stations 19 stations are lie on middle-late
Pleistocene formation; 4 stations lie on Holocene formation, whereas 4 stations lie
on Proterozoic formations as given in Table 1. We observed the majority of seismic
stations lies on the middle-late Pleistocene age site. All waveform data are analysed
carefully to obtained the accurate and precise results. The accelerographs are band
pass filtered in frequency range 0.01 to 20 Hz and then the baseline correction is
applied. The filtered and the baseline corrected waveforms are then used to estimate
the peak ground acceleration (PGA), peak ground velocity (PGV) and peak ground
displacement (PGD). The maximum value obtained from acceleration waveform,
velocity waveform and displacement waveform are used as PGA, PGV and PGD,
respectively.
The acceleration response spectra are determined from each station and then
grouped together according to the same geologically site conditions. The acceleration
response spectra are estimated using fast Fourier transformation (FFT) integration
method. The acceleration response spectra are evaluated from the geometric mean
of two horizontal components and vertical component at 5% damping value. The
obtained acceleration response spectra are then normalized with the first value of
ground acceleration to get normalized acceleration response spectra (NARS). The
normalization of acceleration spectra is applied to both geometric mean of horizontal
components and vertical component. The mean and standard deviation are also calcu-
lated to check the deviation from average acceleration response. The seismic stations
with the same geological site are grouped together. The grouped seismic stations
normalized acceleration response spectra are then undergoing weighted average to
get common normalized acceleration response spectra for specific geological site.
The final obtained acceleration response spectra from the geometric mean of two
horizontal components and vertical component are also compared with the BIS code
for both hard and soft sites.
Table 1 Different accelerometers stations with site geology, number of recorded earthquakes at
each station and recorded magnitude range
Stations Longitude Latitude Geology Site condition Number of Magnitude
earthquake
recorded
ROH 76.593 28.896 Middle to late Soft 3 3.4–4.1
SON 77.01 29.00 pleistocene Soft 1 4.9
JAFR 76.908 28.594 Soft 3 4.2–4.9
NSIT 77.041 28.608 Soft 2 4.2–4.9
GUR 77.029 28.449 Soft 2 4.3–4.9
LDR 77.217 28.583 Soft 1 4.3
RGD 77.123 28.665 Soft 2 3.3–4.9
NTPC 77.304 28.508 Soft 2 3.3–4.9
DU 77.212 28.691 Soft 2 3.3–4.9
ZAKI 77.23 28.64 Soft 2 4.2–4.9
GGI 77.232 28.665 Soft 2 3.3–4.9
PAL 77.33 28.13 Soft 1 4.3
NOI 77.479 28.507 Soft 2 4.2–4.9
IIT 77.192 28.545 Soft 2 3.3–4.9
IMD 77.22 28.588 Soft 3 3.3–4.9
FAR 77.325 28.383 Soft 1 4.2
BAL 77.321 28.342 Soft 1 4.3
BAR 77.255 29.097 Soft 1 4.9
HGR 77.321 28.342 Soft 1 3.3
REW 76.61 28.184 Holocene Soft 1 4.3
MAC 77.295 28.603 Soft 1 4.3
ALIP 77.141 28.795 Soft 3 3.3–4.9
DCE 77.118 28.795 Soft 2 3.3–4.9
JNU 77.165 28.542 Proterozoic Hard 2 3.3–4.9
ANC 77.264 28.539 Hard 2 4.2–4.9
NDI 77.217 28.683 Hard 1 4.3
DJB 77.188 28.652 Hard 2 4.2–4.9
In the present study, the strong ground motion characteristics of Delhi and national
capital region are evaluated using 159 accelerographs recorded at well azimuthally
covered seismic sites located in Delhi, Haryana and Uttar Pradesh states. The seismic
stations are grouped into three categories according to geological aspects of region as
582 Abhishek et al.
Holocene group, middle-late Pleistocene group and Proterozoic group and acceler-
ation response spectra are estimated for each group. Peak ground acceleration, peak
ground velocity and peak ground displacement are evaluated associated with earth-
quake of magnitude 4.9 occurred on 05 march 2012 in Delhi-Haryana boarded region.
It is observed that associated peak ground acceleration (PGA) varies from 2.4 to
30.8 cm/s2 . Peak ground velocity (PGV) varies from 0.07 to 0.88 cm/s, whereas peak
ground distance (PGD) varies from 0.011 to 0.068 cm. Maximum PGA (30.8 cm/s2 ) is
observed at JAFR station which is located 32 km from the epicentre, while minimum
PGA (2.6 cm/s2 ) is observed at ridge observatory (IMD) which is 62 km from the
earthquake epicentre. It is well known fact that peak ground acceleration is function
of magnitude and distance. It follows the law of geometrical spreading which indi-
cate the attenuation of seismic wave as the distance increase between epicentre and
seismic station. In our study, we observed no such relation among many stations.
The stations like SON and JAFR which lies close to Mahendragarh-Dehradun fault
and Delhi-Haridwar ridge experienced maximum ground acceleration, whereas the
station lies on far away from Delhi-Haridwar ridge and opposite side of Sohna fault
experienced low ground motion. We can interpretate this as, the Delhi-Haridwar
ridge and Sohna fault act as barrier for propagation of seismic wave which might be
controlling factor of PGA, PGV and PGD. It is also fact that most of these stations lie
on soft alluvium deposits and experienced high ground acceleration as compare to the
stations with hard deposits which is in good agreement with general seismological
observation.
Sonipat station observed second highest PGA (26.6 cm/s2 ), maximum PGV (0.88
cm/s) and PGD (0.068 cm) as shown in Fig. 2a, b, c. The location of this station is
along the Mahendragarh-Dehradun fault (MDF) and Delhi-Haridwar ridge (DHR).
This can be interpreted as rupture directivity effect of the earthquake as MDF and
DHR may be controlling factor of ground motion or act as barrier for propagation of
strong ground motion beyond them. Bansal and Verma [12] observed the minimum
PAG (2.50 cm/s2 ) at ridge observatory station and maximum PGA (39.4 cm/s2 ) at
Jaffarpur station on transverse component. They observed the site amplification was
not uniform and varies in 3 to 6 range for various sites. Babita Sharma et al. [9]
also observed the rupture effect of 25 April 2015 Nepal earthquake and suggested
the sub-surface material heterogeneity, seismotectonic feature and rupture directivity
affect the variation of strong ground motion parameters.
The effect of local geology on response spectra is also studied in the present work.
The seismic stations are grouped according to their local geology via Holocene age
group (4 stations), middle to late Pleistocene age group (19 stations) and Protero-
zoic age group (4 stations) as given in the Table 1. The normalized acceleration
response spectra (NARS) and weighted average normalized acceleration response
spectra for Holocene age group are shown in Fig. 3a. The average acceleration r–
esponse spectra show maximum spectral amplification of 2.968 at 0.045 s on vertical
component whereas maximum spectral amplification of 2.690 at 0.054 s on horizontal
component. It is also observed that at short period up to 0.098 s, the vertical accel-
eration spectra are higher than horizontal acceleration response spectra, while for
period greater than 0.098 s, it follows the reverse order (horizontal spectra are high
Fig. 2 Distribution of a peak ground acceleration in cm/s2 , b peak ground velocity in cm/s and c
peak ground displacement in cm associated with 5 March 2012 (M: 4.9) earthquake
as compares to vertical spectra). The similar kind of observation is found in middle

to late Pleistocene age group. In this maximum spectral amplification is 2.760 at
0.085 s on vertical component and maximum spectral amplification 2.460 at 0.109 s
is observed at horizontal component as shown in Fig. 3b. The vertical acceleration
response spectra are high as compare to horizontal acceleration response spectra in
the short period less than 0.1153 s. In case of Proterozoic age group Fig. 3c, the
peak spectral amplification on vertical component is 3.28 at 0.060 s, whereas the
peak spectral amplification on horizontal component is 3.54 at 0.068 s. From Fig. 3c,
it is observed that vertical acceleration response spectra are small as compare to
horizontal response spectra in period range 0.055–0.283 s.
We also compared the weighted average normalized acceleration response spectra
for Holocene, Pleistocene and Proterozoic age groups with the acceleration response
spectra for soil and rock site suggest by Bureau of Indian standard code for entire
country as shown in Fig. 3a–c, respectively. We observed the high amplification on
vertical component as compare to horizontal component in Holocene and Pleistocene
age group. On other hand, maximum amplification is observed in horizontal compo-
nent on maximum period as compare to vertical component which is unison to the
584 Abhishek et al.
Fig. 3 Normalized acceleration response spectra for vertical component as well as for geometric
mean of horizontal components and its comparison with Bureau of Standard Indian code 2002
for a Holocene age formation, b middle to late Pleistocene age formation and c Proterozoic age
formation. (Light grey and red colour lines are acceleration response spectra of individual station)
standard norms. It is observed that at period less than 0.09 s the spectral amplifica-
tion is overestimated when compared with Indian standard code whereas for period
greater than 0.2, second the spectral amplification is underestimated than Indian stan-
dard code for both hard and soft rock for Holocene age group as shown in Fig. 3a.
In case of Pleistocene age group, the SA is higher than the Indian standard code for
period less than 0.08 s and lesser for period greater than 0.15 s as shown in Fig. 3b.
A similar kind of observation in Proterozoic age group is shown in Fig. 3c. Hayashi
et al. [8] estimated the average horizontal response spectra for various soil conditions
in Japan and concluded that highest SA is 2.6, 2.5, 3.3 in loose soil, intermediate soil
and stiff soil respectively. Mohraz [13] studied the earthquake response spectra for
different geological conditions California region. He found the average SA is 2.6 on
vertical component and 2.1 on horizontal component for rocky site and 2.3 on vertical
and 2.1 on horizontal component for alluvium site. In Indian scenario, Raghuknath
and Iyenger [14] estimate the seismic spectral acceleration for Peninsular India by
matching the SA for different soil site classes as suggested by NEHRP (2001) with
Indian standard code. Chopra and Choudhary [3] and Sharma et al. [9] also esti-
mated the SA for Gujarat region and boarder area of India to Nepal, respectively. In
present study, the vertical acceleration response spectra are comparatively high from
horizontal spectra in case of Holocene and middle to late Pleistocene age group,
whereas in case of Proterozoic group, the horizontal spectra are high as compared to
vertical which is unison with general seismological observation. This can be inter-
preted as, the station on quaternary sediments experienced high vertical motions.
Richter (1958) explained that the station near to faults or epicentre may experience
larger vertical motion as compare to horizontal motion than elsewhere. In the present
study, small to moderate size earthquakes only have been used as the large/major
earthquakes data for the study region are not available. The obtained results may
be modified by adding more recorded earthquake and simulated earthquake data for
the analysis. The present research work will be helpful for construction of important
structures and the risk due to earthquakes can be minimized as it is very important
for Delhi NCR to build up a safer build environment as in recent years the study
region have experienced tremendous demographic growth.
5 Conclusion
The strong ground motion characteristic of Delhi and national capital region are
estimated using well recorded strong ground motion data by local seismic network.
The effect of local geological conditions on spectral acceleration are also studied.
The estimated strong ground motion parameters associated with 5 March 2012 earth-
quake shows that the Peak ground acceleration (PGA) varies from 2.4 to 30.8 cm/s2 ,
peak ground velocity (PGV) varies from 0.07 to 0.88 cm/s and peak ground distance
(PGD) varies from 0.011 to 0.068 cm. Maximum PGA 30.8 cm/s2 and 26.6 cm/s2 is
observed at JAFR and Sonipat station, respectively, while minimum PGA (2.6 cm/s2 )
is observed at ridge observatory (IMD). We observed PGA distribution is uneven with
respect to the epicentral distance. It is also observed that stations which are closed to
Mahendragarh-Dehradun fault (MDF) and Delhi-Haridwar ridge (DHR) are experi-
enced maximum ground motion and these discontinues may acts as barrier for future
propagation of strong ground motions. The maximum amplitude of acceleration
response spectra for Holocene, middle to late Pleistocene and Proterozoic age group
formation shows peak 2.968 at 0.045 s (V), 2.690 at 0.054 s (H); 2.760 at 0.084 s (V),
2.460 at 0.109 s (H); 3.28 at 0.060 s (V), 3.54 at 0.068 s (H), respectively. The normal-
ized acceleration vertical spectra are high as compared to horizontal spectra for the
short period range and vice versa. The average SA for all three group (Holocene;
middle to late Pleistocene; Proterozoic age) overestimating the Indian standard code
for short period and underestimating for longer periods. In the present study, small
586 Abhishek et al.
to moderate size earthquakes only have been used as the large/major earthquakes
data for the study region are not available. The obtained results may be modified
by adding more recorded earthquake and simulated earthquake data for the analysis.
The present analysis is very helpful for the seismic hazard studies in the Delhi NCR
and may be utilised to improve the design code by characterizing the strong motion
scenarios in the region.
References
1. BIS.: Criteria for earthquake resistant design of structures, part I—general provisions and
buildings. Bureau of Indian Standards. IS 1893 [part I] (2002)
2. Shrestha, B.: Vertical ground motions and its effect on engineering structures: a state-of-the-
art review. In: International Seminar on Hazard Management for Sustainable Development.
Kathmandu, Nepal. November 29–30, (2009)
3. Chopra, S., Choudhury, P.: A study of response spectra for different geological conditions in
Gujarat India. Soil Dynam. Earthq. Eng. 31, 1551–1564 (2011)
4. Biot, M.A.: A mechanical analyzer for the prediction of earthquake stresses. Bull. Seismol.
Soc. Am. 31, 151–171 (1941)
5. Biot, M.A.: Analytical and experimental methods in engineering seismology. Proc ASCE 68,
49–69 (1942)
6. Housner, G.W.: An investigation of the effects of earthquakes on buildings. Ph.D. thesis.
California Institute of Technology, Pasadena, California (1941)
7. Housner, G.W.: Behavior of structures during earthquakes. J. Eng. Mech. Div., Proc ASCE
1959 1959; 85(EM4):109–29
8. Hayashi, S., Tsuchida, H., Kurata, E.: Average response spectra for various subsoil conditions.
In: Proceedings of Third Joint Meeting. U.S. Japan Panel on Wind and Seismic Effects. UJNR,
Tokyo (1971).
9. Sharma, B., Chingtham, P., Sharma, V., Kumar, V., Mandal, H.S., Mishra, O.P.: Characteristic
ground motion of 25th April 2015 Nepal earthquake (Mw 7.9) and its implication for the
structural design code for the border areas of India to Nepal. J. Asian Earth Sci. (133), 12–23
(2017)
10. Srivastava, A.K., Sinha, K.K., Dasgupta, G., Jalote, P.M., Srivastava, M.C., Gupta, S.K.: Report
on Geotechnical project, Delhi. Unpub. Geol. Surv. India Report (1980)
11. Mittal, H., Kumar, A., Ramhmachhuani, R.: Indian national strong motion instrumentation
network and site characterization of its Stations. International Journal of Geosciences (3),
(2012).
12. Bansal, B. K., Verma, M.: The M 4.9 Delhi earthquake of 5 March 2012. Current Science
(102), 12–25 (2012)
13. Mohraz, B.A.: study of earthquake response spectra for different geological conditions. Bull.
Seismol. Soc. Am. 66, 915–935 (1976)
14. Raghukanth, S.T.G., Iyengar, R.N.: Estimation of seismic spectral acceleration in peninsular
India. J. Earth Syst. Sci. 116, 199–214 (2007)
Detection of Liquefaction Phenomenon
from the 2015 Nuweiba Earthquake
Using Remote Sensing Data
Hrik Chaudhury, Abhishek Kumar , and Rishikesh Bharti
Abstract Egypt is a country situated in the northeastern part of the African conti-
nent. It has great historical importance, making this country’s economy depend on
tourism. Egypt faced significant earthquakes throughout history. One of the major
earthquakes occurred in 1995, with a magnitude of 7.8. It left its marks on Nuweiba
city, located at the banks of the Gulf of Aqaba, in the form of liquefaction. This
disaster called for much economic loss in the country. Hence, it is essential to know
that the area is still undergoing liquefaction. The estimation of liquefaction-related
damages is possible using conventional field investigation approaches. However,
conducting field research is difficult due to several factors, including cost, tool limi-
tations, and accessibility issues. In this study, an attempt is made to detect water
content changes which is a manifestation of soil liquefaction, using the 2015 Nuweiba
earthquake of moment magnitude 5.5 in the vicinity of the epicenter. Band ratios and
a statistical technique have been used to analyze pre- and post-earthquake optical
satellite imageries to identify changes in water content. The algorithm has shown
a change throughout the area of the study. Moreover, all the techniques produce
a similar pattern for the immediate vicinity of Nuweiba city, proving the need for
detailed geotechnical discussion.
Keywords Liquefaction · Nuweiba earthquake · Remote sensing
1 Introduction
Egypt has a history of earthquake occurrences. According to reports, thousands of

earthquake incidents have occurred since 2200 BC [1]. USGS earthquake catalog
reports 12 earthquakes having a moment magnitude greater than 5 in the Gulf of
Aqaba region since 1900 AD [2]. Nuweiba city in Egypt experienced a major earth-
quake of moment magnitude 7.2 on November 22, 1995, in the Gulf of Aqaba [3].
H. Chaudhury (B) · A. Kumar · R. Bharti

Department of Civil Engineering, IIT Guwahati, Guwahati 781039, India
e-mail: c.hrik@iitg.ac.in
https://doi.org/10.1007/978-981-99-1459-3_47
588 H. Chaudhury et al.
Fig. 1 Nuweiba earthquake of moment magnitude 5.5 (USGS earthquake catalog) [2]
This 1995 Gulf of Aqaba earthquake had left its mark in the vicinity through the
ground and structural failures. Liquefaction caused ground deformation along the
coastline near Nuweiba city, especially in the hotel areas. Ground fissures even as
wide as 10 cm were reported [4]. These failures caused major economic drawbacks
in the tourist area along the Gulf of Aqaba in the vicinity of Nuweiba city. To ensure
such financial loss does not hinder the country’s growth, it is important to know if
Nuweiba city and its vicinity still tend to undergo such damage.
According to the USGS earthquake catalog [2], a significant earthquake having
moment magnitude 5.5 with a focal depth of 22 km and coordinates of 29.040° N,
34.667° E occurred on Saturday, June 27, 2015, at 15:34:03 UTC in the Gulf of Aqaba
along the Dead Sea fault zone near Nuweiba city [2] (see Fig. 1). This paper aims to
find possible liquefied zones due to the 2015 Nuweiba earthquake by detecting soil
moisture change through geospatial techniques.
2 Study Area
Nuweiba city is located between the coast of the 180 km long Gulf of Aqaba (see
Fig. 2) and the Sinai Peninsula. The Dead Sea fault system, which links the spreading
of the Red Sea seabed with the development of continental rifts, is represented by the
Gulf of Aqaba as its southern end. The Gulf of Aqaba is seismically active because of
the Dead Sea Transform fault. The gulf formed when the Red Sea started rifting into
the land. That displacement is maintained to date, forcing a transverse component of
the Dead Sea Transform [5].
Detection of Liquefaction Phenomenon from the 2015 Nuweiba … 589
Fig. 2 False color composite of the narrowed down study area having a buffer zone of 11 kms
around the epicenter shown by the white hollow circle
3 Dataset and Methodology
According to Papadopoulos (1993) [6], an earthquake of a certain magnitude has

an influence area in terms of liquefaction. The empirical relation presented by
Papadopoulos (1993) [6] is given in Eq. (1). Equation (1) shows the buffer zone
for the 2015 Nuweiba earthquake to be 11 km. As the pre- and post-earthquake
LANDSAT 8 images for the analysis covers a very large area compared to 11 km
square, the study area had to be narrowed down, given in Fig. 2. Moreover, further
closer inspection through Google Earth images reveals that most of our selected
study area is covered in hilly terrain and has no permanent settlement. This narrows
the study area to the coastal region in the vicinity of Nuweiba city (Fig. 3), which
has significance in terms of population and economic aspects.
Mw = −0.44 + 3 × 10−8 × Re + 0.98 × log Re (1)
where
Mw is the moment magnitude of the earthquake.

Re is the epicentral distance from the earthquake.
LANDSAT satellite has a temporal resolution of 16 days and acquires the image
from a height of 705 km. The main goal of this research is to use pre- and post-
earthquake imageries to document moisture changes that occurred in the study area
following the Nuweiba 2015 earthquake. For this purpose, LANDSAT 8 Operational
Land Imager (OLI) data from June 24, 2015, and July 10, 2015, is used. The pre-
Fig. 3 Selected study area

from Google Earth images
and post-earthquake images were acquired at the same time on their respective days.
Moreover, sun elevation and sun azimuth also do not differ significantly. Because
the spectral reflectance values of soil and vegetation change with water content, the
spectral reflectance of liquefied and non-liquefied surfaces will vary. On LANDSAT
8, spectral bands help identify changes in reflectance associated with the moisture
content of the surface. On the other hand, weather factors may affect the surface’s
water content and result in variations. The study area is one of the driest places on
Earth.
Reports from the Center for Hydrometeorology and Remote Sensing show that
there was no rainfall between the period of pre- and post-earthquake image acquisi-
tions [7]. Therefore, any moisture content change should result from the earthquake.
Earlier studies have attempted to employ a geospatial approach to identify changes in
soil moisture, like Ramakrishnan et al. [8], Oomen et al. [9], Baik et al. [10], Chaud-
hury et al. [11] and Mondal and Bharti [12]. The change in moisture is detected in
this work using a variety of indices. These indices contain bands from LANDSAT
8 that are sensitive to the surface’s moisture content. Three types of Normalized
Differential Water Index (NDWI) [13] (NDWI (green,NIR) , NDWI (green, SWIR1) , NDWI
(green, SWIR2) ) and [12] Temporal Differential Liquefaction Index (TDLI) are used for
the analysis. As shown in Eq. (2), NDWI (green, NIR) employs the green band, which is
the 3rd band of LANDSAT 8, and the Near Infrared (NIR) band, which is the 4th band
of LANDSAT 8. NDWI (green, SWIR1) uses the green band (3rd band of LANDSAT 8)
having a wavelength of 550 nm and Short Wave Infrared (SWIR 1) band, which is the
6th band of LANDSAT 8 shown in Eq. (3). NDWI (green, SWIR2) uses the green band
and SWIR 2 band, which is the 7th band of LANDSAT 8 shown in Eq. (4). Also,
TDLI was computed as shown in Eq. (5). The image difference (Indexpre-earthquake and
Indexpost-earthquake ) was calculated using LANDSAT 8 images for the two time periods
before and after the earthquake shown in Eq. (6). An increase in water content after
the earthquake is shown by a positive value, whereas a negative value indicates a
decrease.
Green550 nm − NIR860 nm
NDWI(green,NIR) = (2)
Green550 nm + NIR860 nm
Green550 nm − SWIR1600 nm
NDWIgreen, SWIR1 = (3)
Green550 nm + SWIR1600 nm
Green550 nm − SWIR2200 nm
NDWIgreen, SWIR2 = (4)
Green550 nm + SWIR2200 nm
Pre_earthquake_SWIR2200 nm − Post_earthquake_SWIR2200 nm
TDLI = (5)
Pre_earthquake_SWIR2200 nm + Post_earthquake_SWIR2200 nm
Index = IndexPost_earthquake − IndexPreearthquake (6)
The tasseled cap transformation coefficients developed by Huang et al. [14] for
LANDSAT ETM + reflectance were used. Based on numerous field observations of
soil, impervious surfaces, dense vegetation and moisture content, Huang et al. [14]
created this transformation. The field data were used to determine the rotation of the
principal axes received from principal component analysis (PCA) while maintaining
the orthogonality of the primary axes collected from PCA. Tasseled cap and PCA
differ because the GramSchmidt technique enables the user to choose the calculation
order based on a physical image interpretation. In contrast, PCA organizes the prin-
cipal directions in the data a priori. Using the LANDSAT ETM + image tasseled cap
transform wetness axes, the surface wetness/moisture content difference between
pre- and post-earthquake coverages is assessed.
The images show the change in the NDWI before and after the earthquake that
occurred on June 27, 2015. A more significant difference in the index value indi-
cates an increase in the water content (red and yellow pixels). Tasseled cap wetness
resembles (see Fig. 8) the exact pattern of NDWI (green, NIR) (see Fig. 4). The water
content has increased throughout the majority of the study area, as can be observed
in the image difference of NDWI (green, NIR) (see Fig. 4) and tasseled cap wetness
(see Fig. 8). Tasseled cap wetness gives more moisture change than NDWI (green, NIR).
NDWI (Green, SWIR1) (see Fig. 5), NDWI (Green, SWIR2) (see Fig. 6) and TDLI (See Fig. 7)
are showing more or less similar patterns for the study area with NDWI (Green, SWIR1)
showing more moisture change followed by NDWI (Green, SWIR2) and TDLI. Three of
Fig. 4 NDWI (Green, NIR) pre- and post-earthquake image differences of study area
the NDWIs and TDLI and tasseled wetness image differences give the same pattern
for the pixels in the south near Nuweiba city (Fig. 8).
.
5 Conclusion
The results obtained from NDWI (Green, NIR) and wetness derived through tasseled cap
transformation corroborate well for the moisture change distribution in the study area.
Fig. 5 NDWI (green, SWIR1) pre- and post-earthquake image differences of study area
On the other hand, NDWI (Green, SWIR1) , NDWI (Green, SWIR2) and TDLI produce the
same distribution pattern. Enhanced moisture content is observed near the southern
region of Nuweiba city, as confirmed by all the geospatial techniques. According
to Abuzied et al. [15], the groundwater potential in the vicinity of Nuweiba city is
moderate to good. Therefore, though Nuweiba is a desert city, it has the scope of
increased moisture content due to earthquakes because of the groundwater potential.
Hence, further geotechnical study should be conducted to understand the liquefaction
vulnerability of the area, and mitigation needs to be carried out.
Fig. 6 NDWI (green, SWIR2) pre- and post-earthquake image differences of study area
Fig. 7 TDLI pre- and post-earthquake image difference of study area

Fig. 8 Tasseled cap wetness pre- and post-earthquake image differences of study area
References
1. Sawires, R., Peláez, J.A., Fat-Helbary, R.E., Ibrahim, H.A.: An earthquake catalogue (2200 BC
to 2013) for seismotectonic and seismic hazard assessment studies in Egypt. In Earthquakes
and Their Impact on Society, pp. 97–136. Springer, Cham (2016)
2. USGS Earthquake Catalogue
3. Baer, G., Sandwell, D., Williams, S., Bock, Y., Shamir, G.: Coseismic deformation associated
with the November 1995, Mw= 7.1 Nuweiba earthquake, Gulf of Elat (Aqaba), detected by
synthetic aperture radar interferometry. J. Geophy. Res. Solid Earth 104(B11), 25221–25232
(1999)
4. Abdel-Halim, M.A., Al-Tarazi, B.: Structural and geotechnical aspects of the 1995 Gulf of
Aqaba earthquake. Doboku Gakkai Ronbunshu 2004(759), 15–23 (2004)
5. Hamouda, A.Z.: Recent evaluation of the assessment seismic hazards for Nuweiba, Gulf of
Aqaba. Arab. J. Geosci. 4(5), 775–783 (2011)
6. Papadopoulos, G.A., Lefkopoulos, G.: Magnitude-distance relations for liquefaction in soil
from earthquakes. Bull. Seismol. Soc. Am. 83(3), 925–938 (1993)
7. CHRS data portal
8. Ramakrishnan, D., et al.: Mapping the liquefaction induced soil moisture changes using remote
sensing technique: an attempt to map the earthquake induced liquefaction around Bhuj, Gujarat,
India. Geotech. Geol. Eng. 24(6), 1581–1602 (2006)
9. Oommen, T. et al.: Documenting earthquake-induced liquefaction using satellite remote sensing
image transformations. Environ. Eng. GeoSci. 19(4), 303–318
10. Baik, H., Son, Y.-S., Kim, K.-E.: Detection of liquefaction phenomena from the 2017 Pohang
(Korea) earthquake using remote sensing data. Remote Sensing 11(18), 2184 (2019)
11. Chaudhury, H., Kumar, A., Bharti, R.: Detection of liquefaction phenomenon from the 2017
Tripura earthquake from remote sensing data. In: 2022 IEEE International Geoscience and
Remote Sensing Symposium IGARSS. IEEE (2022)
12. Mondal, S.K., Bharti, R.: Seismic impact around himalayan snow-melt fed lake using InSAR:
a case study for 20 March 2020 MW5.7 Tibet Earthquake. In: 2022 IEEE International
Geoscience and Remote Sensing Symposium IGARSS. IEEE (2022)
13. Özelkan, E.: Water body detection analysis using NDWI indices derived from landsat-8 OLI.
Pol. J. Environ. Stud. 29(2), 1759–1769 (2020)
14. Huang, C., Wylie, B., Yang, L, Homer, C., Zylstra, G.: Derivation of a tasselled cap
transformation based on landsat 7 at satellite reflectance: Int. J. Remote Sens. 23(8), 1741–1748
15. Abuzied, S.M., Yuan, M., Ibrahim, S.K., Kaiser, M.F., Seleem, T.A.: Delineation of ground-
water potential zones in Nuweiba area (Egypt) using remote sensing and GIS techniques. Int.
J. Signal Process. Syst. 4(2), 109–117 (2016)
Estimation of Site Amplification Factor
and Predominant Frequency
in and Around Panchkula City, Haryana,
India
M. Sandhu, R. B. S. Yadav, D. Kumar, and Abhishek
Abstract In this study, we used the horizontal-to-vertical spectral ratio (HVSR)

technique to estimate the site amplification factor and the fundamental/predominant
frequency at some of the sites in the Panchkula city of Haryana state of India. This
technique gained popularity when Nakamura [1] estimated the site amplification
factor by considering the horizontal-to-vertical spectral ratio of microtremor noise
data. This method helps in the seismic microzonation of an area with very less
seismicity. It has been found that the site amplification factor and the fundamental
frequencies at 19 sites in the city vary in the range of 1.9–3.1 and 0.4–1.0 Hz,
respectively, depending upon the local geological conditions. The sites associated
with the riverside show more site amplifications (> 2.8), while sites away from the
river show less amplifications (< 2.1). Similarly, we observed that some sites show
higher predominant frequencies (> 0.85), while some sites show less predominant
frequencies (< 0.5). It has been also observed that some sites show clear H/V peaks,
while some sites show broad peaks due to poor impedance contrast between bedrock
and soil. The variation of site amplification factor and predominant frequency in
the study region exhibits variable geological conditions and site characteristics. This
study is highly useful for civil engineers in designing and constructing high-rise
buildings in this region.
Keywords Site amplification · Predominant frequency · Microzonation ·

Horizontal-to-vertical spectral ratio (HVSR)
M. Sandhu · R. B. S. Yadav (B) · D. Kumar · Abhishek

e-mail: rbsykuk@gmail.com
Abhishek
National Centre for Seismology, Ministry of Earth Sciences, New Delhi, India
https://doi.org/10.1007/978-981-99-1459-3_48
600 M. Sandhu et al.
1 Introduction
It has now been well understood that the earthquake ground motion characteristics
like amplitudes, frequency contents and duration are influenced significantly by the
local site conditions. This further affects the degree and extent of damage patterns
during an earthquake. The ground vibrations due to the same earthquake at one site
have been observed to be stronger than those of another site at a similar distance.
The topographic and local geological conditions termed as site effects are the reasons
for such variations of earthquake ground motions. This depends on the softness of
the rocks as well as the thickness of the sediments above the hard bedrock. The
phenomenon of site effect can be explained by the lower velocity and density of
unconsolidated sedimentary layers as compared to underlying hard bedrocks. The
parameters like intensity and the angle of incidence of the upcoming seismic waves
also play a vital role in modifying the ground motion and might introduce a nonlinear
phenomenon [2] in combination with the local site effect. Site effect can be defined
as the modifications by the surficial layers of soil formations and topography to the
features (amplitude, frequency content and duration) of the seismic waves coming
to the surface which results in the amplification or deamplification of the ground
motion. The amplification factor and resonant frequency are generally measured as
the qualitative and quantitative parameters of the site effects.
The severe consequences of site effects have been observed during the 1985
Michoacan Mexico earthquake, 1988 Spitak, 1989 Lomaprieta and 2001 Bhuj earth-
quakes. It has been shown that a small, localized variation of the thickness of the very
soft clay could originate in very significant differences in the surface ground motions
over short distances, such as those observed during the 1985 Mexico earthquake [3].
It has been suggested that during the 1988 Spitak earthquake, the local site effects
were a major factor in contributing to increased damage levels in the city located on
a broad alluvial basin [3]. During the 1989 Loma Prieta earthquake, the combination
of liquefaction and site amplification was the cause of the extensive damage in the
Marina District of San Francisco [4]. Similarly, during the 2001 Bhuj earthquake, the
site amplification was identified as the cause of heavy damage in Ahmedabad city
which is 300 km away from the source zone [5]. The structures located on sediments
suffered more damage during the 1997–1998 Umbria-Marche seismic sequence than
those of similar type located within a few tens of meters on a rock site [6]. Thus,
for determining the site-specific seismic hazard, the site amplification function plays
a vital role and will be helpful for microzonation studies and simulation of strong
ground motion at the surface of a region. In this study, the site amplification func-
tions have been estimated at different sites in and around Panchkula city. Panchkula
is a district of the state of Haryana, which lies in the foothills of the Himalayas or
Shiwaliks along the main Frontal thrust (MFT) of the Himalayas. The geological
and seismotectonic setting of the area makes it more vulnerable to the occurrences
of earthquakes or other hazards in the region. Being in the vicinity of the capital
of Haryana, Chandigarh and the administrative hub of Haryana, it becomes more
important to assess the site characteristics for the region. Determination of ground
Estimation of Site Amplification Factor and Predominant Frequency … 601
Fig. 1 Geological map of the study area (https://www.sciencedirect.com/science/article/abs/pii/

S2352801X1830081X)
structure is important for seismic hazard analysis and earthquake design of structures
(Fig. 1).
2 Geology and Seismotetonics of the Study Area
Panchkula is the district of Haryana, which is located near the capital of Haryana
(Chandigarh). Panchkula is located approximately between 76°51, East longitude
and 30°41, North latitude. The study area, according to the geological setting, is
located in the Himalayan foothills, i.e., Himalayan Frontal Thrust (HFT) or MFT
or Shivalik Range and Sindhu Plain (Indo-Gangetic Plain) which lies in the seismic
zone IV. Morni hills are the highest point in this area, whose elevation is 337 m above
mean sea level. Due to geomorphological features, the study area has accumulated
large boulders and pebbles. Alluvial plain deposits due to Ghaggar and Sirsa River,
which lies on the Sindhu plain in Punjab. The district lies in the Himalayas boundary
fault zones near to Main boundary thrust (MBT) and Main Central Thrust (MCT),
where earthquakes of moderate (VII) to high (X) intensity have occurred in the past.
Now, the principal tectonic displacement zone in the region is the Himalayan Frontal
Fault system situated at the edge of the Indo-Gangetic plains [7, 8].
3 Methodology
Site amplification and predominant frequency at any site are the key parameters in
designing the structure of a building, which can sustain during an earthquake due to
the seismic ground motions that are amplified in the presence of loose soil sediments
or deposits. The degree of shaking due to the occurrence of an earthquake depends
upon the local geological conditions. There are various methods to evaluate the
site effects based on theoretical and empirical approaches. The theoretical approach
requires geotechnical characteristics of the site such as soil density, wave velocity,
thickness and damping of sediments. The empirical approaches are different from
theoretical approaches as they are based on empirical data. Most of these techniques
are based on analyzing data in the frequency domain. The site amplification can be
measured in two ways: one by using the earthquake waveform data and the other
by using the ambient noise recording data. The influence of source and path effects
has to be removed in order to estimate site effects precisely. These methods can be
separated into two categories: one that uses a reference site (particular location with
no site effects) and the other that uses only one site. These include (i) the standard
spectral ratio (SSR) technique, (ii) generalized inversion scheme (GIS), (iii) the coda
wave technique and (iv) the horizontal-to-vertical spectral ratio (HVSR) technique
[9]. In the present study, we have used the HVSR technique using the ambient noise
data measured by TROMINO 3G as it does not require the reference site for its
estimation.
The horizontal-to-vertical (H/V) spectral ratio method was first presented by
Nakamura [1] to determine site amplification using ambient noise. This method
assumes that the ground motions in the horizontal component at a site are amplified
around the fundamental frequency of a site, while there is no amplification in the
vertical component of ground motion. The H/V ratio technique assumes a half-space
with a horizontal sedimentary layer on top. The vertical component of the transfer
function, hv (f), is given as the ratio between the vertical component of motion on the
surface, Vs(f), and that on the base of the sediments, V B (f), in the frequency domain
[10]:
Hv ( f ) = VS ( f )/VB ( f ) = SV ( f )P( f )Z V ( f )/SV ( f )P( f ) (1)
where Sv (f) represents the source effect, P(f) represents the path effect from the
source to the base sediments and Zv (f) represents the amplification caused by the
site sediments [9]. Similarly, the horizontal component of the transfer function Hu
(f) can be defined as
HU ( f ) = H s( f )/H B ( f ) = Su( f )Z u( f )P( f )/Su( f )P( f ) (2)
where Hs(f) and H B (f) are the horizontal components of the motion at the surface
and the bedrock. Now from Eqs. (1) and (2), the ratio can be given as:
Fig. 2 Estimated H/V curve at the site GHUMTHALA
HU ( f )/HV ( f ) = Z U ( f )/Z V ( f ) = HS ( f )VB ( f )/H B ( f )V s( f ) (3)
Now, Zv (f) = 1 as it is assumed that the vertical component of the ground motion
is free from any influence of the sediments at the site. It can also be assumed that the
spectral amplitude of both the components (H B (f) and V B (f )) at the bedrock, i.e.,
at the base of the sediments is the same and can be given as [10].
HB ( f ) = VB ( f ) (4)
From Eq. (3), the horizontal transfer function can be given as:
Z u( f ) = H s( f )/V s( f ) (5)
The site amplification function at the given site can thus be obtained using Eq. (5).
The Hamming and Tuckey Technique [11] is used for smoothing the H/V ratio
obtained from the above process and will appear as shown in Fig. 2.
The site response characteristics using the ambient noise H/V spectral ratio have been
investigated in this study. Several measurements were made using the Tromino 3G
instrument in the nearby areas of the Panchkula district of Haryana (Northern India).
The frequency range from 0.2 to 20 Hz has been taken for the data analysis in the
study region because most of the engineering structures have a resonant frequency of
less than 10 Hz and hence the selected range will take care of all such components. In
this work, for each site, the recording system operated continuously for 20 min with a
sample rate of 128 Hz. We have acquired data at the 19 different sites in the Panchkula
region with a maximum offset of about 2 km between each successive reading. The
GRILLA software is used to archive, organize, view and analyze the recordings of
TROMINO 3G. Figure 3a–c shows the obtained H/V curve along with the Fourier
spectra of all three recorded components of the ambient noise with the module of
noise filtering. Figure 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14 and 15 shows the estimated site
amplification functions with the predominant frequencies at the different sites. Table
1 gives the estimated predominant frequency and the corresponding site amplification
functions at all 19 sites.
Fig. 3 a Estimated spectral H/V using the HVSR technique, b fourier spectra of all three compo-
nents of recorded ambient noise, i.e., N-S, E-W, vertical c frequency-time graph to eliminate
unwanted frequency
Fig. 4 Estimated spectral H/V and the predominant frequency at Manka ITBP
Fig. 5 Estimated spectral H/V and the predominant frequency at Billa village
Fig. 6 Estimated spectral H/V and the predominant frequency at Panchkula extension
We note that the site amplification function lies in the range from 1.9 to 3.2 at
the predominant frequency. The estimated site amplification function in the region is
found to be consistent with the geology of the area. We have obtained the predominant
frequency in the range of 0.42 ± 0.28 to 0.91 ± 0.17 Hz. The estimated predominant
frequency is less than 1.0 Hz, which reflects the case of a deeper interface with the
two-layer scenario. Furthermore, with more detailed surveying one can estimate the
depth and velocity model of the region.
Fig. 7 Estimated spectral H/V and the predominant frequency at Nadda Sahib Gurdwara
Fig. 8 Estimated spectral H/V and the predominant frequency at enclave
Fig. 9 Estimated spectral H/V and the predominant frequency at Chownki
Figure 16a shows the spatial distribution of the estimated predominant frequencies
at various sites. The spatial distribution of the estimated amplification levels corre-
sponding to the predominant frequency (0.42 to 0.91 Hz) which in turn corresponds
to tall buildings to high-rise buildings is shown in Fig. 16b.
Fig. 10 Estimated spectral H/V and the predominant frequency at Birghaghar
Fig. 11 Estimated spectral H/V and the predominant frequency at Lana Billa village
Fig. 12 Estimated spectral H/V and the predominant frequency at Manka village
Fig. 13 Estimated spectral H/V and the predominant frequency at Paras hospital
Fig. 14 Estimated spectral H/V and the predominant frequency at sector 21
Fig. 15 Estimated spectral H/V and the predominant frequency at Upper Nada
Table 1 Estimated parameters at 19 different stations

S. No Site name Latitude (°N) Longitude (°E) Site Predominant
amplification frequency (Hz)
1 Mhtctay 30.3915 76.5325 2.6 0.91 ± 0.17
2 Mnkafct 30.3850 76.5343 2.1 0.42 ± 0.28
3 Biriaghr 30.2180 76.5247 2.0 0.50 ± 0.08
4 Gmthala 30.4350 76.5345 2.5 0.44 ± 0.08
5 Chownki 30.4246 76.5345 2.8 0.44 ± 0.07
6 Prshspt 30.4123 76.5254 2.0 0.94 ± 0.33
7 Sec-21 30.4015 76.5147 1.9 0.53 ± 0.14
8 Enclave 30.3915 76.5325 2.6 0.91 ± 0.09
9 Nrgudra 30.4142 76.5243 2.5 0.50 ± 0.34
10 Manka Itbp 30.634 76.9066 2.5 0.84 ± 0.06
11 Billa village 30.649 76.935 2.4 0.41 ± 0.30
12 Lana Billa 30.628 76.9225 2.2 0.84 ± 0.17
village
13 Upper 30.6925 76.8972 2.6 0.50 ± 0.27
Nadda
14 Panchkula 30.6451 76.8791 3.2 0.81 ± 0.06
extension
15 Chandi 30.727 76.8927 2.2 0.50 ± 0.90
Mandir
16 Jhiriwala 30.673 76.8830 2.8 0.53 ± 0.20
17 Manka 30.635 76.9041 2.3 0.84 ± 0.05
village
18 Chiken 30.821 77.006 2.4 0.68 ± 0.08
village
19 HMT 30.780 76.910 2.5 0.89 ± 0.12
5 Conclusions
The site amplification level at 19 different sites in and around the Panchkula region
has been computed in this study. The horizontal-to-vertical spectral ratio technique
has been used for this purpose. The average value of the site amplifications ranges
from 1.9 to 3.2 for the predominant frequency, which lies in the range of 0.42 to
0.91 Hz. The spatial distribution of the predominant frequencies and the site ampli-
fication levels at a predominant frequency corresponding to the natural frequency of
the tall and high-rise story buildings have been shown. These maps are useful for
the evaluation of seismic hazards to the different story buildings. The site ampli-
fication functions estimated in this study may be used for the simulation of the
(a) (b)
Fig. 16 Spatial distribution of a estimated predominant frequencies and b site amplification

function functions corresponding to high-rise buildings at all the sites
site-specific ground motions in the region as well as for urban development plan-
ning. The estimated values in this study will also be useful for civil engineers in
developing earthquake-resilient infrastructure in the region.
Acknowledgements The authors are thankful to the reviewers for positive and constructive sugges-
tions that improved the quality of the manuscript. The authors are also thankful to RUSA 2.0 for
financial support.
References
1. Nakamura, Y.: A Method for dynamic characteristics estimations of subsurface using

microtremors on the ground surface. QR RTRI 30, 25–33 (1989)
2. Pitilakis, K.: Site effects (recent advances in earthquake geotechnical engineering and
microzonation). A. Ansal. Dordrecht, Kluwer Academic Publishers 1:139–197 (2004)
3. Papageorgiou, A.S., Kim, J.: Study of the propagation and amplification of seismic waves in
Caracas valley with reference to the July 29, 1967 earthquake response: SH Waves. Bull. Seism.
Soc. Am. 81(6), 2214–2233 (1991)
4. Bonilla, L.F., Steidl, J.H., Lindley, G.T., Tumarkin, A.G., Archuleta, R.J.: Site amplification
in the San Fernando valley, California: variability of site-effect estimation using the S-wave,
coda, and H/V methods. Bull. Seism. Soc. Am. 87(3), 710–730 (1997)
5. Mahajan, A.K., Kumar, S., Kamal.: Macroseismic field observations of January 26th, 2001
Kachchh earthquake and its Seismotectonics. J. Asian Earth Sci. 23(1), 17–23 (2004)
6. Rovelli, A., Scognamiglio, L., Marra, F., Caserta, A.: Edge-diffracted 1-sec surface waves
observed in a small-size intramountain basin (Colfiorito, central Italy). Bull. Seism. Soc. Am.
91, 1851–1866 (2001)
7. Nakata, T.: Active faults of the Himalaya of India and Nepal. GSA Special Papers 232, 243–264
(1989)
8. Yeats, R.S., Nakata, T., Farah, A., Fort, M., Mirza, M.A., Pandey, M.R., Stein, R.S.: The
Himalayan frontal fault system, Annales Tectonicae, Special issue, supplement to 6, 85–98
(1992)
9. Sandhu, M., Kumar, D., Teotia, S.S.: Estimation of site amplification functions for the National
Capital (Delhi) Region India. Nat Hazards 85(1), 171–195 (2017)
10. Castro, R.R., Mucciarelli, M.F., Petrungaro, C.: S-wave site response using horizontal to vertical
spectral ratio. Bull. Seismol. Soc. Am. 87, 256–260 (1997)
11. Bath, M.: Spectral Analysis in Geophysics, vol. 563. Elsevier Scientific Publishing Company,
Amsterdam (1974)
Use of GIS for Hypsometric Analysis
for Determining Erosion Proneness
of Mandakini Watershed, Lesser
Himalaya, Uttarakhand, North India
James Xavier Paul and Daya Shanker
Abstract An attempt has been made for the assessment of erosion proneness which
is quite essential in tectonically active, highly fragile and environmentally sensitive
hilly regions. The assessment will not only help in knowing erosion proneness but
also supports in adopting the best practices for integrated watershed management.
The hypsometric analysis was performed to know the geological stages of the devel-
opment of erosional landscapes that reveal the health or condition of a watershed.
The hypsometric integral was estimated from the graphical representation of the
measured contour height and enclosed area, and using the empirical formula. The
study was carried out to assess erosion susceptible areas of the Mandakini River
Watershed, which forms a tributary of the Alaknanda River catchment of Lesser
Himalaya, Uttarakhand. Six sub-watersheds were delineated from the Mandakini
Watershed for performing the hypsometric analysis using contours generated from
the DEM in a GIS environment. The hypsometric integral values were quantified
by the elevation-relief method for all the sub-watersheds and are ranging between
0.46 (B3) and 0.50 (B4). Further, it was found out that almost all the sub-watersheds
are comparatively matured and erosional processes are in the course of stabilization.
The study reveals that the sub-watershed primarily B1 of the Mandakini Watershed
is susceptible to erosion. Moreover, the past earthquakes occurred near by area can
also lead to the instability of the zones. Therefore, suitable remedial measures such
as structural and non-structural methods may be adopted to mitigate soil erosion and
also in enhancing sustainable conservation and management practices.
Keywords Hypsometric analysis · Lesser Himalaya · Erosion proneness ·

Mandakini watershed · GIS
J. X. Paul (B) · D. Shanker

Earthquake Engineering Department, Indian Institute of Technology Roorkee, Roorkee, India
e-mail: jxavierpaul@eq.iitr.ac.in
https://doi.org/10.1007/978-981-99-1459-3_49
614 J. X. Paul and D. Shanker
1 Introduction
Hypsometric analysis helps to understand the various driving factors that influence
the topographical features of a drainage basin. It assesses the data especially related
to water structures at different elevation levels within an area [1]. It is not only used
for evaluating the smaller water bodies but also the larger bodies covering the whole
Earth, and even other terrestrial bodies. Langebein [2] was the first who introduced
that the hypsometric analysis to ascertain the overall slope and relative forms of a
watershed. Hypsometry is a measure of the proportion of area covered at various
elevations within an area of interest. Likewise, the hypsometric curve represents the
spread over the surface of a particular area of its elevation [1]. It is the important
tool to estimate the evolution of geomorphic landforms and to infer the various
driving factors that are acting on landscape structures of a drainage basin (Horton
[3]; Schumn 1956; Markose and Jayappa [4]).
Hypsometry can be explained by a “hypsometric curve” (HC) or as an integral
parameter known as a “hypsometric integral” (Hsi). The HC is related to the overall
volume of soil mass concentration in a drainage basin and the total sum of erosion
taken place in a drainage basin in comparison to the left-over mass [5]. The shape of
the HC forms an important component which is used in measuring the evolutionary
stages of a drainage basin. It is a plot of a continuous function of dimensionless
distribution of relative area of drainage basin [1, 6]. The elevational distribution of an
area has been widely used for topographical assessment since it is more advantageous
in portraying three-dimensional data by a two-dimensional representation. The shape
of the HC for different watersheds with comparably similar hydrologic conditions
may provide information about past mass movement in the watersheds. Thus, the
shape of the HC also describes the past fluctuations in the slope of the concentrated
watershed.
Ritter et al. [7] described that HC and Hsi are the two main components that
capture the state of erosion and health of a watershed. Thus, the intensity of exposure
is indicated by the Hsi value and shape of the HC [8]. Strahler [1] viewed that the HC
shows the phases of the erosion cycle of landforms by studying different catchments,
they were divided into different classes based on the shape of the HC as (i) peneplain
or distorted for a drainage basin in which the curve is concave upward, (ii) S-shaped
curve for the mature basin, where the curve is convex toward the bottom, indicating
a low height and concave toward the top, indicating a high height and (iii) convex
curve upwards for a basin in youth stage. It is noted that the shape of the curves is
constantly changing in the initial stages of the geomorphological cycle and stabilizes
after reaching the stage of maturity. The HC influences the erosion properties of a
catchment and also indicates the erosion cycle. A suitable model for interpreting the
HC is given here [7], (Fig. 1).
Like the HC, the Hsi describes the “cycle of erosion” and the progress in the
geologic stages of the watershed’s development [1]. And, useful for the following
prioritization of achieving sustainable soil and water conservation practices [9, 10].
The Hsi values provide a comprehensive knowledge of the erosion condition and
Use of GIS for Hypsometric Analysis for Determining Erosion … 615
Fig. 1 Conceptual model representing hypsometric analysis and curves (modified after [7])
limits, as proposed by [1] are: (i) in-equilibrium (young) stage, if the Hsi ≥ 0.6, (ii)
equilibrium (mature) stage, if the Hsi is between 0.3 and 0.6, and (iii) monadnock
(old) stage, if the Hsi ≤ 0.3. The Hsi shows an inverse correlation with various
morphometric parameters such as channel gradient, drainage density, total relief
and slope [1]. Thus, Hsi is useful to explain the soil degradation that occurred in
a watershed during the geologic past due to various hydrologic and erosive factors
[11].
Until recently, hypsometry analysis was used in various parts of the world for
different studies to estimate erosion and set priorities for watersheds [11–21].
The hilly watersheds of the Himalayan regions are liable to erosion due to persis-
tent rainfall during the monsoon, pre-monsoon and post-monsoon seasons. In this
context, the HC and Hsi are two important parameters useful for assessing the status
of watersheds in ecologically sensitive and fragile ecosystems such as those of the
Himalayas. During the literature review, it was found that few studies have been
conducted in the Himalayan region using hypsometry to assess the condition of
watersheds [9, 22–24]. However, estimating and collecting data in the rough terrain
of the Himalayan region is very tedious. The advent of new technologies such as the
Geographic Information System (GIS) and remote sensing has facilitated the esti-
mation process with greater precision [10]. Based on the literature study and to fill
a research gap, an attempt was made to conduct a hypsometric analysis in a water-
shed of the Lesser Himalaya that forms a major tributary of an ecologically highly
sensitive area. The study is conducted for the determination of erosion behavior and
health status of the Mandakini watershed and its six sub-watersheds.
2 Study Area
The Mandakini River is a tributary of the Alaknanda River in the Indian state of
Uttarakhand. The river located between the Rudraprayag and Sonprayag areas and
originates from the Chorabari Glacier. The river joins the Songanga at Sonprayag
and flows past the Madhyamaheshwar Hindu temple at Ukhimath. At the end of
its course, it joins the Alaknanda, which flows into the Ganges. The study area lies
between latitudes 30°26, 6,, N to 30°32, 10,, N and longitudes 79°2, 60,, E to 79°7, 40,,
E (Fig. 2). The Mandakini basin lies between 3800 and 6090 m above sea level. The
climate is generally cooler than mainland India, with maximum temperatures of 30°–
35 °C and a minimum of 0°–8 °C. Humidity is relatively high, especially during the
monsoon season. The Mandakini region is seismically and environmentally sensitive,
being located in a collision zone. Geologically, it consists of parts of the crystalline
rock groups of the higher Himalaya and parts of the lesser Himalaya. These rocks are
very susceptible to erosion and are therefore one of the main causes of the numerous
landslides in the region. Sandy skeletal soils, coarse clay, fine clay and mesic clay
soils predominate in the study area. Tectonically, the area is a very active zone, as
evidenced by the presence of several overthrusts and several northeast to southwest
trending faults/lineaments.
3 Materials and Methods
The 30 m resolution Shuttle Radar Topography Mission Digital Elevation Model

(SRTM DEM) was downloaded from the USGS Earth Explorer portal (https://earthe
xplorer.usgs.gov/). And the DEM was used to delineate the natural drainage layer
and boundaries of the six sub-basins along with their stream numbers and stream
orders listed in the figure, using hydrologic tools (Fig. 3). Based on the delineated
sub-watersheds, the DEM for the Mandakini sub-watersheds was created using the
“Extract Mask” tool in the ArcGIS environment (Fig. 4). The slope maps were
prepared for the six sub-watersheds of Mandakini Watershed by using slope tool of
the spatial analyst module in the ArcGIS platform (Fig. 5).
3.1 Hypsometric Curve (HC) Estimation
Hypsometric analysis of six sub-watersheds of the Mandakini River (B1, B2, B3, B4,
B5, and B6) was conducted to determine the different stages of the erosion cycle to
Fig. 2 Study map of the Mandakini watershed with subdivision of its six sub-watersheds
Fig. 3 Stream numbers and stream orders of six sub-watersheds of Mandakini watershed
Fig. 4 Digital elevation model (DEM) of the six sub-watersheds of Mandakini watershed
Fig. 5 Slope map of the six sub-watersheds of Mandakini watershed

which the sub-watersheds were subjected. HCs for sub-watersheds were constructed
by plotting relative area (the area above a given contour (a) relative to the total area
of the catchment (A)) on the abscissa and relative height (height of a given contour
(h) from the basal level to the maximum height of the catchment (H)) on the ordinate.
The resulting HC provides information about the volume of land mass located below
or above the reference basal plane (e.g., Markose and Jayappa [4]).
3.2 Hypsometric Integral (Hsi) Estimation
The Hsi is generally calculated from the HC, where it is accurately demonstrated
[25]. The elevation-relief ratio (E) proposed by them is demarcated as follows:
Elev(mean) − Elev(min)
E ≈ Hsi =
Elev(max) − Elev(min)
where
E equivalent to Hsi
Elev(mean) weighted mean elevation of the delineated watershed,
determined from the conspicuous contours derived from
the DEM; and
Elev(max) and Elev(min) are the maximum and minimum elevations in the water-
shed.
The values derived from HC and Hsi are helpful in determining progress in the
stages of exogenic geologic evolution or geological weathering. The criteria proposed
by [1] for determining the stages of watershed development were used in determining
the geologic stages of watershed development as a function of Hsi values. When the
Hsi value is ≤ 0.35, it indicates the Monadnock stage of the landform of a watershed,
where the HC is distorted and the area is represented by projected Monadnock hills,
when the Hsi value is between 0.35 and 0.5, it indicates an equilibrium and mature
fragmentation stage of landform within the watershed, indicating a stable state of
development; and finally, when the Hsi value is ≥ 0.5, it indicates that the watershed
is in the juvenile or equilibrium stage of development [26].
The results of this study are discussed in the following subsections: hypsometric
curve, hypsometric integral, prioritization, conservation and watershed management.
Table 1 Hypsometric integral values of the six sub-watersheds of Mandakini watershed

Sub-watersheds Area Max Min Mean Hypsometric Geological
(km2 ) elevation elevation elevation integral (Hsi) stages
(m) (m) (m)
B1 63.0 2258 1016 1636.8 0.50 Late
youthful
B2 38.0 2113 949 1530 0.49 Early
mature
B3 61.3 2046 960 1462 0.46 Mature
B4 78.4 2729 920 1768 0.46 Mature
B5 49.7 2128 883 1476 0.47 Mature
B6 42.0 1854 849 1302 0.45 Mature
4.1 Hypsometric Curve
For the six sub-watersheds of Mandakini Watershed, hypsometric curves had been
prepared based on the hypsometric parameters (Table 1; Fig. 6). Later, the shoulders
of the plotted curves were interpreted as an important source for the study of erosion
progress in a catchment area [1]. The interpreted shoulders of most of the sub-
watersheds indicate that they have a concave shape and have passed from the youth
stage to a mature stage. Thus, the results of the study are consistent with those of
other watersheds in the western Himalayas, where most watersheds show a transition
from the younger to the more mature stage [9]. From the interpretation of the curves,
it is clear that the erosion progress varies a little in the different sub-watersheds of
the Mandakini Watershed.
4.2 Hypsometric Integral
Runoff and mass movement are the two main hydrological processes responsible for
precipitation events in a watershed system. The Hsi value plays an important role in
measuring erosion processes, even if it is used indirectly to estimate soil losses in
a watershed system. In small watersheds, HCs are convex and values of Hsi are in
the unity range compared to large watersheds, indicating active slope processes. In
large watersheds, the integral approaches zero, HCs are concave and fluvial processes
dominate [21]. As given in Table 1, the Hsi values of the sub-watersheds of Mandakini
Watershed are in mature stage and are in the transformation phase of peneplanation
or in the geological phase of deterioration. Strahler [1] observed that hypsometry
describes the age of watershed. Moglen and Bras [16] found that the hypsometric
distribution is influenced by the subsurface rock formation.
From soil degradation, it can be inferred that most sub-watersheds were formed
by incision of channel beds, downward movement of quantifiable topsoil and rock
Fig. 6 Hypsometric curves of the six sub-watersheds of Mandakini watershed
masses from higher elevations and removal of soil masses and erosion of stream
banks. Ritter et al. [7] examined that sub-watershed that are in the mature phase have
slow erosional activity, except under certain circumstances with high storm intensity
and peak runoff.
4.3 Relationship with Past Earthquakes
Three earthquakes which have occurred in the last two decades in the neighboring
area, viz. 4.5 magnitude with epicenter distance 22 km NE of Rudraprayag earth-
quake of November 19, 1998; 4.5 magnitude with epicenter distance 13 km NNE
of Rudraprayag earthquake of March 12, 2001 and 4.6 magnitude with epicenter
distance 20 km NE of Rudraprayag earthquake of September 05, 2005 (https://earthq

uake.usgs.gov/earthquakes). Hence, earthquake density is frequent in the neighbor
areas which leads to the instability of the watershed basin and also make it the more
proneness for the erosion.
4.4 Prioritization, Conservation and Management

of Watershed
From the hypsometry analysis, it is clear that the study area is susceptible to high
runoff. Therefore, it is imperative that the necessary measures be taken to regu-
late surface runoff. The Hsi values show that all six sub-watersheds are affected by
maximum total surface runoff. Accordingly, the Hsi values can be used to priori-
tize the sub-watersheds. When sub-watersheds are identified, appropriate measures
can be taken to contain and control surface runoff and the resulting erosion. It is
very important that both structural and non-structural measures be taken to contain
the process of sediment loss and conserve water by implementing integrated water-
shed management practices. Structural control measures include the construction of
retention dams, tillage pods, contour walls and other structures to prevent flooding
and promote water conservation measures. In addition, sustainable agricultural prac-
tices such as strip cropping, crop rotation and agroforestry will be introduced. Non-
structural measures include an early warning system, especially to predict flooding,
and awareness programs to discourage traditional agricultural practices.
5 Conclusions
Hypsometric analysis is useful to determine the complex nature of denudation activity

and also to evaluate the change in morphological features in a watershed. It includes
the HC and Hsi, which serve as indicators for evaluating runoff and associated
erosion in a watershed, and particularly in a hilly area where much of the terrain
is inaccessible. Therefore, it is beneficial to know the erosion status and prioritize
the sub-watersheds to apply conservation strategies. The application in the present
study showed that all sub-watersheds of Mandakini Watershed are comparatively
mature and erosion processes are stabilizing. The results of Hsi values showed that
sub-watershed B1 is at high risk of erosion unlike the other associated sub-basins
of the Mandakini Watershed. Therefore, this sub-basin requires extreme care and
appropriate remedial measures such as structural and non-structural methods can be
initiated as a priority to minimize soil erosion and promote water conservation for
integrated management of the basin.
References
1. Strahler, A.N.: Hypsometric (area-altitude) analysis of erosional topography. Geol. Soc. Am.
Bull. 63, 1117–1141 (1952)
2. Langebein, W.B.: Topographic characteristics of watersheds. USGSWater Supply Paper, 947-C,
pp. 127–157 (1947)
3. Horton, R.E.: Erosive development of streams and their catchments, hydrophysical approach
to quantitative morphology. Geol. Soc. Am. Bull. 56, 275–370 (1945)
4. Markose, V.J., Jayappa, K.S.: Hypsometric analysis of the Kali River Basin, Karnataka, India,
using geographic information system. Geocarto Int 26(7), 553–568 (2011)
5. Hurtrez, J.E., Lucazean, F., Lave, J., Avouac, J.P.: Investigation of the relationship between
basin morphology, tectonic uplift and denudation based on the study of an active fold belt in
the Siwalik Mountains. (Central Nepal). J. Geophys. Res. 104, 779–796 (1999)
6. Chow, V.T.: Hand Book of Applied Hydrology. McGraw Hill Book Company, New York (1964)
7. Ritter, D.F., Kochel, R.C., Miller, J.R.: Process Geomorphology. McGraw Hill, Boston (2002)
8. Weissel, J.K., Pratson, L.F., Malinverno, A.: The length scaling properties of topography.
Geophys. Res. 99, 13997–14012 (1994)
9. Singh, O., Sarangi, A., Sharma, C.M.: Methods for estimating hypsometric integral and its
significance for erosion status of Northwestern lesser Himalayan watersheds. Water Resour.
Manage. 22, 1545–1560 (2008)
10. Manjare, B.S., Pophare, A.M.: Identification of groundwater development zones in the Morna
river sub-basin, central India. J. Geosci. Res. 52(2), 139–145 (2020)
11. Bishop, M.P., Shroder, J.F., Bonk, R., Olsenholler, J.: Geomorphic change in high mountains:
a western Himalayan perspective. Glob. Planet. Change 32, 311–329 (2002)
12. Harsha, J., Ravikumar, A.S., Shivakumar, B.L.: Evaluation of morphometric parameters and
hypsometric curve of Arkavathy river basin using RS and GIS techniques. Appl. Water Sci.
10(86). (2020). https://doi.org/10.1007/s13201-020-1164-9
13. Huang, X.J., Niemann, J.D.: Modelling the potential impact of groundwater hydrology on
long-term watershed development. Earth Surf. Process. Landf. 31, 1802–1823 (2006)
14. Lamsoge, B.R., Vijesh, V.K., Pophare, A.M., Katpatal, Y.B.: Determination of erosion
susceptibility of WR -2 watershed using hypsometric analysis. J. Geosci. Res. 3(1), 85–89
(2018)
15. Mishra, N.: Hypsometric integral—a basis for determining erosion status and priority number
of unmeasured watersheds. J. Soil Water Conserv. 32, 38–45 (1988)
16. Moglen, G.E., Bras, R.L.: The effect of spatial heterogeneities on geomorphic expression in a
model of basin evolution. Water. Resour. Res. 31, 2613–2623 (1995)
17. Ohmori, H.: Changes in the hypsometric curve due to mountain building resulting from
simultaneous tectonics and denudation. Geomorphology 8, 263–277 (1993)
18. Pradhan, K., Senapati, P.C.: Hypsometric analysis of some selected watersheds of Hirakund
catchment. J. Soil Water Conserv. 30, 183–185 (2002)
19. Shukla, D.P., Dubey, C.S., Ningreichon, A.S., Singh, R.P., Mishra, B.K., Singh, S.K.: GIS
based morpho-tectonic studies of Alaknanda River basin: a precursor for hazard zonation. Nat.
Hazards 71, 1433–1452 (2014). https://doi.org/10.1007/s11069-013-0953-y
20. Sivakumar, V., Biju, C., Deshmukh, B.: Hypsometric analysis of Varattaru River Basin of Harur
Taluk, Dharmapuri Districts, Tamil Nadu, India using geomatics technology. Intl. J. Geomat.
Geosci. 2, 241–247 (2011)
21. Willgoose, G., Hancock, G.: Revisiting the hypsometric curve as an indicator of form and
process in transport-limited catchment. Earth Surf. Process. Landf. 23, 611–623. (1988) https://
earthexplorer.usgs.gov/
22. Goel, A.K., Singh, J.K.: Hypsometric analysis for the foothills of the Shivalik. Indian J. Soil
Conserv. 28, 84–85 (2000)
23. Sharma, S., Mahajan, A.K.: GIS-based sub-basin prioritization through morphometric analysis
in the outer Himalayan region of India. Appl. Water Sci. 10, 1–11 (2020). https://doi.org/10.
1007/s13201-020-01243-x
24. Singh, O.: Hypsometry and erosion susceptibility: a case study in the lesser Himalayan
watersheds. J. Soil Water Conserv. 8(2), 53–59 (2009)
25. Pike, R.J., Wilson, S.E.: Elevation-relief ratio, hypsometric integral and geomorphic area-
altitude analysis. Geol. Soc. Am. Bull. 82, 1079–1084 (1971)
26. Kusre, B.C.: Hypsometric analysis and watershed management of the Diyung watershed in
northeast India. J. Geol. Soc. India 82, 262–270 (2013). https://doi.org/10.1007/s12594-013-
0148-x
Understanding the Structure
and Tectonic Configuration of Bengal
Basin for Earthquake Magnitude
Prediction
Mir Fazlul Karim and Daya Shanker
Abstract The Bengal Basin is a living model of an active and complex geological
entity consisting of dynamic deltaic depositional complex, heterogenous and multi-
dimensional tectonic deformational settings from three tectonic plates. The earth-
quake risk of the basin is not rationally anticipated due to the absence of adequate
seismic and geodetic data, subsurface geological maps, insufficient representative and
instrumental data acquisition system. For a geometrical array and vector-mapping
of tectonic deformation, intensive data search is done from published sources. An
investigation on crustal configuration using available seismic data, velocity images,
and travel time tomography along number of profiles is done. Based on these studies,
the basin is classified into five geotechnical domains. An attempt is made to under-
stand the scenario of sediment input, tectonic and gravitational stress distribution,
and their pattern of deformation vectors for earthquake magnitude prediction. The
crustal study indicates that the basin is severely fragmented, and topography of the
sedimentary sequences is shaped and stressed by various sizes of graben and horsts.
The basin received enormous volume of sediments, which resulted in deposition
of very thick clastic sediments. A seismic zoning map is prepared using the five
major geotechnical domains with maximum possible magnitude of earthquakes to
be occurred. The 1918 Srimangal event of Mw ≈ 7.5 is the largest recorded earth-
quake and no seismogenic structure has been found to cause great earthquakes in
Bangladesh.
M. F. Karim (B)
Director (Ex), Geological Survey of Bangladesh and GeoEastern Inc. Inc., Massachusetts, USA
e-mail: geologist.karim@gmail.com
D. Shanker
Department of Earthquake Engineering, IIT, Roorkee, Uttarakhand 247667, India
e-mail: d.shanker@eq.iitr.ac.in
https://doi.org/10.1007/978-981-99-1459-3_50
626 M. F. Karim and D. Shanker
Keywords Bengal basin · Tectonics · Seismicity · Earthquake
1 Introduction
Bangladesh and its surrounding regions, including West Bengal and the north-east
India, occupy the most active tectonic regime of the world. Seismic risk in Bangladesh
is one of the major socio-economic concerns due to very high population density,
unplanned urbanization, and non-engineered construction practices. We emphasize
a rational prediction of the state of seismic threats in the region and not to apprehend
overscored earthquake magnitude. Overscored and uncertain earthquake prediction
would increase unnecessary recurring costs in the engineering practices and earth-
quake risk management system for the citizens. The Bengal Basin is mostly built
of the Ganges–Brahmaputra Meghna (GBM) delta and assumed to have the thickest
sedimentary cover of the present world. The GBM delta is the largest prograding
delta consisting of varied floodplain sediments from most dynamic, densely arrayed
intricate river system, and older terraces. The delta occupies a major part of basin
which is one of the largest and youngest sedimentary basins actively and tectonically
stressed by the Indian, Tibetan, and Burmese plate motions [1].
The historical records of earthquakes in and around Bangladesh indicate that
generally at every two decades of interval there are earthquakes of intensities greater
than X in MM scale in the northern and north-eastern part of the country [2].
The damages due to these earthquakes are widespread. The seismicity of these
events is deeply associated with tectonic behavior of the subducting Indian plate.
It is believed that the country is neo-tectonically active for being located near the
syntax of three tectonic plates and Himalayan Mountain Building System. The
Himalayan range is seismically most active zone of Asia. Four great earthquakes
of magnitude exceeding 8 during 1897, 1905, 1934, 1950 and another 10 earth-
quakes exceeding magnitude 7.5 have occurred in the Himalayan belt during the
last 100 years. Previous studies indicate presence of active faults in and around the
Bengal Basin [3, 4]. These faults are capable to cause moderate earthquakes. The
Himalayan Arc to the north and northeast and Myanmar Arc to the east and southeast
have generated number of devastating great earthquakes in the past and shaken the
entire Indian subcontinent. Recurrence of similar shake in this region would cause
unmanageable destruction. The region is placed in a high seismic zone in the Global
Seismic Hazard Map and by Lamont–Doherty Earth Observatory (LDEO) [6].
The chronicle of studies of tectonics of the Bengal Basin System (BBS) is not that
long and most of the studies were based on limited field and meager instrumental
data and theoretical assumptions. Recently, several papers and instrumental data are
available over the internet having a detailed quantitative database, mechanical, and
geodynamic analysis of the BBS, consisting of methodical investigations. During the
last few decades, an aggressive advancement has taken place for studies and under-
standing of geodynamic behavior of the basin system. The finding [5] of decollement
structures and development of Neogene accretionary prism complex in an oblique
Understanding the Structure and Tectonic Configuration of Bengal … 627
subduction environment is a very important turning point in the understanding of

tectonic mechanism of BBS. The Lamont–Doherty Earth Observatory (LDEO) scien-
tists executed a revolutionary sophisticated study in the BBS for a decade for the first
time using Global Positioning System (GPS) and computer models [6]. The scien-
tists described a huge, locked subduction interface or megathrust, where the front
is extended to the megacity of Dhaka from the tectonically folded sediments in the
east. Though the physical attributes and geometrical dimensions and configurations
are not well defined. The LDEO study concluded with likelihood of occurrence of
great earthquakes from the accumulated elastic strain across the megathrust zone.
The study concluded with a prediction of occurrence of a great earthquake (Mw 8.2
to 9) in Bangladesh and North-East Indian region at any time. This prediction has a
lot of uncertainties which were not mentioned by the authors [6] in the publication
and turned an issue that weighed heavily with Bangladesh geologists and geotech-
nical engineers. We believe that this is a subject of further studies for a coherent
understanding and engineering practices in the region.
1.1 Data and Methods
This work is mainly based on detailed field investigations throughout Bangladesh

and neighboring states of India. As Bangladesh doesn’t have adequate instrumental
geophysical and GPS database, we used required GPS data from Nevada Geodetic
Laboratory [7] Map Browse GPS Stations and UNAVCO. Earthquake magnitude,
locations, and hazard maps are retrieved from IRIS, USGS and Global Earthquake
Model (GEM). The deep crustal and Moho configuration data are used from published
sources including the seismic, geodetic data and subsurface geological maps.
2 Geology of Bangladesh
A larger part of Bangladesh is occupied by the world’s largest delta formed by the
Ganges–Brahmaputra Meghna River system. It is almost entirely formed of vast
alluvial plain except some strips of asymmetrically and plunged folded hills on the
north-east and eastern margins. The alluvial plains have the elevation from about
90 m in the north-western part of the country to ~ 0 m along the coastal plains.
Apart from these, there are three tracts of Terraces in the country with a maximum
elevation of 40 m, raised concurrently during the Himalayan mountains final uplift.
About 80% of the country is covered by Holocene deposits; consisting of fluvio-
deltaic sediments, where upper top materials (10–20 m) are very sensitive to cyclic
stresses and vulnerable to annual flooding. The 12% area is exposed with tertiary
sedimentary rocks where seismic response is very much complex due to intricate
geological structures and the 8% area are raised terraces and formed of stiff and hard
clay residuum, Fig. 1.
Fig. 1 Geological map of Bangladesh (Modified by Karim (2004) from Alam et al. (1990))
The oldest exposed rock is the Tura Sandstone of Paleocene age but older rocks
like Mesozoic, Paleozoic, and Precambrian Basement have been encountered in the
drill holes in the north-western part of the country [9], Karim 2004). Mainly sand-
stone, siltstone, shale, and claystone represent the tertiary rocks. The field investi-
gation indicates that the highly oxidized Pleistocene sediments and weathered soils
are exposed in three isolated uplifted blocks. The Holocene deposits, consisting of
unconsolidated sand, silt, and clay of varying amounts, are the products of active
alluvial, fluvio-deltaic, or coastal depositional environment. All these materials will
behave differently during various degrees of ground shaking.
3 Tectonics of Bengal Basin
Bangladesh shares the Bengal Basin with West Bengal with equal degree of seismic
risks. The evolution of the basin is associated with the movements of Indian, Eurasian
and Myanmar platelets. Through a long geological time (Permian to Recent) the base-
ment of Bengal Basin, below a thick sedimentary cover has been severely faulted
and fractured. The remote sensing image analysis indicates that the modern-day
topography very clearly reflects the gradient of deep basement or crustal configura-
tion (Fig. 2). We have determined that many of the Bangladesh rivers are structurally
controlled and follow the old faults or lineaments. We detected deep seated crustal low
and high by investigating available published tomographic images, seismic section,
(Figs. 2 and 3), and rheological characteristics of the Neogene sedimentary cover
[10]. We expect presence of similar vertical detachments or staggered crustal blocks
all along the thick sediments. These detachments cause frequent moderate to low
magnitude earthquakes. We studied a swarm of small to moderate earthquakes in
2021 and found linear alignments of the epicenters which reflects the subsurface
crustal structural trends.
Following a prediction of 8.2–9 M earthquake in Dhaka Megacity by Steckler
et al. [6], we performed a comparative review of the geodynamic and geothermal
gradient and geo-tectonic structures. Our study indicates that the shallow thermal
gradient is getting cooler toward east [15] or folded belt, whereas the deep crustal
Fig. 2 An illustrative tectonic map of Bengal Basin and adjoining regions showing important
tectonic elements and crustal divisions. A Schematic tectonic map of Bengal Basin and surrounding
terrains of Indian shield covering Bangladesh and part of India and Myanmar (political boundaries
not shown). The terrain analysis portrayed a unique westward sinusoidal progression of the folds
due to steady uniform compressional stresses from the east with a convergence vector (∼N27 °E)
[11]. B The crustal low or Graben and crustal high or Horst in the Bengal Basin are identified,
shown in a satellite image with their surficial scars [12]. C Based on seismic and earlier studies (by
Geological Survey of Bangladesh), Bangladesh is divided into five major geotechnical domains:
1. Stable shelf, 2. Continental slope and Hinge Zone, 3. Deep central trough (Sylhet-Hatiya), 4.
Eastern Fold Belt of Chittagong Tripura and 5. Dauki Fault Zone [13]
Fig. 3 Three geological sections along A-A’, B-B’ and C-C’ are prepared from the velocity model
cross sections of the Bengal Basin (modified after [18] and [19]. The section A-A’ (750 km)
apparently does not show any vertical or horizontal lineation and indicate moderately isotropic
shear wave velocity pattern at the used resolution. We compared and determined gradual vertical
variation indicating 20 km thick upper crust and 20 km thick lower crust (after Karim et al. [12],
Red arrow for Shillong popup mark). The section B-B’ shows a significant variation in shear wave
velocities, both vertically and laterally. The basin profile clearly reflects the subsiding geometry of
Bengal Basin. The high angle shear zone close to Chandina terrace shows trace of vertical lineation
and flexures. Presence of low density sedimentary thick deposits indicates a rapid sedimentation
below 3 km of depth. A drastic change in the depositional environment in the central region of the
basin is remarkable. The velocity section C-C’ shows an extraordinary and heterogenous structural
pattern associated with both depositional and tectonic environments. It appears that the buoyant
mantle materials have pushed the Shillong massif up and crushed the neighbouring continental
crust slabs into multiple horsts and grabens (?)
and mantle are cooler westward from the Indo-Myanmar subduction slope or under
Bengal Basin [16]. The thermal variation possibly causing overall density and fric-
tional variation of the tectonic slabs under heterogenous sedimentary sequences. We
assume that the GPS velocity vectors require further calibration for geothermal and
physico-mechanical of tectonic movement in the Burmese–Bangladesh folded zone.
An extensive field search for any signatures of Subduction Zone was carried out and
no such structures were found in the regions. This work did not find any geological
evidence to justify earthquake greater than 7.5 magnitude within the periphery of
Bengal Basin under huge gravitational load or vertical stress of average > 15 km
of sediments. The detailed structural and tectonic morphological studies along with
deep petroleum drilling data analysis and available geophysical data did not find
presence of any such megastructures capable of generating great earthquakes. Hence

considering the distal source structure, that caused the popularly known Assam earth-
quake of 1897 8.5 Magnitude earthquake, a test model was run at the USGS (pers.
cmmun. John Whitney, former USGS). The analysis does not indicate any higher
amplification factors in the existing ground conditions of Bangladesh.
It is identified that the slowed down tectonic stresses helped to accommodate the
thick tertiary folded sedimentary belt out of thrusts from the Burmese plate while
the energy of the thrust slowly reduced and helped in keeping the petroleum deposits
of Bengal Basin less disturbed. The gradually merging folded sediments started
pushing the NE-SW trending hinge zone, tilted and raised the deeper basin deposits
including Madhupur Residuum (Fig. 1). Similarly, the Shillong massif, a crustal slab
of Bengal Basement thrusted up where the conjugate slabs slipped or sank down to
the south, e.g., Sylhet Trough, [17]. We performed a comparison of the exhumation
and uplift sequences of Shillong plateau for understanding of Bengal basin crustal
topography. We identified that tectonic fragmentation or crustal disintegration has
occurred throughout the basement [18] of Bengal Basin in the form of graben and
horsts, since a very long geological time and deposition of enormous sediments.
The heterogenous depositional and tectonic partition of undetermined set of graben,
horst, and trenches in the basin (e.g., Surma basin or Surma graben, Fig. 2).
4 Crustal Configuration and Sediment Thickness
For a detailed crustal investigation for the Bengal Basin, we used published reports
and research papers and found that the Bengal Basin is severely fragmented, and
topography of the sedimentary sequences is shaped by various sizes of grabens and
horsts.
We are very much convinced in identifying the Sylhet Trough as a tectonic graben
and active subsiding basin. The received huge volumes of Neogene sediments, which
resulted in deposition of measuring between 14 km and over 16 km in thickness.
The tectonic deformation patterns of Bengal Basin are assembled in the sediments
of various time series where both the thick-skinned and thin-skinned deformation
are present and characterized by tomographic anomalies. In conjunction, we used
Bouguer gravity anomaly and aeromagnetic anomalies of Bangladesh published by
the Geological Survey of Bangladesh (GSB) and United States Geological Survey
(USGS) to correlate recent available high-resolution images and deep geophysical
data and found satisfactorily positive correlation among the maps and geological
sections. We used an interpretation done by [19] and Center for Earthquake Research
and Information at The University of Memphis, for ground motion characteristics
of the Bengal Basin region in Bangladesh. A computer simulation was run by the
authors for a series of historical seismic records and scenario earthquakes. We found
this study on seismic velocity model for the region very appropriate for our under-
standing of true geometrical boundary condition. It presented the initial version of a
three-dimensional model built upon a geology-based representation of the geometry
of the Bengal Basin and simple approximations of the basin deposits, crustal, and
background structures (Fig. 3).
The results from a low-frequency ground motion simulation (f ≤ 0.5 Hz) from
a Mw 5.1 Chandpur Earthquake compared with available basic qualitative charac-
teristics, finite-element parallel code used by Huda and [19]. The results are very
satisfactory with the known geometry of Bengal Basin crustal configuration and
exposed bed rock in the Shillong Massif regime. The horizontal and vertical discon-
tinuities clearly comparable with the known stratigraphic stackings and intensity
of folding in the east. The extent and shape of all tectonic divisions are positively
correlated.
5 The Geometry of Tectonic Motion Vector in and Around

Bengal Basin
Present study involves a detailed geological field mapping, remote sensing image
processing and geotechnical analysis throughout the Bengal Basin and NE Indian
region for understanding and identification of seismic source areas [12]. Most of the
seismic sources and crustal (both clastic and crystalline) layers reflect that lateral and
vertical discontinuities are at various locations subjected to multiple segmentation
and torsional shear stress having complex tectonic motion vectors. The differen-
tial tectonic motion over the Bengal Basin segments is potentially influenced by
the complex and multidirectional regional compression forces derived from various
directions (Fig. 5 insets A-E). The oblique and transverse convergence of the Sunda
and Indian plates and the intense compressive stress rotates from NE-SW to E-W
near the Bengal Basin along the inner and northern arc. This rotation is consistent
with the splitting deformation reflected in the rotation of the relative displacement
vector from the SSW direction, where the Sunda-Burma motion is toward Burma-
India subduction, i.e., NNW directional motion. As a result of this segmentation,
the main belt-parallel fault zones show a differential movement along N-S arc with
strike-slip and oblique reverse dextral slip displacement.
This study finds a differential pattern of tectonic motion as monitored by various
GPS measurement group. The multiple convergent and divergent vectors make a
jargon of the stress environment (inset H in Fig. 4) in and around Bengal Basin. This
would require further substantial studies before a prediction or superfluity of extraor-
dinary earthquakes [6] in the region. The general relationships among geological
structures, tectonic motion vector, earthquake dynamics, and deformation character-
istics of rocks have complex interaction over an uncertain time and space and requires
to well analyzed for a rational seismic risk management in Bangladesh (Fig. 5).
Fig. 4 This diagram shows a clarification of the regional seismotectonic interaction over the Bengal
Basin, the crustal and mantle dynamics, existing stress pattern. Multiple convergent and divergent
vectors make a jargon of the stress environment centering deeper part of the Bengal Basin. Modern
days GPS motion data can be contradictory to the major NNE-SSW India-Eurasian (inset A) and
E-W Sunda-India convergent (inset B, E, F), while syntaxial torsional vector in the NE India at
the northern tip (insets B, C, D) of Bengal Basin plays severe deformation causing shallow depth
moderate size earthquakes. The N-S strike-slip faults (insets E, F, G, H) over the India-Bruma
subduction splay generate deep earthquakes due to strike-slip partitioning in an oblique convergence
zone between the India Plate and the West Myanmar Block (see also Fig. 5 for systematic clusters
of earthquakes). [Ref. for Inset diagrams. A Socquet et al. [26], B Karim [11] C Schellart (2019) D
Barber et al. [20], E and F Earnest [24] G Khin et al. [25]. The Inset H is prepared for this paper]
6 Exploring the State of Moho and Deep Crustal

Configuration of the Bengal Basin
Our study along the Dauki fault region and reviewing of published papers, data [10]
and [21] and receiver function analysis indicate presence of continental crust under
the Brahmaputra Valley and the Shillong Plateau, whereas the crust beneath Bengal
Basin is identified as a complex of amalgamated continental to transitional type.
Toward the north of the plateau, the crustal thickness increases by 8–10 km with
Moho flexure of 30°. The southern section is downfaulted by12–13 km across the
Dauki Fault. The plateau uplift is mediated by thrust faulting on the reactivated Dauki
Fault and back thrusting on the north dipping Oldham Fault. We extensively studied
and reviewed the deep geology and Bengal Basin crustal and mantle dimensions, and
configuration from available published geophysical research papers, reports, maps,
and data; and found out that most of the Bengal Basin has experienced severe frag-
mentation and sinking of extended Indian craton along the southern neighborhood
of Shillong craton.
Fig. 5 Versus structure along different profiles marked in the seismicity map of Bangladesh, NE
India, and Myanmar (Burma). Showing Moho discontinuity, sedimentary sequences, and crustal
boundaries. (Modified from Mitra et al. [21] and Kumar et al. [10]) Abbreviations—IBR Indo-Burma
Ranges; SM Shillong Massif; ITSZ Indus-Tsangpo Suture Zone; STD Southern Tibetan Detachment;
MCT Main Central Thrust; MBT Main Boundary Thrust; HFT Himalayan Frontal Thrust; JF
Jamuna fault; SF Sylhet fault; DBF Dhubri fault; DKF Dhansiri-Kopili fault; DF Dudhnoi fault;
EHZ Eocence Hinge Zone; OF Oldham fault; PCF Po Chu fault; PF Purlang Fault
The tomographic image of shear wave along six sections of Bengal Basin is clearly
showing crustal and mantle geometry of Bengal Basin. Section B shows upheaval
of mantle that pushed up a part of Indian craton to form the present-day Shillong
Plateau and caused sinking of southern portion of craton to preserve the continental
crust under thick sediments of Bengal Basin [10].
7 Seismic Zonation Maps
A Seismic Zoning Map of Bangladesh (SZMB) is prepared (Fig. 6) using Global

Earthquake Model [22] in conjunction with consideration of seismotectonic behavior,
spatial distribution of earthquakes in the region, and potential seismic source zones
and geological attributes.
The maximum possible magnitude of earthquakes to occur within the Bengal
Basin is Mw ≥ 7.5 considering an occurrence period of last 100 years in northeast
India and its adjoining regions. It is found in this study that there are no seismogenic
structures in the Bengal Basin that can generate earthquakes greater than 7.5 magni-
tude. We have identified a gross anomaly in the results of tectonic structures done
by various researchers from different geographical locations and chose a rational
Fig. 6 Proposed seismic zoning map A of Bangladesh considering tectonic setup, potential seismic
source area, distribution of past earthquakes, crustal and basement geometry, and neotectonics
progression in and around Bengal Basin B. Present SZMB as published with Bangladesh National
Building Code, 2021
seismic source map of Bengal Basin and neighboring region (Seismic source map—
after [2] and Panthi et al. [2]). The recently published SZMB as included in the
Bangladesh National Building Code 2021 reasonably correlates with the proposed
SZMB (modified from Global Earthquake Model and Global Seismic Hazard Map,
version 2018.1, Pagani et al. [22]).
8 Discussion
The seismic or earthquake risk management in Bangladesh is one of the major socio-
economic concerns owing to its high population density together with tectonic setting
due to complex syntaxial stress–strain and multidimensional differential deformation
conditions from Indian, Tibetan, and Burmese tectonic plates. Bengal Basin System
is a unique and dynamic natural laboratory for studying an active deltaic depositional
environment under a continuous influence of tectonic and gravitational stress pattern.
The earthquake risks in Bangladesh are poorly understood due to absence of
required instrumental seismic data acquisition facilities and institutional research and
development systems. This study identifies seismic source areas, possible configu-
ration, and crustal thickness of Bengal Basin. It is found that the crust is severely
segmented during the rapid break up of Indian craton and rising of Shillong Plateau.
The segmentation pattern is a matter of further studies for seismic hazard analysis.
The basin is filled up with heterogeneous deposits in a unique deltaic environment.
Possibly the crustal segmentation later facilitated to weaken the tectonic movements
as the basin devoid of giant structures which would not be able to accommodate much
strain to generate earthquakes larger than magnitude 7.5 in Bangladesh. This finding
will provide an improved understanding and considerations in the geo-engineering

practices and the earthquake risk of the country will be rationally anticipated.
References
1. Alam, M.: Tectonic classification of Bengal Basin. Geol. Soc. Am. Bull. 83(2), 519–522 (1972)
2. Shanker, D., Panthi, A., Singh, H.N.: Long-term seismic hazard analysis in northeast Himalaya
and its adjoining regions. Geosciences 2(2), 25–32 (2012). https://doi.org/10.5923/j.geo.201
20202.04
3. Karim, M.F., Khan, S.I.: Report on Geological Investigation of Rangamati Earthquakes.
Unpublished report of Geological Survey of Bangladesh, July August (2003)
4. Khan, A.A., Hoque, M.M., Shaharier, K.M., Hoque, M.: Convergent tectonics and sedimenta-
tion in the Eastern Margin of the Indian Plate with emphasis on the Bengal Basin Bangladesh.
J Geol 21, 9–22 (2002)
5. Sikder, A.M.: Tectonic evolution of eastern folded belt of Bengal Basin. Unpubl. PhD Thesis,
Dhaka Univ., p. 175 (1998)
6. Steckler, M., Mondal, D., Akhter, S., Seeber, L., Lujia F., Gale LJ., Emma M., Michael,
H.: Locked and loading megathrust linked to active subduction beneath the Indo-Burman
Ranges. Nature Geosci. 9. (2016). https://doi.org/10.1038/ngeo2760
7. Blewitt, G., Hammond, W.C., Kreemer, C.: Harnessing the GPS data explosion for interdisci-
plinary science, Eos, 99. 2018. https://doi.org/10.1029/2018EO104623
8. Alam, M.K., Hasan, A.K.M.S., Khan, M.R., Whitney, J.W. 1990. Geological map of
Bangladesh. Geological Survey of Bangladesh, Dhaka. Digitally compiled by F.M. Persits,
C.J. Wandrey, R.C. Milici, (USGS), A. Manwar, (DG, GSB), (2001)
9. Coleman, J.M.: Brahmaputra River: channel processes and sedimentation. Sediment Geol
Special Issue 3(2/3), 131–237 (1969)
10. Kumar, A., Kumar, N., Mukhopadhyay, S., Klemperer, S.L.: Tomographic image of shear wave
structure of NE India based on analysis of Rayleigh wave data. Front. Earth Sci. 9, 680361
(2021). https://doi.org/10.3389/feart.2021.680361
11. Karim M.F.: Country Paper: High-Level Expert Group Meeting on Technical Options for
Disaster Management Systems: Tsunamis and others. 22–24 June 2005, UNESCO, Bangkok,
Thailand. (2005)
12. Karim, M.F., Rahman, M.Z., Sikder, A.M., Shanker, D.: Tectonic and Geotechnical Review
for Engineering Design and Seismic Risk Management in Bangladesh. Poster Presentation at
2021 SCEC Annual Meeting. 2021, 08. Poster #055, Earthquake Engineering Implementation
Interface (EEII). https://www.scec.org/meetings/2021/am/poster/055
13. Karim, M.F., Kayal J.R., Shanker, D., Khandaker, N.I., Sikder, A.M., Rahman, M.Z., Hassan,
M.Q.: Tectonic and geotechnical review of Bengal Basin for Seismic Risk assessment in
Bangladesh. Geological Society of America. Abstracts. 53(6). https://doi.org/10.1130/abs/202
1AM-371261
15. Akbar, M.A.: An assessment of the geothermal potential of Bangladesh. Geothermal Training
program Report 2011 Orkustofnun, Grensasvegur 9(5) IS-108 Reykjavik, Iceland. (2011)
16. Beardsmore G.R., Cull, J.P.: Crustal heat flow a guide to measurement and modelling.
Cambridge University Press. (2001). https://doi.org/10.1017/CBO9780511606021
17. Johnson, S.Y., Alam, A.M.N.: Sedimentation and tectonics of the Sylhet Trough. Bangladesh.
Geol. Soc. Am. Bull. 103, 1513–1527 (1991)
18. Biswas, S., Coutand, I., Grujic, D., Hager, C., Stockli, D.: Exhumation and uplift of the Shillong
plateau and its influence on the eastern Himalayas: new constraints from apatite and zircon
(U-Th-[Sm])/He and apatite fission track analyses. Tectonics, American Geophysical Union
(AGU), (2007)
19. Huda, M., Taborda, R.A.: Three-dimensional regional-scale earthquake ground motion simu-
lation for the Bengal Basin. Center for Earthquake Research and Information. In: 16th World
Conference on Earthquake Engineering Santiago Chile, January 9th to 13th, (2017)
20. Barber, A.J., Zaw, K., Crow, M.J. (eds.): Myanmar: geology, resources and tectonics. Geolog-
ical Society, London, Memoirs, 48, 19–52. 2017. https://doi.org/10.1144/M48.2 © 2017 The
Author(s). Published by The Geological Society of London. Publishing disclaimer: www.geo
lsoc.org.uk/pub_ethics
21. Mitra, S., Priestley, K.F., Borah, K., Gaur, V.K.: Crustal structure and evolution of the Eastern
Himalayan plate boundary system, Northeast India. J Geophys Res Solid Earth 123, 621–640
(2018). https://doi.org/10.1002/2017JB014714
22. Pagani, M., Garcia-Pelaez, J. Gee, R., Johnson, K. Poggi, V. Styron, R. Weatherill, G. Simionato,
M. Viganò, D. Danciu, L. Monelli, D.: Global earthquake model (GEM) seismic hazard map
(version 2018.1-December 2018). https://doi.org/10.13117/GEM-GLOBAL-SEISMIC-HAZ
ARD-MAP-2018.1
24. Earnest, T.C.A., Sunilkumar, K.: Silpa sinking slab stress and seismo-tectonics of the Indo-
Burmese arc: a reappraisal. Tectonics (2021). https://doi.org/10.1029/2021TC006827
25. Khin, K., Moe, A. Myint. M, Aung. K.P.: Dextral transgressional shearing and strike-slip
partitioning developments in the Central Myanmar Basin during the collision between the
India Plate and West Myanmar Block. Journal of Asian Earth Sciences. (1998). https://doi.org/
10.1016/j.jaesx.2021.100055
26. Socquet, C. Vigny, N. Chamot-Rooke, W. Simons, C. Rangin, B. Ambrosius. India and Sunda
plates motion and deformation along their boundary in Myanmar determined by GPSA. (2006)
Exploring an Alternate Perspective
of the Importance Factor for Seismic
Design of Structures
Narsiram Gurjar and Dhiman Basu
Abstract Contingent to the risk to human life in the event of failure structural
systems is designed with an additional factor, commonly known as the importance
factor (Cl 7.2.3) (IS-1893 (2016) Indian standard criteria for earthquake resistant
design of structures, Part-1: General provisions and buildings. In: Bureau of Indian
Standards, New Delhi.). In other words, importance factor is aimed to augment the
design seismic hazard to reduce the probability of failure of a structure contingent
on the associated risk to human life. Most standards recommend a factor (such
as 1.5, 1.2 and 1.0 by Indian standard) based on the importance of a structure. The
important question that arises here is, given the varying level of epistemic uncertainty
contingent on the fundamental period and regional seismicity, are we maintaining
the underlying objective of uniform mean annual rate of exceedance while using a
constant importance factor (such as that specified by most seismic standards)? This
study offers an alternate perspective of increasing the design seismic hazard while
maintaining the same underlying principle in an effort to comprehend the underlying
problem. The ratio of fractile-to-weighted mean hazard is envisioned in this study
as an alternate perspective of the importance factor used in seismic design, which
appears to be dependent on the regional seismicity and fundamental period (but not on
the soil sites). Besides, PSHA of the North-east Indian region is carried out using the
logic tree approach and the results of few cities are presented here for the purpose of
comparative illustrations. Finally, the importance factor is given by a linear function
of the time period with contour maps of location-dependent constants.
Keywords Importance factor · Weighted mean and fractile seismic hazard · Logic
tree · Contour map
N. Gurjar (B) · D. Basu

Department of Civil Engineering, Indian Institute of Technology Gandhinagar, Gandhinagar, India
e-mail: narsiram.gurjar@iitgn.ac.in
D. Basu
e-mail: dbasu@iitgn.ac.in
https://doi.org/10.1007/978-981-99-1459-3_51
640 N. Gurjar and D. Basu
1 Introduction
Contingent to the risk to human life in the event of failure (or impaired function-
ality!), the structural systems in ASCE 7–16 [1] are categorized into four groups with
Category-I and Category-IV indicating the lowest and highest risk levels, respectively
(Cl 1.5) [1]. Conditional probability of failure given the design shaking, i.e., the
target reliability is specified against each category along with an importance factor.
Eurocode 8 [2] imposes a target reliability through an importance factor against each
structure considering the risk of its failure or impaired performance (decided by the
National Authorities). Along the same line, the design seismic hazard per Indian
standard [3] considers the importance factor (Cl 7.2.3) [3] which has the interpre-
tation of accounting for the risk associated with failure (or impaired performance)
of the structure. In other words, importance factor is aimed to augment the design
force and, alternatively, the design seismic hazard to reduce the probability of failure
of a structure contingent on the associated risk to human life. Figure 1 explains a
schematic illustration of the need for importance factor. While aiming to augment the
design seismic force, most standards recommend a factor such as 1.5, 1.2 and 1.0 by
Indian standard based on the importance of the structure, and this paper explores the
alternate viewpoint of increasing the design seismic hazard. Assuming the ordinary
structures are designed with mean hazard (that corresponds to importance factor 1.0),
the important structures may be designed against ‘mean plus x times standard devia-
tion (μ + xσ )’ hazard, e.g., 84 and 99 percentiles. Although the choice of percentile
needs rigorous risk assessment, for the purpose of illustration, the ratio of 84.13th
percentile to weighted mean and 99.86th percentile to weighted mean is considered
as the two importance factors, I84 and I99 , respectively, in this paper.
Probabilistic seismic hazard assessment for seven states of North-east Indian
region is carried out using the logic tree approach for this purpose. Sample illus-
tration of results is reported in terms of different fractile and weighted mean hazard
(and their ratio) for some of the major cities of the region at National Earthquake
Hazard Reduction Program (NEHRP) soil site class-C (average Vs30 = 550 m/s)
with 2475 and 475 years of return period. Finally, it is recommended that the impor-
tance factor of a structure can be considered as time period and location dependent,
in terms of a ratio of fractile (importance level) to weighted mean hazard from site-
specific probabilistic seismic hazard assessment (PSHA), instead of a time period
and location-independent factor as per Indian standard recommendation. However,
decision of fractile levels for differently important structures is beyond the scope of
this study.
Exploring an Alternate Perspective of the Importance Factor for Seismic … 641
Statement! Earthquake NO! YES!

The buildings should of certain
be designed for DBE magnitude?
level of hazard Intensity measure
(IM) with x% PE
What is in y years (or re-
DBE? turn period)
Hazard level is fixed for a particular category of structures (say buildings)
Do all the building have

same importance (irrespec- NO!
tive of their use)? Different buildings have different im-
portance based on risk to human lives
What shall we in the event of failure
do?
Design the important struc-
Why?
tures for higher forces
Higher return ↓ Risk =

period? Higher forces ↓ Hazard (↑ Hazard level) x
NO! from where? ↓ Exposure (change location, NO!) x
(Already decided) ↓ Vulnerability (↓ probability of failure)
Present study (Alternate

IS 1893 (and most
perspective)
other standards)
Consider higher epistemic
Just use the importance
uncertainty in the seismic
factor accordingly
hazard characterization
How do we get
and ?
How do we get this?

Interesting? Why 1.2 and 1.5 only?
See you next No information!
Fig. 1 Schematic illustration of the need for importance factor

2 Probabilistic Seismic Hazard Assessment
Seismic hazard estimation is one of the important steps to quantify the possibility of
hazard at any particular site. The assessment provides a basis for proper planning and
safe design of important structures in a seismically exposed region. North-east Indian
region is one of the extreme seismically active regions all around the world. Prob-
abilistic seismic hazard assessment for seven states of North-east region is carried
out. Seismic hazard is characterized in terms of weighted mean and fractile repre-
sentation of hazard using the logic tree approach. Recurrence relation parameter,
magnitude and distance probability distribution, maximum magnitude and ground
motion predictive equations (GMPEs) are considered the source of epistemic uncer-
tainty. Detailed description of the hazard computation with GMPE rule and logic tree
for accounting epistemic uncertainty in seismic hazard characterization is explained
by Gurjar and Basu (2022) [4, 5], and only a brief description is provided here for
the ready reference.
2.1 Hazard Computation with GMPE Rule and Logic Tree
Logic tree diagram used for construction of model hazard curves (MHCs) in PSHA is
given in Gurjar and Basu [4]. The logic tree is mainly divided into two sections, i.e.,
Earthquake Rupture Forecast (ERF) and Sub-Earthquake Rupture Forecast (SERF).
The ERF consists of four subsections: (a) distance probability distribution as magni-
tude dependent and independent, (b) recurrence relation parameters as straight-line
fitting (linear regression) and log-likelihood, (c) magnitude distribution as given by
Gutenberg-Richter, (1944) [6] and Main and Burton [7] and (d) maximum magni-
tude as maximum of five alternatives from Kijko and Singh [8]; Gupta [9]; and
proportional to fault length. GMPE rule is defined in the SERF for the consideration
of epistemic uncertainty due to selection of GMPEs. The rate corresponding to the
intensity measure (IM) value is contributed by GMPE rule; i.e., all ruptures (m − r
pair) in qth source zone will be governed by ONLY vth of the aq number of possible
candidates for GMPE, whereas other ruptures in any other source zone (1, L but
= q) will be contributed from the weighted attributes of all possible GMPEs in the
respective source zone. Note that GMPE rule is defined for the SERFs and hence
remains same in all ERFs.
Weighted mean and fractile representation of hazards are estimated in this study
using the logic tree approach of PSHA. Sample illustration of the PSHA results is
reported in terms of the weighted mean and fractile (50, 84.13 or 99.86th percentile,
1.0E+01
1.0E+00 1.0E+00
1.0E-02 1.0E-01
1.0E-02
1.0E-04 1.0E-03
λ
λ
Mean
1.0E-06 1.0E-04 50th P
1.0E-05 84.13th P
99.86th P
1.0E-08 1.0E-06
0.01 0.1 1 0.01 0.1 1
Sa (g) Sa (g)
Fig. 2 Hazard curves at site class-C, T = 0 for Imphal West (Manipur)
2.0 2.0
50th P
84.13th P
1.5 1.5 99.86th P
Mean
Sa (g)
Sa (g)
1.0 1.0
0.5 0.5
0.0 0.0
0.01 0.1 1 0.01 0.1 1
Time Period (sec) Time Period (sec)
Fig. 3 Site-specific uniform hazard spectra (UHS) accounting for importance factor at site class-C,
10% PE in 50 years, for Imphal West (Manipur)
etc.) representation of hazard curve (Fig. 2) and site-specific UHS accounting for
importance factor (Fig. 3) for one city (Imphal West district, Manipur) of the North-
east region.
Although the choice of percentile needs rigorous risk assessment, for the purpose
of illustration, the ratio of 84.13th percentile to weighted mean and 99.86th percentile
to weighted mean is considered as the two importance factors, I84 and I99 , respec-
tively, in this paper. Table 1 features the variation of 475-years I84 against time period
at several locations for the site class-C. Also included are the similar data for I99 in
Table 2. Similar exercise is also carried out for the 2475-year return period but not
included in the table for brevity.
The importance factor I84 varies in the range of 1.1 ~ 2.1 for 2475-years return
period and 1.1 ~ 1.8 for 475 years return period over a period range of 0 ~ 3 s for
the entire North-east region (2302 site locations considered at 0.1° grid spacing).
Similarly, the I99 varies in the range of 1.2 ~ 3.8 and 1.2 ~ 3.1 for 2475 years
and 475 years return period, respectively. Further, these ratios are noted as nearly
insensitive to the site class. For example, a variation in the order of ± 0.2 is noted
Table 1 84.13th percentile-to-weighted mean hazard ratio (I84 ) variation with time period and location
644
S. No. City Lat (˚N) Long (˚E) 84.13th percentile/weighted mean

T=0s T = 0.1 s T = 0.5 s T=1s T=2s T=3s
1 Papum Pare 27.17 93.70 1.19 1.33 1.26 1.34 1.32 1.44
2 Changlang 27.76 96.62 1.12 1.26 1.17 1.26 1.29 1.35
3 Lohit 27.90 96.17 1.15 1.33 1.19 1.26 1.30 1.37
4 Kamrup Metropolitan 26.08 91.64 1.23 1.33 1.26 1.32 1.38 1.49
5 Nagaon 26.30 92.70 1.18 1.26 1.33 1.30 1.33 1.41
6 Dhubri 26.14 89.93 1.22 1.33 1.34 1.47 1.41 1.71
7 Imphal West 24.82 93.90 1.11 1.12 1.13 1.15 1.21 1.29
8 Senapati 25.32 94.15 1.12 1.14 1.14 1.15 1.24 1.34
9 Thoubal 24.64 94.00 1.11 1.11 1.12 1.15 1.18 1.26
10 East Khasi Hills 25.37 91.75 1.20 1.32 1.23 1.31 1.40 1.50
11 West Garo Hills 25.57 90.22 1.34 1.45 1.42 1.51 1.49 1.74
12 North Garo Hills 25.90 90.49 1.30 1.43 1.41 1.51 1.45 1.69
13 Aizawl 23.88 92.90 1.17 1.21 1.17 1.21 1.26 1.30
14 Lunglei 22.88 92.70 1.16 1.20 1.19 1.21 1.25 1.27
15 Champhai 23.45 93.18 1.15 1.14 1.15 1.17 1.22 1.27
16 Dimapur 25.76 93.78 1.19 1.19 1.16 1.21 1.27 1.37
(continued)
N. Gurjar and D. Basu
Table 1 (continued)
T=0s T = 0.1 s T = 0.5 s T=1s T=2s T=3s
17 Kohima 25.75 94.11 1.13 1.11 1.14 1.18 1.24 1.34
18 Mon 26.80 94.98 1.13 1.13 1.15 1.20 1.27 1.38
19 West Tripura 23.90 91.40 1.16 1.29 1.25 1.29 1.34 1.39
20 South Tripura 23.23 91.56 1.20 1.26 1.27 1.27 1.31 1.35
21 North Tripura 24.08 92.26 1.16 1.25 1.21 1.25 1.29 1.34
Exploring an Alternate Perspective of the Importance Factor for Seismic …
645
Table 2 99.86th percentile-to-weighted mean hazard ratio (I99 ) variation with time period and location
646

T=0s T = 0.1 s T = 0.5 s T=1s T=2s T=3s
1 Papum Pare 27.17 93.70 1.50 1.47 1.88 2.39 2.39 2.93
2 Changlang 27.76 96.62 1.31 1.48 1.44 1.96 2.10 2.70
3 Lohit 27.90 96.17 1.34 1.56 1.49 2.01 2.15 2.76
4 Kamrup Metropolitan 26.08 91.64 1.64 1.71 1.83 2.23 2.33 2.94
5 Nagaon 26.30 92.70 1.45 1.52 1.74 2.18 2.16 2.66
6 Dhubri 26.14 89.93 1.53 1.56 1.87 2.35 2.40 2.99
7 Imphal West 24.82 93.90 1.25 1.27 1.33 1.46 1.57 2.00
8 Senapati 25.32 94.15 1.28 1.30 1.36 1.50 1.64 2.13
9 Thoubal 24.64 94.00 1.26 1.27 1.31 1.39 1.51 1.86
10 East Khasi Hills 25.37 91.75 1.64 1.75 1.75 2.11 2.25 2.88
11 West Garo Hills 25.57 90.22 1.62 1.71 1.85 2.25 2.36 2.99
12 North Garo Hills 25.90 90.49 1.56 1.67 1.82 2.24 2.31 2.90
13 Aizawl 23.88 92.90 1.43 1.68 1.43 1.68 1.69 2.07
14 Lunglei 22.88 92.70 1.37 1.44 1.49 1.71 1.68 2.03
15 Champhai 23.45 93.18 1.35 1.35 1.43 1.56 1.60 1.91
16 Dimapur 25.76 93.78 1.41 1.56 1.40 1.72 1.83 2.41
(continued)
N. Gurjar and D. Basu
Table 2 (continued)
T=0s T = 0.1 s T = 0.5 s T=1s T=2s T=3s
17 Kohima 25.75 94.11 1.28 1.27 1.37 1.66 1.79 2.34
18 Mon 26.80 94.98 1.32 1.32 1.33 1.66 1.86 2.51
19 West Tripura 23.90 91.40 1.67 1.69 1.76 2.02 1.98 2.46
20 South Tripura 23.23 91.56 1.47 1.56 1.63 1.90 1.84 2.27
21 North Tripura 24.08 92.26 1.60 1.63 1.66 1.94 1.88 2.32
Exploring an Alternate Perspective of the Importance Factor for Seismic …
647
for both I84 and I99 , respectively. Hence, importance factor using this alternative
perspective depends on the fundamental period, seismicity of the site but not on the
site class.
The location- and time period-dependent importance factor for 475 years return
period can be represented as

I84 = M84 T + C84
f or T = 0 − 3 s
I99 = M99 T + C99
Here, M84 , C84 , M99 , and C99 are the location-dependent constants. The location-
dependent constants (site specific) can be determined by fitting a linear relation
as represented in Fig. 4 for East Khasi Hills district of Meghalaya. The contour
maps of location-dependent constant (M84 , C84 , M99 , and C99 ) for the entire North-
east region are determined in a similar manner and shown in Fig. 5a–d. Figure 5a–d
illustrates the spatial variability of location-dependent constants similar to the hazard
map of a region.
The observation presented above is somewhat startling with a chance of being
misconstrued in absence of a separate discussion. As a sample illustration let us
consider two buildings with fundamental period T = 1.0 s (~10 story), located at a
site with hard soil (NEHRP site class-C) in the East Khasi Hills (Meghalaya) area
of North-east India. Let both the structures be of different risk to human life in the
event of failure (or impaired functionality!), e.g., a business community center and
a hospital building. An importance factor of 1.2 and 1.5 should be considered for
the business community center and hospital building, respectively, per the Indian
3.0
84.13th P/ Mean
2.8 99.86th P/ Mean
Linear (84.13th P/ Mean)
2.6
Linear (99.86th P/ Mean)
I99 = 0.414T + 1.603
2.4
R² = 0.964
Percentile /Mean
2.2
2.0
1.8 I84 = 0.109T + 1.203

R² = 0.910
1.6
1.4
1.2
1.0
0 0.5 1 1.5 2 2.5 3 3.5
Time Period (sec)
Fig. 4 Site-specific period-dependent importance factor for 475 years return period, Site class-C
at East Khasi Hills (Meghalaya)
(a)
(b)
Fig. 5 Contour maps of location-dependent constants for 475 years return period, site class-C, a
M84, b C84, c M99, d C99
standard regardless of the fundamental period and regional seismicity. However,

defining the importance factor as I84 and I99 at the fundamental period, one may
show 1.31 and 2.02 (Fig. 4) for the business community center and hospital building,
respectively. Although at T = 0 s, I84 = 1.2 and I99 = 1.6 for East Khasi Hills
are nearly same as Indian standard, a significant effect of this new perspective of
importance factor on the design spectra (specially at higher time period) can be
clearly witnessed from Fig. 6. Hence, the consideration of a constant importance
factor may underestimate the design forces, especially for the high-rise buildings
(c)
(d)
Fig. 5 (continued)
with higher fundamental period. The step-by-step procedure for the computation
of alternate perspective of the importance factor can best be explained through a
flowchart as presented in Fig. 7.
Greater importance factor per this new perspective implies larger epistemic uncer-
tainty in the seismic hazard characterization and not the higher return period. Given
the varying level of epistemic uncertainty contingent on the fundamental period
and regional seismicity, a constant importance factor such as that specified by most
0.6 0.8
IS1893 IS1893
0.7
0.5 IS1893 ꓫ 1.2 IS1893 ꓫ 1.5
IS1893 ꓫ I84 0.6 IS1893 ꓫ I99
0.4 0.5
Sa (g)
Sa (g)
0.3 0.4
0.2 0.3
0.2
0.1 0.1
0.0 0.0
0 1 2 3 0 1 2 3
Time Period (sec) Time Period (sec)
Fig. 6 Design spectra at East Khasi Hills (Meghalaya) with different importance factors
seismic standards may defeat the underlying objectivity. Further, a constant impor-
tance factor also defeats the objective of uniform mean annual rate of exceedance
if higher mode participation is non-negligible. Based on the discussion presented
above, the fractile hazard definition proposed in this paper scientifically accounts for
the objectives of importance factor though choice of appropriate fractile contingent
on the importance of structure requires a rigorous risk assessment, which is beyond
the scope of this study.
4 Summary and Conclusion
The importance factor is aimed to augment the design seismic hazard to reduce
the probability of failure of a structure contingent on the associated risk to human
life. A constant importance factor in seismic design recommended by most seismic
standards may defeat the underlying objectivity of uniform risk-based seismic
design. An alternate perspective while maintaining the same underlying principle
is recommended in this paper that defines the ratio of fractile-to-weighted mean-
representation of hazard as the importance factor, which gives an important infor-
mation about location and time period dependent importance factor, rather than
a constant. Choice of appropriate fractile however needs rigorous risk assessment
which is beyond the scope of this paper.
PSHA of the North-east Indian region is carried out using logic tree approach with
due consideration of epistemic uncertainty contributed from several sources, and the
results of few cities are presented here for the purpose of comparative illustrations.
Results are reported in terms of different fractile and weighted mean hazard (and
their ratio) for some of the major cities of the region. The location- and time period-
dependent importance factor for 475 years return period is defined as a linear function
of time period and contour maps of location-dependent constants.
STEP-1
Carry out the site specific
PSHA using logic tree ap-
proach STEP-2
Compute the model haz-
ard curves (at different
time periods) resulting
from each logic tree
branch
STEP-4
Compute the cumulative STEP-3
distribution function of the Compute the site specific
IM vector at each time pe- UHS from horizontal dis-
riod considering cumulative section of each model haz-
branch factor (from logic ard curve at a particular
tree) as the cumulative prob- return period (say DBE)
ability
STEP-5 STEP-6
Compute the weighted Take the ratio of different frac-
mean and fractile repre- tiles (say 84th or 99th) to weighted
sentation of site specific mean and fit a linear relation, to
UHS get and in terms of a
Note: The site specific function of the time period
UHS accounting for im-
portance factor can also
be used for design
STEP-7
Repeat the STEP-1 to -6 at all the
possible locations of the region and
compute the location dependent
constants of the linear relation (
STEP-8 or ) and plot the contour map
Compute the site-specific time
period dependent importance fac-
tor ( or ) and multiply
with the code specific design
spectra
Fig. 7 Flowchart for alternate perspective of the importance factor
Acknowledgements This research is funded by the Ministry of Earth Sciences, Seismology Divi-
sion, Government of India, under the Grant No. MoES/P.O. (Seismo)/1(370)/2019, and the financial
support is acknowledged.
References
1. American Society of Civil Engineers: Minimum design loads and associated criteria for buildings
and other structures. ASCE/SEI 7, Reston, VA (2016)
2. EN 1998–1: Eurocode 8: Design of structures for earthquake resistance—Part 1: General rules,
seismic actions and rules for buildings (2004)
3. IS-1893: Indian standard criteria for earthquake resistant design of structures, Part-1: General
provisions and buildings. In: Bureau of Indian Standards, New Delhi (2016)
4. Gurjar, N., Basu, D.: Epistemic uncertainty in PSHA and seismic hazard characterization using
logic tree approach: part I-developing the framework. Pure Appl. Geophys. (2022). https://doi.
org/10.1007/s00024-022-03143-4
logic tree approach: part II-implementation over North-East India. Pure Appl. Geophys. (2022).
https://doi.org/10.1007/s00024-022-03148-z
6. Gutenberg, B., Richter, C.F.: Frequency of earthquakes in California. Bull. Seism. Soc. Am. 34,
185–188 (1944)
7. Main, I.G., Burton, P.W.: Information theory and the earthquake frequency-magnitude distribu-
tion. Bull. Seism. Soc. Am. 74, 1409–1426 (1984)
8. Kijko, A., Singh, M.: Statistical tools for maximum possible earthquake magnitude estimation.
Acta Geophys. 59, 674–700 (2011)
9. Gupta, I.D.: The state-of-the-art in seismic hazard analysis. ISET J. Earthquake Technol. 39(4),
311–346 (2002)
Simplified Damping Modification Factor
for Vertical Response Spectra
Ravi Kanth Sriwastav and Dhiman Basu
Abstract Damping modification factor of vertical response spectra (DMFV ) for

1%, 2%, 8% and 10% damping levels with that at 5% of the critical is explored for
a period range of 0–4.0 s. The dependence of DMFV on seismological parameters
(magnitude, epicentral distance and soil type) is studied and is shown to be weak
and insignificant for practical purposes. Simplified relation for computing DMFV as
a function of the period is proposed irrespective of any seismological parameters.
Keywords Vertical earthquake shaking · Damping modification factor
1 Introduction
Significant effect of the vertical ground motion on the performance of certain rein-
forced concrete (RC) structures has been observed in the past [6, 9, 12–15]. The
effects of vertical components in seismic design can be included by defining a design
vertical response spectrum. Notable progress has been made over the past decade in
the characterization of vertical ground motion.
The procedures for construction of the vertical spectrum available in the literature
can be grouped into two broad categories. The ground motion predictive model
(GMPM) for direct construction for the vertical response spectrum is referred to as
the first type. The second type refers to median prediction of the vertical-to-horizontal
spectral ratio (V/H) through a GMPM. The product of V/H and the corresponding
horizon tal spectrum is used to calculate the vertical spectrum. Another type of
vertical spectrum constructed by simplifying the first two types was also proposed in
an earlier work [5]. Most building codes (For example, IS1893:2016 Part 1, ASCE
7:16 [2] and EC8-part 1: 2008 [7]) have recognized the effect of vertical components
R. K. Sriwastav (B) · D. Basu

Department of Civil Engineering, Indian Institute of Technology Gandhinagar, Gandhinagar, India
e-mail: ravi.sriwastav@iitgn.ac.in
D. Basu
e-mail: dbasu@iitgn.ac.in
https://doi.org/10.1007/978-981-99-1459-3_52
656 R. K. Sriwastav and D. Basu
and provide a simple recommendation for constructing the design vertical spectrum
with the design horizontal spectrum as a basis. Recently, Sriwastav and Basu [18]
proposed a simplified V /H spectrum using a large database of recorded events for the
computation of vertical spectra for both scenario- and intensity-based assessments.
Most vertical response spectra proposed in the literature and that recommended
by the design codes are for a damping ratio (ξ ) of 5%. However, the vertical modes in
the structures are relatively stiffer. It is anticipated that the damping ratios related to
these modes would differ from the horizontal modes. Therefore, a factor is generally
needed to convert the 5% damped vertical spectral ordinate to that associated with an
arbitrary damping ratio. Such a factor may be referred to as the vertical damping modi-
fication factor (DMF). Design codes specify a constant DMF for different damping
ratios irrespective of period and seismological parameters. As an example, Eurocode
8-Part 1 (2004) [Clause 3.2.2.2(3)]
/ / specifies the DMF for both horizontal and vertical
components as DMF = 10 (5 + ξ )(≥ 0.55). ASCE 7:16 (Table 17.5-1) recom-
mended a range of DMF between 1.25 and 0.5 for damping ratios up to 50% of the
critical. The possible dependence on the seismological parameters and period of the
structure is thus not accounted for in such grossly simplified recommendations.
Few recent studies including Akkar et al. [1], Rezaeian et al. [17] and Xiang and
Huang [19] extensively addressed the issues pertaining to the DMF of vertical compo-
nent (DMFV ). Akkar et al. [1] developed DMFV model to modify the 5%-damped
vertical pseudo-spectral acceleration (PSA) of Akkar et al. [1] for generating PSA at
other damping levels between 1 and 50%. The dataset in this study consisted of 1041
accelerograms from 221 shallow crustal earthquakes recorded in the Mediterranean
region and the Middle East. Mean DMFV was proposed as a function of M W , RJB
and Vs30, damping ratio and period. Rezaeian et al. [17] also developed a model for
DMFV for a range of damping ratios between 0.5% and 30%. Ground motion records
from the PEER NGA West2 database [3] were used, and median DMFV was shown
to depend upon the magnitude, distance, damping ratio and spectral period (but not
Vs30). Xiang and Huang [19] proposed a regional (Japanese) DMFV utilizing 3198
ground motion records from the Japanese seismic database. The range of damping
ratio studied were 0.5%, 1%, 2%, 3%, 4%, 6%, 7%, 8%, 9%, 10%, 12.5%, 15%,
17.5%, 20%, 25%, 30%, 35% and 40%. The peaks (for ξ less than 5%) or valleys
(for ξ more than 5%) of DMFV spectra were observed at a spectral period of 0.12 s.
It further deviates from unity with the increase in earthquake magnitude and epicen-
tral distance. The dependence of hypocenter depth and site category on DMFV was
reported to follow no distinct trend. [Therefore, there is no clear consensus on the
dependence of DMFV on seismological parameters.]
Additionally, the DMFV models in the literature are too complex to use in design
practice, and hence, a simplified DMFV model is needed with a sufficient level of
accuracy. This paper tries to develop such a simplified model based on a large database
of recorded events. We took five damping ratios namely 1%, 2%, 5%, 8% and 10%
of the critical to study the dependence of seismological parameters on DMFV using
a comparatively larger dataset. A simplified DMFV model for these damping levels
Simplified Damping Modification Factor for Vertical Response Spectra 657
is also proposed. Following section discusses the ground motion dataset used in the
paper.
2 Seismic Events and Ground Motions Considered
The ground motion database in this paper is similar to that used in Sriwastav and
Basu [18]. A set of 5962 records from PEER NGA West2 database with (i) MW 5.0–
8.0; (ii) epicentral distance (R)-0–200 km and (iii) Vs30-180–1500 m/s have been
selected. This dataset is classified into a number of triplets defined by M-R-Vs30.
An interval of 0.5 is used to construct magnitude bins. Six distance bins are formed
namely (i) 0 < R ≤ 20 km; (ii) 20 < R ≤ 50 km; ; (iii) 50 < R ≤ 75 km; (iv)
75 < R ≤ 100 km; ; (v) 100 < R ≤ 150 km and (vi) 150 < R ≤ 200 km. Three
bins for site category are constructed per NEHRP soil classification [8]; (i) medium
soil (180–360 m/s), (ii) hard soil (360–760 m/s) and (iii) rock (760–1500 m/s). A
total of 76 triplets are formed. The absolute acceleration spectrum is considered as
IM of the vertical component.
3 Median Vertical Spectrum for Scenario-Based

Assessment
Let the vertical response spectra corresponding to damping ratio ‘ξ ’ and that at 5% of
the critical are denoted by Vξ and V0.05 , respectively. Also, assume that the spectral
ordinates (at any period T ) conditioned to an M-R (magnitude-distance) bin follows a
lognormal distribution and logarithmic mean of Vξ and V0.05 are represented by μln Vξ
and μln V0.05 , respectively. Thus, vertical spectral ordinate corresponding to damping
ratio ‘ξ ’ at any period can be computed as
Vξ (T ) Vξ /
Vξ (T ) = V0.05 (T ) = V0.05 (T ) ⇒ ln Vξ = ln V0.05 + ln Vξ V0.05 .
V0.05 (T ) V0.05
(1)
/
It follows from Eq. (1) that ln Vξ V0.05/ is also normally distributed with
μln(Vξ / V0.05 ) as the logarithmic mean of ln Vξ V0.05 . Further, μln Vξ (T ) is given by
μln Vξ (T ) = μln V0.05 (T ) + μln(Vξ / V0.05 ) (2)

/
Denoting θ(Vξ / V0.05 ) as the median of Vξ V0.05 with ln θ(Vξ / V0.05 ) = μln(Vξ / V0.05 ) ,
it may be calculated for a given M-R bin as
/
θ(Vξ / V0.05 )(T ) = ex p E ln Vξ V0.05
/
= ex p(E[ln Vξ ] ex p(E[ln(V0.05 )]
/
⇒ θ(Vξ / V0.05 )(T ) = θVξ (T ) θV0.05 (T ) (3)
where E[·] denotes the first moment. The ratio of median vertical response spectrum
/ level (θVξ (T ) ) to that at 5% of critical (θV0.05 (T ) ) is, therefore, the same
at any damping
as median Vξ V0.05 spectrum.
3.1 Damping Modification Factor Using a Set of Recorded

Events
Damping modification factor for vertical shaking (DMFV ) is defined here as the ratio
of median vertical spectrum at any damping level (θVξ (T ) ) to that at 5% of critical
/
D M FV (T ) = θVξ (T ) θV0.05 (T ) (4)
The process of constructing DMFV is briefly discussed here. Given an M-R-Vs30

triplet, the median vertical response spectra at five damping levels (1%, 2%, 5%, 8%
and 10%) are constructed. Next, the DMFV is computed for each damping ratio. This
is followed for all the M-R-Vs30 triplets. Examples are presented for a few triplets
covering a different range of magnitude and distance for medium, hard and rock soil
sites in Figs. 1, 2 and 3, respectively. The median vertical spectrum for all damping
ratios is presented in the left panel of the figures. The right panel presents the DMFV
for the associated damping ratios. The shape of DMFV for a particular damping ratio
is observed to be nearly the same regardless of the triplet characteristics.
0.50 1.80
1% 1%
0.40 2% 1.60 2%
5% 5%
8% 1.40 8%
0.30
10% 10%
Sa/g
1.20
DMF
0.20
1.00
0.10 0.80
0.00 0.60
0.001 0.01 0.1 1 0.001 0.01 0.1 1
Period (s) Period (s)
Fig. 1 Vertical response spectra (left) and damping modification factor (DMFV ) (right) for a triplet
defined by M: 6.5–7.0 R: 20–50 km Vs30: 180–360 m/s [medium soil sites]
0.30 1.80
1% 1%
2% 1.60 2%
5% 5%
0.20 8% 1.40 8%
10% 10%
Sa/g
1.20
DMF
0.10 1.00
0.80
0.00 0.60
0.001 0.01 0.1 1 0.001 0.01 0.1 1
defined by M: 7.0–7.5 R: 50–75 km Vs30: 360–760 m/s [hard soil sites]
0.06 1.80
1% 1%
2% 1.60 2%
5% 5%
0.04 8% 1.40 8%
10% 10%
DMF
Sa/g
1.20
0.02 1.00
0.80
0.00 0.60
0.001 0.01 0.1 1 0.001 0.01 0.1 1
defined by M: 6.0–7.0 R: 50–100 km Vs30: 760–1500 m/s [rock sites]
4 Dependence of Damping Modification Factor

on Seismological Parameters
Guided by the similarity of DMFV shape irrespective of triplet characteristics, sepa-

rate comparisons are performed in this section to verify the dependence of DMFV on
M, R and Vs30. Figures 4 and 5 present a few representative plots of such compar-
ison. The plots are shown separately for the four damping levels (1%, 2%, 8% and
10%) in each figure. A comparison of the DMFV for different magnitudes given a
particular distance and site class is presented in Fig. 4. Similarly, a comparison of
DMFV for distances given a magnitude and site class is conducted. However, it is not
presented for brevity. Weak dependence of DMFV on the magnitude and distance is
observed. Figure 5 presents the comparison against soil sites regardless of magnitude
and epicentral distance. DMFV is observed to (i) decrease with Vs30 for ξ < 0.05 and
(ii) increase with Vs30 for ξ > 0.05. However, this effect of soil category on DMFV
is small. Therefore, for all practical purposes, it may be regarded as insignificant.
1.9 1.6
1.6
1.3
DMF
DMF
1.3
1.0
1.0 M5.25 M5.75 M5.25 M5.75
M6.25 M6.75 M6.25 M6.75
M7.25 M7.75 M7.25 M7.75
0.7 0.7
0.001 0.01 0.1 1 0.001 0.01 0.1 1
a) 1% damping R:20-50;Vs30:180-360 b) 2% damping R:100-150;Vs30:180-360
1.2 1.2
M5.25 M5.75 M5.25 M5.75
M6.25 M6.75 M6.25 M6.75
1.0 M7.25 M7.75 1.0 M7.25 M7.75
DMF
DMF
0.8 0.8
0.6 0.6
0.001 0.01 0.1 1 0.001 0.01 0.1 1
c) 8% damping R:50-75; Vs30:180-360 d) 10% damping R:75-100km; Vs30: 180-360
Fig. 4 Dependence of DMFV on earthquake magnitude for a 1%, b 2% and c 8% d 10%
5 Simplified Damping Modification Factor for V/H
Since, DMFV is observed to be weakly dependent on all the seismological parameters

considered here (M, R and Vs30), all the triplets are clubbed and the DMFV is
computed to study its relationship with the period for each damping level. DMFV
spectra for all damping levels have a distinct shape and may be defined through
their shape in three period ranges. Following simplified relation is observed to best
represent the DMFV computed using the recorded ground motions
| |
| 1.0 T ≤ 0.02s |
| |
|
DMFV (T ) = | c1 + c2 e c3 T
0.02s <T ≤ 0.2s ||. (5)
| c1 + c4 T T > 0.2s |
Here, c1 , c2 , c3 and c4 are the regression coefficients. These are provided in Table 1
for all damping levels. Figure 6 presents the comparison of DMFV computed using
recorded events and that generated using simplified expression per Eq. (5). The
difference is negligible for all practical purposes. Therefore, using the framework
followed in this paper, a simplified DMFV model for a wide range of damping ratios
may be developed for possible recommendation in the seismic code.
1.8 1.4
1.6
1.2
1.4
DMF
DMF
1.2 Medium Soil 1% Medium Soil 2%
1.0
1.0 Hard Soil 1% Hard Soil 2%
Rock 1% Rock 2%
0.8 0.8
0.001 0.01 0.1 1 10 0.001 0.01 0.1 1 10
a) 1% damping b) 2% damping
1.2 1.2
Medium Soil 8% Medium Soil 10%
Hard Soil 8% Hard Soil 10%
1.0 Rock 8% 1.0 Rock10%
DMF
DMF
0.8 0.8
0.6 0.6
0.001 0.01 0.1 1 10 0.001 0.01 0.1 1 10
c) 8% damping d) 10% damping
Fig. 5 Dependence of DMFV on site class for a 1%, b 2% and c 8% d 10%
Table 1 Regression coefficient for damping modification factor of vertical component

Damping ratio (%) C1 C2 C3 C4
1 1.654 −1.59 −41.84 −0.098
2 1.362 −0.67 −30.72 −0.046
8 0.842 0.26 −24.76 0.015
10 0.777 0.38 −26.40 0.017
1.75
1%
1% fitted
1.50
2%
DMF
2% fitted
1.25
8%
1.00 8% fitted
10%
0.75 10% fitted
0.001 0.01 0.1 1 10
Period (s)
Fig. 6 Recorded and fitted median damping modification factor for four damping levels; 1%, 2%,
8% and 10%
6 Assessment of Proposed Damping Modification Factor
The proposed model is compared with three DMFV GMPMs available in the liter-
ature namely Akkar et al. [1] (AK13), Rezaeian et al. [17] (RZ14) and Xiang and
Huang [19] (XH18). Taking an example triplet defined by M: 5.5–6.0 R: 100–150 km
Vs30: 180–360 m/s, 5% damped median vertical spectrum is computed using ground
motions from PEER Database [3]. Also computed is the 1% damped median vertical
spectrum. Utilizing the 5% damped median vertical spectrum, the corresponding 1%
damped spectrum is computed using the proposed model and that per AK13, RZ14
and XH18. Figure 7a presents the comparison of these predicted median vertical
spectra. Also included in the comparison are the vertical spectra computed per EC8-
part 1 and ASCE 7–16 recommendations. A similar comparison is provided for 2%,
8% and 10% damping ratios in Fig. 7b–d, respectively.
Different inferences are observed from the above comparisons for damping ratios
less than and greater than 5%. For damping ratios less than 5%, both the design codes
underestimate the median vertical spectra for period greater than 0.03 s, whereas the
prediction is overestimated for shorter periods. However, for damping ratios greater
than 5%, the codal prediction is significantly lower than the recorded spectrum for
0.09 0.08
Rec
Pre 0.06
0.06 EC8
PEER
Sa/g
Sa/g
0.04
0.03 AK13
RZ14 0.02
XH18
0.00 0.00
0.0001 0.001 0.01 0.1 1 0.001 0.01 0.1 1
0.05 0.04
0.04
0.03
0.03
Sa/g
0.02
Sa/g
0.02
0.01
0.01
0.00 0.00
0.001 0.01 0.1 1 0.001 0.01 0.1 1
c) 8% damping d) 10% damping
Fig. 7 Assessment of the predicted median vertical spectrum for different damping ratios with
GMPMs (keeping the recorded median vertical spectrum as the reference) for example triplet M:
5.5–6.0 R: 100–150 km Vs30: 180–360 m/s (constructed using PEER database)
0.09 0.08
Rec
Pre
0.06
0.06 EC8
Sa/g
PEER
Sa/g
0.04
0.03 AK13
RZ14 0.02
XH18
0.00 0.00
0.0001 0.001 0.01 0.1 1
0.001 0.01 0.1 1
Fig. 8 Assessment of the predicted vertical spectrum for different damping ratios with GMPMs
(keeping the recorded median vertical spectrum as the reference) for example triplet M: 5.5–6.0 R:
50–100 km Vs30: 180–360 m/s (constructed using PESMOS-COSMOS database)
period less than 0.03 s, whereas the prediction is comparatively better for longer
periods. The median spectrum computed using the GMPMs and the proposed models
are nearly equal and closely match the recorded median vertical spectrum for all
damping levels. A similar comparison with (for the same triplet) is provided in Fig. 8
using the ground motions from PESMOS [15] and COSMOS [4] databases which
are not utilized in the development of the DMFV model. Only 1% and 10% damped
spectra are demonstrated for brevity. The above inferences also hold for this dataset.
7 Demonstration of Utility of Damping Modification Factor
The proposed damping modification factor may be used for computing the median
vertical spectrum for both scenario- and intensity-based assessments given the asso-
ciated 5% damped spectrum. The utility of damping modification factor is demon-
strated here through two examples: (i) site-specific spectrum and (ii) code-based
spectrum.
7.1 Site-Specific Spectrum
The PSHA of the horizontal component performed by Gurjar and Basu [11, 10]
for Northeast India is revisited for computing the uniform hazard spectrum (UHS)
and conditional mean spectrum (CMS) using the generalized causal rupture (GCR)
approach [11, 10]. Two cities namely Golaghat in Assam and Noney in Manipur are
selected and the ‘weighted mean representation of UHS and GCR-CMS (hereafter
referred to as UHS and H-CMS, respectively)’ at a conditional period of 1.25 s are
computed in hard soil site (Vs30: 360–760 m/s) for the 2475 year return period using
the logic tree approach with GMPE rule.
V/H spectra are computed for all the generalized causal rupture scenarios (mean
M-R) contributing to the H-CMS using [18]. Weighted mean representation of V/H
spectrum is computed using GCR and logic tree weights applied to each V/H spectra.
The 5% damped vertical spectrum consistent with H-CMS (denoted by V_H-CMS)
is computed as the product of weighted mean representation of V/H spectrum and
H-CMS. This vertical spectrum is denoted as ‘vertical spectrum consistent with
horizontal seismic hazard’.
V_H-CMS is validated through a suite of ground motions selected and scaled to
match the H-CMS. Two period ranges namely i) 0.1 s–1.5T H and ii) 0.1 s–2.0T H are
studied, where T H (=1.25 s) is the fundamental horizontal period.
The same set of scale factors is utilized to scale the associated vertical components
of the suite as well. Next, the average of scaled vertical spectra (Avg_V ) and the
average of scaled horizontal spectra (Avg_H) are computed. The ratio of average
vertical-to-average-horizontal spectra, (denoted by V /H), are computed for both the
period ranges and are compared with the weighted mean representation of V /H
spectrum in Fig. 9a, c for Golaghat and Noney cities, respectively. The comparison
of the average vertical spectra with V_H-CMS is also presented in Fig. 9b, d for these
cities. The use of the wider period range (0.1–2.0T H ) is seen to better represent the
vertical spectrum consistent with the horizontal seismic hazard (V_H-CMS). This
is expected as the spectral shape of H-CMS is representative of the hazard and the
ground motions selected using a wider period range better represent the target.
With V_H-CMS as the 5% damped vertical spectrum, the vertical spectra associ-
ated with other damping ratios are computed utilizing the proposed DMFV model.
A comparison of these vertical spectra is provided in Fig. 10a, b for Golaghat and
Noney cities. In another example, the intensity-based vertical spectrum is computed
using the product of horizontal UHS and the intensity-based (envelope) V/H spec-
trum proposed by Sriwastav and Basu [18]. The proposed DMFV model is again
utilized to construct the vertical spectrum corresponding to other damping ratios. A
comparison of these spectra is provided in Fig. 11a, b for these two cities.
7.2 Code Specific Spectrum
The selected cities (Golaghat and Noney) lie in the same seismic zone per the zonation
map of IS1893:2016 (Part 1). Thus, the design horizontal spectrum is the same
for both cities. Design vertical spectrum is computed using the product of this 5%
damped horizontal spectrum (for hard soil site) and the intensity-based (envelope)
V/H spectrum per Sriwastav and Basu [18]. Vertical spectra corresponding to other
damping ratios are also computed using the proposed DMFV model and compared
in Fig. 12.
0.90 0.28
0.75
0.21
0.60
Sa/g
Sa/g
0.45 0.14
0.30
V/H [Weighted mean] 0.07 V_H-CMS
0.15 V/H (0.1 s ~ 1.5TH) Avg_V (0.1 s ~ 1.5TH)
V/H (0.1 s ~ 2.0TH) Avg_V (0.1 s ~ 2.0TH)
0.00 0
0 1 2 3 4 0 1 2 3 4
a) V/H spectrum for Golaghat-Assam b) Vertical spectrum for Golaghat-Assam
0.80 0.24
0.60 0.18
Sa/g
0.40 Sa/g 0.12
0.20 V/H [Weighted Mean] 0.06 V_H-CMS

V/H (0.1 s ~ 1.5TH) Avg_V (0.1 s ~ 1.5TH)
V/H (0.1 s ~ 2.0TH) Avg_V (0.1 s ~ 2.0TH)
0.00 0
0 1 2 3 4 0 1 2 3 4
c) V/H spectrum for Noney-Manipur d) Vertical spectrum for Noney-Manipur
Fig. 9 Comparison of average vertical-to-horizontal spectrum with the weighted mean representa-
tion of V/H (left panel) and comparison of average vertical spectrum with the H-CMS (right panel).
Period range in the legend denote the range used in ground motion section/scaling
0.50 0.50
1%_V 1%_V
2%_V 2%_V
0.40 0.40
5%_V (V_H-CMS) 5%_V (V_H-CMS)
8%_V 8%_V
0.30 10%_V 0.30
Sa/g
10%_V
Sa/g
0.20 0.20
0.10 0.10
0.00 0.00
0 1 2 3 4 0 1 2 3 4
a) City 1: Golaghat-Assam b) City 2: Noney- Manipur
Fig. 10 Comparison of the vertical spectrum for different damping levels where 5% damped vertical
spectrum is computed as the product of (weighted mean representation of) CMS and envelope of
scenario-based V/H spectrum [18]
2.80 2.80
1%_V 1%_V
2%_V 2%_V
2.10 5%_V (UHS * V/H) 2.10 5%_V (UHS * V/H)
8%_V 8%_V
1.40 10%_V 1.40 10%_V
Sa/g
Sa/g
0.70 0.70
0.00 0.00
0 1 2 3 4 0 1 2 3 4
a) City 1: Golaghat-Assam b) City 2: Noney- Manipur
Fig. 11 Comparison of the vertical spectrum for different damping levels where 5% damped vertical
spectrum is computed as the product of (weighted mean representation of) UHS and intensity-based
(envelope) V/H spectrum [18]
Fig. 12 Comparison of the 1.50

1%_V
vertical spectrum for 2%_V
different damping levels 1.20
5%_V_Pre
where 5% damped vertical
0.90 8%_V
spectrum is computed as the 10%_V
Sa/g
product of code-based 0.60

horizontal spectrum per
IS1893:2016 (Part 1) and 0.30
intensity-based (envelope)
V/H spectrum [18] 0.00
0 1 2 3 4
Period (s)
8 Summary and Conclusion
Damping modification factor of vertical response spectra (DMFV ) for 1%, 2%, 8%
and 10% damping levels with that at 5% of the critical is explored for a period
range of 0–4.0 s. DMFV is studied for 76 triplets, each defined by a range of M-
R and Vs30. The dependence of DMFV on seismological parameters (magnitude,
epicentral distance and soil type) is observed to be weak and insignificant for all
practical purposes. Simplified relations for computing DMFV as a function of the
period is proposed for damping levels 1%, 2%, 8% and 10% irrespective of any
seismological parameter. The proposed DMFV is demonstrated to be used for both
scenario- and intensity-based assessments. A simplified model using the framework
adopted in this paper may be developed for a larger range of damping factors for
possible recommendation in the seismic codes.
Acknowledgements This research is funded by the Ministry of Earth Sciences, Seismology Divi-
sion, Go I, under the Grant No. MoES/P.O. (Seismo)/1(370)/2019 and the financial support is
acknowledged.
References
1. Akkar, S., Sandıkkaya, M.A., Ay, B.Ö.: Compatible ground-motion prediction equations for
damping scaling factors and vertical-to-horizontal spectral amplitude ratios for the broader
Europe region. Bull. Earthq. Eng. 12(1), 517–547 (2014)
2. American Society of Civil Engineers: Minimum design loads and associated criteria for
buildings and other structures. ASCE/SEI 7, Reston, VA (2016)
3. Ancheta, T.D., Darragh, R.B., Stewart, J.P., Seyhan, E., Silva, W.J., Chiou, B.S.J., Wooddell,
K.E., Graves, R.W., Kottke, A.R., Boore, D.M., Kishida, T.: NGA-West2 database. Earthq.
Spectra 30(3), 989–1005 (2014)
4. Archuleta, R.J., Steidl, J., Squibb, M.: The COSMOS Virtual Data Center: A web portal for
strong motion data dissemination. Seismol. Res. Lett. 77(6) (2006)
5. Bozorgnia, Y., Campbell, K.W.: The vertical-to-horizontal response spectral ratio and tentative
procedures for developing simplified V/H and vertical design spectra. J. Earthq. Eng. 8(02),
175–207 (2004)
6. DiSarno, L., Elnashai, A.S. Manfredi, G.: Seismic Response of RC Members Subjected to the
2009 L’Aquila (Italy) Near-Field Earthquake Ground Motions. Report No. 10-01, Mid-America
Earthquake Center (2010)
7. EN 1998-1: Eurocode 8: design of structures for earthquake resistance–Part 1: General rules,
seismic actions and rules for buildings (2004)
8. FEMA 450.: NEHRP recommended provisions for seismic regulations for new buildings and
other structures-Part 1: provisions. Building seismic safety council. Washington, DC
9. Ghobarah, A., Elnashai, A.S.: Contribution of vertical ground motion to the damage of RC
buildings. In: Proceedings of the Eleventh European Conference on Earthquake Engineering
(pp. 9–13). AA Balkema, Rotterdam (1998)
the logic tree approach: part II, Implementation over North-East India. Pure Appl. Geophys.
(2022)
the logic tree approach: part I, Developing the framework. Pure Appl. Geophys. (2022)
12. Hosseinpour, F., Abdelnaby, A.E.: Effect of different aspects of multiple earthquakes on the
nonlinear behaviour of RC structures. Soil Dyn. Earthq. Eng. 92, 706–725 (2017)
13. Kim, S., Kim, S.J., Chang, C.: Analytical assessment of the effect of vertical ground motion on
RC frames designed for gravity loads with various geometric configurations. Adv. Civil Eng.
(2018)
14. Mwafy, A.M., Elnashai, A.S.: October. Vulnerability of code-compliant RC buildings under
multi-axial earthquake loading. In: 4th International Conference on Earthquake Engineering,
Taiwan, Paper No. 115 (2006)
15. PESMOS. Department of Earthquake Engineering, IIT Roorkee, https://pesmos.com/ [last
visited 10/02/2020]
16. Papazoglou, A.J., Elnashai, A.S.: Analytical and field evidence of the damaging effect of
vertical earthquake ground motion. Earthq. Eng. Struct. Dynam. 25(10), 1109–1137 (1996)
17. Rezaeian, S., Bozorgnia, Y., Idriss, I.M., Abrahamson, N.A., Campbell, K.W., Silva, W.J.:
Damping scaling factors for vertical elastic response spectra for shallow crustal earthquakes in
active tectonic regions. Earthq. Spectra 30(3), 1335–1358 (2014)
18. Sriwastav, R.K., Basu, D.: Vertical spectra consistent with horizontal seismic hazard. Soil Dyn.
Earthq. Eng. 157, 107242 (2022)
19. Xiang, Y., Huang, Q.L.: Damping modification factor for the vertical seismic response
spectrum: a study based on Japanese earthquake records. Eng. Struct. 179, 493–511 (2019)
December 01, 2020, Haridwar,
Earthquake: Fault Plane Solution
and Tectonic Implications
Pooja Mahto and S. C. Gupta
Abstract On December 01, 2020, a small-sized earthquake (M L = 4.0) occurred

25 km northwest of Haridwar after a gap of about 45 years. The location of the
earthquake falls on Main Frontal Thrust (MFT) about 105 km southwest of the 1991
Uttarkashi earthquake (M w = 6.8). The earthquake was sourced in the crust-mantle
boundary (40.6 km). Data from the local broadband seismic network around the Tehri
region have been used for the study of the source mechanism. The fault plane solution
of this event suggests that deformation occurred by strike-slip faulting with a signifi-
cant reverse component. The trend/plunge of the major and minor principal axes P/T
are 201°/10° and 296°/24°, respectively. This paper is based on the determination
of the focal mechanism through the moment tensor inversion solution of waveforms
derived from the records of the local event, employing ISOLA software. The moment
tensor solution is considered a stable and more robust method to quantify the fault
orientation of faults that slipped due to an earthquake and the method yields more
accurate results than any other standard approach. A numerical technique, developed
based on forward modeling, is used for the inversion of the observed waveforms for
the components of the moment tensors and the earthquake source-time function
(STF(t)). We used f min of 0.15 Hz and f max of 0.55 Hz to perform the waveform
inversion of the earthquake.
Keywords Moment tensor analysis · Source parameter · Focal plane solution ·

Seismicity
P. Mahto (B) · S. C. Gupta

e-mail: pmahto@eq.iitr.ac.in
S. C. Gupta
e-mail: s.gupta@eq.iitr.ac.in
https://doi.org/10.1007/978-981-99-1459-3_53
670 P. Mahto and S. C. Gupta
1 Introduction
The current knowledge of the crustal structure allows waveforms to be modeled

at low frequencies only. The local earthquake (around 300 km) is constrained by
natural noise. As a result, M = 4.0 waveforms can be integrated over a limited
frequency and distance range. Also, the number of usable stations is small, so few
waveforms seem to provide a relevant focal mechanism [2]. When the epicentral
distances are relatively small, and the frequency is low, then the waveforms are
less dependent on the crustal model. The synthetic seismograms which have been
calculated for the region in the present study provided almost identical waveforms.
The focal mechanism solution through MT inversion requires different filters to find
out the usable frequency range and avoid the disturbances of instruments. It requires
visualization of space and time-variation of the correlation between real and synthetic
seismograms, plotting waveform fit, and beach-ball representation.
A small-sized earthquake of magnitude M L = 4.0 occurred on December 01, 2020,
about 25 km northwest of Haridwar at a focal depth of 40.6 km. The earthquake has
been recorded by a VSAT-based local seismic network operative in the Tehri region
around Tehri dam which was deployed by the Department of Earthquake Engineering,
Indian Institute of Technology Roorkee to monitor the local seismicity of the region
and sponsored by THDC India Ltd., Rishikesh.
In the present study, the computer-code ISOLA is used, with the combination of the
computational speed of FORTRAN code and the user comfort of MATLAB devel-
oped by Sokos and Zahradnik [18], and the local earthquake records are inverted
to estimate the MT solutions. The determination of focal mechanisms solved by
moment tensor waveform inversions shows stable and much more accurate results.
Here, the deviatoric tensor is used because the inverse problem remains linear
and does not create problems during the inversion process. The digital data has
acquired through Broadband sensor 3ESPS (120–50 Hz) and short-period CMG-6 T
(30–100 Hz) coupled with 24-bit DAS (DM24-53) at a sampling interval of 100
samples/sec/component.
The present study lies in the Sub-Himalayan part of the Himalayas. The sub-
Himalayas constitute the foothills part of the Himalayas and lie between the HFT
and MBT. Geographically, Sub-Himalayas constitute the Siwalik range. Rivers have
carried and deposited sandstone and mudstone from the Himalayan mountains for
over 5000 years in this region. Approx. over a million years ago, the Himalayan
Front Fault (or Thrust), which is still active, uplifted the foothills of the Siwalik
Range (of the Himalayas). HFT is the boundary between the Siwalik range and the
North Indian plains. During the uplifting of the Siwalik range by the Himalayan
Frontal Fault, river valleys were created in the Sub-Himalayan zone; these basins
are called duns (doons), and well-known examples include Dehra Dun in India and
Chitwan in Nepal.
December 01, 2020, Haridwar, Earthquake: Fault Plane Solution … 671
2 Geology and Tectonic of the Area
Since the collision of the Indian Plate and Eurasian Plate during the Tertiary Period,
the Himalayan arc has formed along a plate boundary that is 2500 km long and
continues to do so at a rate of 37–44 mm each year [5, 12, 8]. Over the past century,
there have been several significant Himalayan earthquakes, including four great earth-
quakes (M ≥ 8.7). The chronological order of occurrence of them; the 1897 Shillong
earthquake of M 8.7 (Milne et al. 1911), the 1905 Kangra earthquake of M 7.8
(Gutenberg and Richter et al. 1954), the 1934 Bihar-Nepal border earthquake of
M 8.3 (Tandon and Srivastava et al. 1974, 75), and the 1950 Assam earthquake of
M (Tandon and Srivastava et al. 1974, 75). These great earthquakes have ruptured
approximately 300 km long sections of the Himalayan Plate Boundary in the past
100 years.
The Garhwal Himalaya which forms the western part of the Himalayan mountain
ranges is characterized by moderate to large-sized earthquake activity. In the last
three decades, two moderate-sized earthquakes namely the Uttarkashi earthquake of
1991 and the Chamoli earthquake of 1999 were experienced by this region [8] that
falls in the Seismic Zone IV of the seismic zoning map of India (IS: 1893-2002).
Based on the statistical study of the local earthquakes of this region, Khattri et al.,
1999, gave the opinion that there is a high probability to generate a great earthquake
in the future. Focal mechanism solutions are being used extensively to study the
type of displacement and the relative motion between the boundaries of different
plates. In the early days, focal mechanisms have been revealed as a powerful tool
in attaining seismotectonics and the state of stress, but their reliability of results
for Garhwal Himalaya, where there is small-to-moderate earthquakes are occurring
continuously, is inappropriate due to the lack of coverage of seismic stations.
In response to thrust tectonics in the detachment zone along the northward dipping
Indian plate, almost all of the higher magnitude earthquakes (M ≥ 5.0) in the
Himalayas have originated at a depth of 10–20 km near the Main Central Thrust
(MCT) [12]. It is referred to as the area that lies between the locked, shallow fault
segments that ruptured during strong earthquakes and the smoothly sliding deeper
aseismic zone.
The source, path, and instrument response are three effects that are combined to
create the ground motion record of an earthquake at a station. Therefore, to determine
the source mechanism of each fault, an earthquake must be deconvoluted.
There are multiple north-dipping overthrusts in the Himalayan tectonic zone,
which is a plate collision boundary. These overthrusts originate at a decollement
surface that dips 15 degrees to the north, at a depth of between 12 and 20 km [17].
Solutions of fault planes in moderately sized earthquakes indicate upthrust along
shallowly dipping planes originating from the north (e.g., [13]).
The Tethyan Himalayan Sequence (THS), the Higher Himalayan Crystalline
Sequence (HHCS), the Lesser-Himalayan Sequence (LHS), and the Sub-Himalayan
Sequence (SHS) are the major tectonic divisions of the Garhwal Himalaya, and
they are stacked from north to south by the thrust and detachment faults viz. the
Southern Tibetan Detachment (STD), the Main Central Thrust (MCT), the Main
Boundary Thrust (MBT), and the Himalayan Frontal Thrust (HFT) [6, 16, 20, 21].
According to Molnar and Tapponnier et al. [12], the region is subjected to seismic
activity due to the northward motion of the Indian plate against the Tibet block of
the Eurasian plate at a rate of around 0.05–0.06 m/year, which deforms the rocks
to create the higher Himalayas. The majority of the earthquakes that occur in the
Kumaun-Garhwal Himalaya occur nearby or to the north and south of the MCT.
Several seismological studies and efforts have been done in India and elsewhere
to study the complex seismicity patterns in this zone of continental-continental colli-
sions. Fault plane solutions to earthquakes in this region have been carried out by
various authors using data from various sources. Molnar and his coworkers from 1973
to 1975 worked out the solutions by using P-wave first motion using long-period seis-
mometers. Since then, this region has carried out a continuous approach to study fault
orientations. Recently, a study has been done by Prasath and his team [15] to estimate
the upper crustal stress and seismotectonic of the Garhwal Himalayas.
3 Methodology and Qualitative Analysis
3.1 Moment Tensor Inversion
The waveform inversion of the event is a powerful tool for estimating the focal mech-
anism of the event as it provides information on the source properties and the defor-
mation in the source region that generates the seismic waves to examine the solution,
we perform the waveform inversion of December 01, 2020, Haridwar earthquake
using a tool called ISOLA. Using least-squares inversion and iterative deconvolution
of the entire waveform, ISOLA, a MATLAB-based program, determines the focal
MT of an earthquake [9, 18]. The moment tensor and Green’s tensor are used to
relate the event waveforms to the source and medium parameters [1]. Because the
current study is focused on small-scale earthquakes, it analyzes the near-field effects
at local stations. If the time and position of the source are to be determined, the least-
square method is used along with the spatio-temporal grid-search method in MT
inversion. ISOLA code can use both local as well as regional events. The computer-
code ISOLA by Sokos and Zahradnik [18] is used to estimate waveform inversion in
several modes: full MT inversion, DC-constrained MT inversion, and deviatoric MT
inversion. There are deviatoric and volumetric parts to the full moment tensor, and it
is challenging to obtain the volumetric part. Considering [22, Benetatos et al. 2013]
as references, the decomposition in ISOLA for the inversion process namely volu-
metric (ISO), compensated linear vector dipole (CLVD), and double-couple (DC),
such that
ISO% + DC% + CLVD% = 100%.

In the present study, the deviatoric tensor is performed because the inverse problem
remains linear and does not create problems during the inversion process. The devia-
toric tensor can be decomposed into the double-couple (DC) and non-double-couple
(CLVD) components. The relative sizes of double-couple and non-double-couple
(CLVD) parts are (1–2t) and (2t), respectively, where 1, t–1, and –t are the normalized
MT eigenvalues. The term 100 (1–2t) is referred to as the double-couple percentage
DC%.
During waveform inversion, DC% is calculated for all inversion. Green’s function
is calculated by the discrete wavenumber method [4, Bouchon et al. 2003]. The overall
variance reduction provides the match between the observed and synthetic data by
the relation
Σ
Varred = 1 − A/R; where A = (Ri − Si )2 ,
Σ
R = (Ri )2 , with R and S denoting the observed and synthetic data, along with
a summation of overall samples, components, and stations.
The ISOLA code enables histories of complicated ruptures by multiple point-
source subevents, with each event represented by a delta function [24].
4 Data Set and Analysis
In the present study, data from the December 01, 2020, earthquake that has been
recorded in the Sub-Himalaya close to HFT has been analyzed. A local seismo-
logical network of 18 stations around the Tehri dam in the Garhwal Himalaya has
recorded this earthquake (Fig. 1). The Department of Earthquake Engineering at IIT
Roorkee manages and maintains the network, which is sponsored by THDC India
Ltd., Rishikesh. The location of the event marked on the tectonic map of the region
is shown in Fig. 1. The short-period sensor CMG-6 T frequency range 30 s–100 Hz
and the broadband sensor CMG-3ESPCS frequency range (120 s–50 Hz) have been
used to collect the digital data for this earthquake at 20 stations each at a rate of
100 samples/sec/component. The location of the earthquake has been determined
through the HYPO71 computer program used in SEISAN software.
This step estimates the Green function using the trial source position. The best fit
between the observed and the synthetic waveforms is determined using a grid search
by ISOLA software. The grid of trial points can be set up on a plane, in a line, with
different strikes and dips, or it can be fixed at the epicenter position to find the event’s
optimum depth. Figure 4 illustrates the focal depth estimate for an event for which
the highest correlation and DC% values were determined using a vertical grid-search
method. Here, the frequency range (0.15 < f < 0.55 Hz) is used for investigation, and
the frequency above 0.55 Hz cannot be used as above this range, the model misses
the necessary detail of the crustal structure Therefore, the permitted ranges for both
frequency and distance are unfortunately quite normal. A narrow range also reduces
Fig. 1 18-station seismological network operative in the Tehri region along with the location of
the Dec. 01, 2020, earthquake. Tectonics after GSI (2000)
data and makes users more vulnerable to noise, which is a drawback. Instrument
corrections and baseline corrections for the specified event were applied during the
raw data preparation in ISOLA. The spectral amplitude of the earthquake has been
utilized in the present study for the MT inversions.
The final step of analysis is the waveform inversion, which is performed at the
best centroid position of each event. The final output shows the correlation between
the observed and synthetic waveforms as shown in Fig. 2. For each event, the optimal
depth is found, and by fixing this depth, the optimal epicenter position is calculated.
The best location of the epicenter is one where the correlation coefficient has the
maximum value in the horizontal plane.
Based on the above-mentioned analysis procedure, out of 18 stations, only 4
stations are used for inversion. The waveform of the other stations is not inverted
due to the instrumental disturbances and poor azimuthal coverage.
5 Results
In the present study, data from the December 01, 2020, earthquake has been analyzed
using a moment tensor inversion solution. Based on the analysis, we have obtained
earthquake source parameters viz., hypocentral location consisting of centroid loca-
tion, and centroid depth. It has been identified that earthquakes with M L > 3.0 hold
Fig. 2 Waveform fit for the moment tensor solution for Dec. 01, 2020, earthquake. The plot of
comparisons between the original (black waveform) and synthetic (red waveform) for four stations
used for inversion. Inversion band 0.15–0.55 Hz is used for analysis
a good signal-to-noise ratio (SNR) to perform waveform inversion. The earthquake

here is of magnitudes M L = 4.0. The value of the signal-to-noise ratio is found to
be 3, which gives a good estimate over the frequency range of 0–15 to 0.55 Hz.
The variation between the calculated Mw and the local magnitude (M L ) is 0.01. The
centroid depth estimated has a difference of 4.4 km to the hypocenter of the earth-
quake. Figure 3 depicts the inverted earthquake event’s seismic moment (Mo) as
1.147e + 15 N-m. The moment magnitude (M w ) obtained through the MT inversion
of the earthquake is 4.01.
Fig. 3 Moment tensor solution of the event 01.12.2020 estimated using single source inversion in
ISOLA software
The fault plane solution of this earthquake has been obtained from the MT inver-
sions using only 4 seismological stations of the network as shown in Fig. 2. The fault
parameters of this earthquake along with beach-ball presentation are given below.
From the obtained fault plane solutions, it has been found that the solution is strike-
slip fault with thrust components. The epicentral location of the earthquake is located
around the HFT zone as shown in Figs. 3, 4, and 5.
For an event, when the correlation and the DC% are showing the highest values
in the vertical grid search, the focal depth estimate is considered to be found. Here,
the values of the correlation coefficient of moment tensors with trial focal depths are
shown in Fig. 4.
Based on the above analysis, finally, following results have been obtained:
Origin time: 04:11:50.83
Centroid location: Lat. 30.10 N, Long. 78.020 E
Centroid depth (km): 45 km
Seismic moment: 1.15e+15 N-m
Moment magnitude (Mw ): 4.01
DC %: 49.1
CLVD%:50.9
Plane 1: Strike: 71, Dip: 80, Rake: 25
Plane 2: Strike: 336, Dip: 66, Rake: 169
P-axis: Azimuth: 201, Plunge: 10

Fig. 4 Correlation of original and synthetic waveforms performed at a trial source depth. DC%
represent by colors coding for 01/12/2020 earthquake and the focal mechanism during the inversion
process for the optimum time step for this event (+0.36 s) and various depths are shown
Fig. 5 Map showing beach-ball representation of FPS solution of the event marked on the tectonics
map of the region
T-axis: Azimuth: 296, Plunge: 24
Type of solution: Strike-slip with thrust faulting
The MT solution being with the previously available solutions studied in the
Chamoli region with M > 3.0 earthquake showed the strike-slip MT solutions are
well supported by the occurrence of a couple of strike-slip fault plane solutions [11].
But being very less availability of noise-free data along HFT and the Ganga foredeep
area, the solution provided in the present study seems important and useful.
6 Discussion
Garhwal Himalaya is a region with active earthquakes and is well-known for its inter-
plate seismic zone of thrust regime. A detailed and accurate analysis of seismic
sources is always performed using moment tensor inversion. The earthquake has
been generated due to both double-couple and non-double-couple mechanisms. The
high value of CLVD shows that there may be a presence of complex tectonic features
which could be responsible for this earthquake, and it could be a multiple fractures
zone showing numerous cracks.
The inversion is showing a good correlation between centroid location and
hypocentral location. And, the value of variance reduction is near 1 confirming that
the trial source modeling at this depth is giving the most stable result causing this
earthquake. The value of CLVD is very much equal to double-couple, it may show
the presence of a geothermal province (may be hot springs), or may be the result of
a crack opening under stress and strain and causing this earthquake.
Further, the MT solution provided in this study being consistent will be impor-
tant as very less earthquake activity with magnitude three and above occur in the
region particularly along HFT which is considered the boundary between the Siwalik
foothills and the Ganga foredeep plains. The centroid depth will also give clue the
depth of Moho in this zone.
Acknowledgements Authors acknowledge with thanks to the THDC India Ltd., Rishikesh for
sponsoring the study. Authors are also thankful to the Head, Department of Earthquake Engineering,
Indian Institute of Technology Roorkee and Prof. M. L. Sharma, P.I. of study for providing the data
for carrying out the study and allowing to use departmental facilities to complete the study. Mrs.
Pooja Mahto also expresses gratitude toward Dr. Arup Sen, Dr. Sanjay Kr. Jain and other team
members of the Seismological Observatory, Department of Earthquake Engineering, IIT Roorkee
for providing the help and suggestions at various stages of this study.
References
1. Aki, K., Richards, P.G.: University Science Books. Sausalito, California (2002)
2. Antolik, M., Dreger, D., & Romanowicz, B. (1999). Rupture processes of large deep-focus
earthquakes from inversion of moment rate functions. Journal of Geophysical Research: Solid
Earth, 104(B1), 863-894.
3. Benioff, H., Gutenberg, B., & Richter, C. F. (1954). Progress Report, Seismological Laboratory,
California Institute of Technology, 1953. Eos, Transactions American Geophysical Union,
35(6), 979-987.
4. Coutant, O.: Program of numerical simulation AXITRA. LGIT (in French), Universite Joseph
Fourier, Grenoble, Res. Rep (1989)
5. Gansser, A.: Geology of the Himalayas (1964)
6. Jain, A.K.: Continental subduction in the NW-Himalaya and Trans-Himalaya. Italian J. Geosci.
136(1), 89–102 (2017)
7. Kanaujia, J., Kumar, A., Gupta, S.C.: Three-dimensional velocity structure around Tehri region
of the Garhwal Lesser Himalaya: constraints on the geometry of the underthrusting Indian plate.
Geophys. J. Int. 205(2), 900–914 (2016)
8. Kayal, J.R., Ram, S., Singh, O.P., Chakraborty, P.K., Karunakar, G.: Aftershocks of the March
1999 Chamoli earthquake and seismotectonic structure of the Garhwal Himalaya. Bull. Seismol.
Soc. Am. 93(1), 109–117 (2003)
9. Kikuchi, M., Kanamori, H.: Inversion of complex body waves—III. Bull. Seismol. Soc. Am.
81(6), 2335–2350 (1991)
10. Kumar, R., Gupta, S.C., Kumar, A.: Effect of azimuthal coverage of an earthquake on moment
tensor solutions estimated by waveform inversion. Arab. J. Geosci. 8(8), 5713–5726 (2015)
11. Kumar, N., Paul, A., Mahajan, A.K., Yadav, D.K., Bora, C.: The Mw 5.0 Kharsali, Garhwal
Himalayan earthquake of 23 July 2007: source characterization and tectonic implications. Curr.
Sci. 102(12), 1674–1682 (2012)
12. Molnar, P.: A review of seismicity and the rates of active underthrusting and deformation at
the Himalaya. J. Himalayan Geol. 1, 131–154 (1990)
13. Ni, J., Barazangi, M.: Seismotectonics of the Himalayan collision zone: geometry of the under-
thrusting Indian plate beneath the Himalaya. J. Geophys. Res.: Solid Earth 89(B2), 1147–1163
(1984)
14. Paul, A., Prasath, A.R., Singh, R.: Slip heterogeneities evaluated for earthquakes M> 4.0 using
waveform modelling in the Garhwal region of Central Seismic Gap in Northwest Himalaya,
India. Him. Geol. 36, 153–160 (2015)
15. Prasath, R.A., Paul, A., Singh, S.: Upper crustal stress and seismotectonic of the Garhwal
Himalaya using small-to-moderate earthquakes: implications to the local structures and free
fluids. J. Asian Earth Sci. 135, 198–211 (2017)
16. Richards, A., Argles, T., Harris, N., Parrish, R., Ahmad, T., Darbyshire, F., Draganits, E.:
Himalayan architecture constrained by isotopic tracers from clastic sediments. Earth Planet.
Sci. Lett. 236(3–4), 773–796 (2005)
17. Seeber, L., Armbruster, J.G.: Great detachment earthquakes along the Himalayan arc and
long-term forecasting. Earthquake Pred.: An Inter. Rev. 4, 259–277 (1981)
18. Sokos, E.N., Zahradnik, J.: ISOLA a Fortran code and a Matlab GUI to perform multiple-point
source inversion of seismic data. Comput. Geosci. 34(8), 967–977 (2008)
19. Sokos, E., Zahradník, J.: Evaluating centroid-moment-tensor uncertainty in the new version of
ISOLA software. Seismol. Res. Lett. 84(4), 656–665 (2013)
20. Srivastava, P., Mitra, G.: Thrust geometries and deep structure of the outer and lesser Himalaya,
Kumaon and Garhwal (India): implications for evolution of the Himalayan fold-and-thrust belt.
Tectonics 13(1), 89–109 (1994)
21. Valdiya, K.S.: The two intracrustal boundary thrusts of the Himalaya. Tectonophysics 66(4),
323–348 (1980)
22. Vavryčuk, V.: Inversion for parameters of tensile earthquakes. J. Geophys. Res.: Solid Earth.
106(B8), 16339–16355 (2001)
23. Vavryčuk, V.: Tensile earthquakes: theory, modeling, and inversion. J. Geophys. Res.: Solid
Earth 116(B12) (2011)
24. Zahradnik, J., Jansky, J., Plicka, V.: Detailed waveform inversion for moment tensors of M∼4
events: examples from the Corinth Gulf, Greece. Bull. Seismol. Soc. Am. 98(6), 2756–2771
(2008)
Local Seismicity Around Tehri Dam,
Garhwal Himalaya
M. L. Sharma, S. C. Gupta, J. P. Narayan, J. Das, A. Sen, S. K. Jain,

A. K. Jindal, Subhash Patel, Prajawal Tandekar, Avichal Rastogi,
Rajeev Vishnoi, Atul Jain, Virendra Singh, and S. K. Saxena
Abstract Tehri dam with a height of 260.5 m is the highest earth and rock fill dam
in India. It is situated on the confluence of the Bhagirathi and the Bhilangana rivers
in the Garhwal Himalaya and located in the highly strained region of the north-
western Himalaya. The region around Tehri dam lies in zone IV and V as per the
seismic zoning map of India where the 1991 Uttarkashi earthquake Mb ~ 6.6 and
the 1999 Chamoli earthquake Mb ~ 6.4 have occurred. The local seismicity in the
environs of Tehri dam is being monitored for last more than two and half decades.
For this purpose, a local seismological network was deployed by the Department
of Earthquake Engineering in September 1993 under the scheme of Department
of Science and Technology (DST). The network has been upgraded time to time.
Presently, 18 stations of state-of-the-art seismological network is being operated
around Tehri dam reservoir area. The spatial variation of local seismicity follows
the trend of surface trace of MCT in the Garhwal Lesser Himalaya. The focal depth
distribution of seismic events along and across the strike direction of the regional
tectonic features reveal the confinement of activity within 10 to 15 km. In the present
paper, attributes of local seismicity for more than two and half decades have been
presented for Tehri dam. In addition, the changes in seismicity due to dam reservoir
as a part of reservoir-induced seismicity (RIS) has been studied and no correlation of
seismicity activity with the dam reservoir filling has been observed even after about
seventeen years filling/drawdown of dam reservoir.
Keywords Microearthquake · Seismicity · MBT · MCT
M. L. Sharma · S. C. Gupta (B) · J. P. Narayan · J. Das · A. Sen · S. K. Jain · A. K. Jindal ·

S. Patel · P. Tandekar · A. Rastogi
e-mail: s.gupta@eq.iitr.ac.in
M. L. Sharma
R. Vishnoi · A. Jain · V. Singh · S. K. Saxena
THDC India Ltd, Rishikesh, Uttarakhand, India
https://doi.org/10.1007/978-981-99-1459-3_54
684 M. L. Sharma et al.
1 Introduction
Microearthquake networks are powerful tool in studying the nature and state of
tectonic processes for mapping active faults for hazard evaluation. The moderniza-
tion of seismological instrumentation has led to the development of seismological
networks for monitoring of seismicity around important engineering projects such as
HE projects. Such networks may contain different types of instrumentation depending
on the objectives of the study. While broadband instruments are used to record the
teleseismic and/or regional events, the short period seismometers are preferred to
record the local events due to their high frequency contents. Such networks may also
contain some of the strong motion recorders which are generally deployed in the
dam bodies to study the response of the structure during the earthquake [1]. These
instruments may become the part of the network when the data is collected at a central
recording station or may be deployed as an individual unit. The basic processing of
the data is carried out to locate the events and to estimate the size and the physical
process in terms of the focal mechanism. The advance processing may contain the
estimation of the temporal and spatial variation of seismicity, their source parameters
and the physical modeling of the medium and the source in the region [2]. Further,
specific studies may also be carried out which relates to the medium characterization
in terms of its attenuation properties [3], 3D velocity structures [4], seismic hazard
parameter estimations and the risk assessment for the region. The results of these
studies are used to interpret them in terms of hazard for the region may it be for earth-
quake resistant designs or for studies like reservoir-induced seismicity and other by
products being used by disaster managers.
The Garhwal Himalaya that forms the part of north-western Himalayan mountain
ranges is characterized by moderate to large sized earthquake activity. In the last
two decades, two moderate sized earthquakes namely the Uttarkashi earthquake of
1991 and the Chamoli earthquake of 1999 were experienced by this region that
falls in the Seismic Zone IV of seismic zoning map of India (IS: 1893-2002). To
monitor the contemporary local seismicity covering parts of the Garhwal Himalaya,
the earthquake data has been collected through the continuous operation of a local
seismological network that commenced in the region during 1993. A general pattern
of ambient tectonic strains prevailing in the region is attributed to the collision of
India plate and Eurasia tectonic plate resulting in high level of inter-plate seismicity.
This paper contains the various attributes of contemporary local seismicity around
Tehri dam in the Garhwal Himalaya. Further, the paper also contains its seismic
hazard parameter analysis along with its RIS behavior.
2 Seismological Network
The region around Tehri dam falls in the Lesser Himalaya of the Garhwal Himalaya.
It is composed of a mountain belt having an average elevation of about 2000 m and is
Local Seismicity Around Tehri Dam, Garhwal Himalaya 685
made up of fossiliferous Riphean sediments overridden by several thrust sheets which

have traveled from north to south in response to the geodynamic processes that gave
birth to the Himalaya. To the south, the Lesser Himalaya is separated from the Sub-
Himalaya by the Main Boundary Thrust (MBT), while in the north it is separated from
the Higher Himalaya by the Main Central Thrust (MCT). The Main Central Thrust
(MCT) is demarcated as a structural feature of great geotectonic significance and is
marked as a boundary separating the Lesser Himalaya from the Higher Himalaya
[5]. Besides the two prominent thrust boundaries trending northwest to southeast and
running parallel to the Himalayan arc, there are structural features trending transverse
to the Sub-Himalaya in the form of subsurface ridges and depressions, buried under
the alluvium cover of the Indo-Gangetic Plains [6, 7]. Therefore, the location of
Tehri dam that falls between two major tectonic features, the MBT and the MCT.
Besides that, the region encompasses several local tectonic features, e.g., Sriguri
thrust, Dunda thrust, Aglar thrust, Uttarkashi thrust, Srinagar thrust in the Garhwal
Lesser Himalaya and transverse structural features in the Sub-Himalaya SW of dam
site [8].
In view of presence of several tectonic features, structural features as well as the
area around Tehri dam falls in seismic zone IV and V, as per the seismic zoning
map of India [9], several local seismological networks were operated in the region
time to time by several institutions to monitor the local seismicity of the region (e.g.,
Gaur et al. [10], Kumar et al. [11], Kumar [12], Khattri et al. [13], Kumar et al.
[14]). The Department of Earthquake Engineering, IIT Roorkee has deployed its
first seismological network in 1970 with three observatories. Later, a six-station local
seismological network has been operated from 1987 onwards as part of DST project
that later funded by the THDCIL since 1995. This network has been expanded and
upgraded time to time and being operated continuously since last more than 27 years.
Presently, a 18-station VSAT-based local seismological network is operative in the
region around Tehri dam reservoir. Out of 18 stations, 9 stations are operated in
radio telemetry mode while 9 stations are operated in VSAT mode and real time
data is being received continuously to two central recording stations (CRSs) [15].
Figure 1 shows the locations of 18 stations of the seismological network along with
the location of Tehri dam and reservoir area. The network has been deployed with
the objective to collect small magnitude earthquake activity on long-term basis to
study the contemporary local seismicity and to delineate active seismic source zones
in the environs of Tehri dam reservoir and also to provide data of local seismicity in
the region around Tehri dam before, during and after impounding of dam reservoir to
allow to study the changes in the local seismicity if any, during and after impounding
of Tehri dam reservoir.
3 Data Set
The network, deployed by Department of Earthquake Engineering, IIT Roorkee and

funded by the THDC India Limited, Rishikesh has acquired one of the most precious
Fig. 1 Map showing 18-station seismological network around Tehri dam along with major tectonic
features in the region. Tectonics after GSI [8]
data sets on local seismicity from the Garhwal Himalaya. Such data set is one of
the assets for seismotectonic studies which is indispensable for seismic hazard and
risk assessments. The present study is based on the local earthquake data collected
continuously for last more than two and half decades from 1995 to 2021. Local
earthquakes with magnitude two and above (ML ≥ 2.0) recorded within 200 km
from each station, i.e., (S-P) time less than 25.0 s., have been considered for analysis.
Only those events having standard errors less than 0.6 s in origin time (erms), less
than 6.0 km in epicentral distance (erh) and less than 6.0 km focal depth (erz) have
been considered estimation of hypocenter parameters, preparation of seismicity map
and drawing fault plane solution, etc.
The hypocenter parameters of events have been computed employing the
HYPO71PC and HYPO71 program given in SEISAN computer program [16, 17]
and using the velocity model given by Chander et al. [18], Kumar et al. [19], and
Kanaujia et al. [20] for the region (Table 1).
Table 1 Velocity model

P-wave velocity (km/sec) Depth to the top of the layer (km)
adopted in the study
3.0 0.0
5.2 1.0
6.0 16.0
7.91 46.0
4 Data Analysis
4.1 Spatial Variation of Local Seismicity
Based on the above analyses procedures, hypocenters of 2224 events having magni-
tude two and above and recorded within 200 km of Tehri dam (i.e., ML ≥ 2.0 and S-P
≤ 25 s) have been considered suitable for location and other interpretation. Based on
the epicentral locations of these events, a seismicity map has been prepared and shown
in Fig. 2. The distribution of seismic activity has brought out that the most of the
contemporary local seismicity is non-uniformly distributed and occurs in the Lesser
Himalaya and the Higher Himalaya (EQ: 2021-04). An endeavor has been made to
associate the present seismicity with the existing tectonic features. Such correlation
is important from seismic hazard point of view where the tectonic features have to be
redefined as seismogenic features based on the contemporary seismicity [20]. The
seismic hazard studies have been carried out in Himalayas using such data [21]. The
seismicity associated with the existing features may include some of the seismicity
part which is due to the features in their vicinity or may be considered as their interac-
tions and/or transfer of stress regime during the dynamism of stress release process,
i.e., earthquake occurrence in this area [22–25].
Fig. 2 Map showing epicenters of events around Tehri dam located during 1995 to 2021. Two
depth cross-sections marked AA’ and BB’ are also shown. Tectonics after Seismotectonics Atlas
[8]
Based on the distribution (pl. see Fig. 2) about 85% of seismic activity is located
along the segment of MCT and follow the trend of surface trace of the MCT. The
distribution of activity defines about 400 km long and about 60 km wide NW–
SE elongated local seismicity zone extending from the Himachal Himalaya in NW
direction to Dharchula in the Kumaon Himalaya in SE direction, while 15% of
remaining activity scattered between the Srinagar thrust and MBT in the Lesser
Himalaya and between the MBT and HFT in the Sub-Himalaya as well as in the
Indo-Gangetic plains.
4.2 Depth Distribution and Depth Sections of Seismicity
The focal depth distribution of seismic activity shows that about 80% of activity
occurred at shallow depths ranging between 5 and 20 km. There is a significant
reduction in the occurrence of events having focal depths above 20 km (Fig. 3).
Therefore, based on this sample of data set, it is attributed that there are maximum
chances of occurrence of earthquake at shallow depth up to 20 km [26]. To study
the depth-wise variation of seismic activity along traverses, two depth sections AA’
and BB’ are drawn. These traverses have been drawn along and across the major
tectonic features in the region (Fig. 2). Hypocenters of events, confined to a narrow
zone approximately ± 40 km wide on either side of these traverses, are projected
onto a vertical plane passing through these traverses.
Depth section AA’ passes in the close proximity of Tehri dam and runs across the
regional strike directions of the MFT, MBT, MCT and Srinagar thrust (Figs. 2 and
5). The seismic activity along this section primarily confined to the three tectonic
domains namely the Sub-Himalaya, the Lesser Himalaya and the Higher Himalaya.
Fig. 3 Histograms showing

depth distribution of seismic
activity in the Garhwal
Himalaya
Fig. 4 Depth cross section of seismic activity along traverse AA’
Baring few events, most of the activity along this section confined to depth up to
50 km. The focal depth of events in the Sub-Himalaya between the HFT and MBT
occur at around 12–50 km, while activity in the vicinity of dam site occurs at depth
between 10 and 20 km. This activity seems to be related mainly to Srinagar thrust
that dips in south-west direction. Further, NE of the section, the most of activity in
the vicinity of MCT occurs at depths between near surface to 20 km, while further
to the northeast of the MCT in the Higher Himalaya, events occur either at depth
around 10–20 km or at depths above 45–55 km (Fig. 4). The activity at shallow
depths around 10–15 km in the Lesser Himalaya and the Higher Himalaya seems to
be associated with the upper part of the detachment surface of the under-thrusting
Indian plate, whereas the activity at a level between 30 and 45 km seems to signify
the depth of the brittle-ductile transition zone. The activity at this depth seems to
define the geometry of the under-thrusting India plate.
The depth section BB’ depicts the focal distribution of seismic activity along the
regional strike direction of the major tectonic features in the region (Fig. 5). In this
section, events are widely distributed all along section and focal depths of events
vary from surface to 50 km forming the synclinal structure of activity close to the
surface at both the end. Focal depths of most of events fall between 10 and 20 km.
This activity occurs in the Lesser Himalaya either in the vicinity of the MCT or to
the south of the MCT. Most of the activity that falls in the north of the MCT in
the Higher Himalaya occurs at shallow depths between 10 and 15 km. Therefore, it
is observed that the activity in the Garhwal Lesser Himalaya occurs at an average
depth of about 15 km, whereas in the Higher Himalaya, the activity occurs at an
average depth of about 10 to 12 km. The events at very shallow depths (<10 km)
on both the end of section represent the seismic activity of the part of the boundary
between Himachal Himalaya and the Garhwal Himalaya and the Garhwal-Kumaon
Lesser Himalaya and also the activity of two earthquake swarms that occurred in the
vicinity of Uttarkashi during 1995 and 1996. This activity seems to be related to the
local tectonic features as, the Uttarkashi thrust and Dunda thrust.
Fig. 5 Depth cross section of seismic activity along traverse BB’
4.3 Focal Mechanism
Focal mechanism describes the geometry and mechanism of fault during an earth-
quake. It is determined by drawing fault plane solution (FPS) from the polarities of
first motion of seismic wave arrived at network stations. It is defined by three angles,
the strike angle, φ, which is the angle of azimuth with respect to north, the dip angle,
δ, showing steepness of the fault and the rake or slip angle, λ, which indicates the
direction of motion within the fault plane relative to the horizontal or direction in
which hanging wall block moves during rupture. FPS is generally drawn under the
assumption that the events located in the same area/cluster are occurred along a
similar fault/thrust therefore, consists of a similar focal mechanism. However, some
studies have reported that events occurring closely in space and time do not share
common focal mechanism (e.g., [27]). On account of this, considerable care and
judgment have been taken in drawing FPS.
Here, 26 events (2021) have been considered for drawing FPS in the region. Map
showing locations of these events and their respective FPS in the form of ‘beach ball’
diagram are depicted in Fig. 6. From the figure, it is evident that events depicted all
types of focal mechanisms viz., normal, reverse and strike slip along with the different
combinations such as normal and reverse faulting with strike-slip component.
5 Seismic Hazard Parameter—b-Value
Evaluation of seismicity using available data on past earthquakes is most commonly

based on the Gutenberg-Richter’s [28] relationship. According to which the annual
occurrence rate N with magnitude equal to and greater than M in a source zone
can be described by log N = a − bM, where N is the cumulative frequency of
occurrence of earthquakes of magnitude M and greater, and a and b are constants for
a given source zone. The seismicity parameter ‘b’ gives the idea about the relative
seismicity with respect to the size, and the parameter ‘a’ gives the total seismicity
occurring in a region greater than magnitude 0. The higher the value of b, the higher
is the probability of occurring low magnitude earthquakes in the region, while the
low b-value indicates the relatively higher probability of larger size magnitudes in
the region. The G-R relationship has been worked out for the Garhwal region using
P axis in degrees: 51.73 68.20
T axis in degrees: 249.28 20.88
P axis in degrees: 20.06 73.29 P axis in degrees: 165.00 40.00 P axis in degrees: 353.96 10.04 P axis in degrees: 353.18 14.96
North T axis in degrees: 306.21 4.77 T axis in degrees: 165.00 50.00 T axis in degrees: 272.59 40.27 T axis in degrees: 196.07 73.83
North North North North
B North
U´
P
B
T T T U
U
U´ T
P U
U T
U T
U
P B B
P
B U´
U´
P
U´ U´
stereographic lower B P
fault plane in degrees: 170.20 24.70 75.50
auxiliary plane in degrees: 334.31 66.14 96.57
stereographic lower stereographic lower stereographic lower stereographic lower
fault plane in degrees: 230.60 51.80 69.50 fault plane in degrees: 75.00 85.00 90.00 fault plane in degrees: 229.50 70.50 38.00 fault plane in degrees: 254.50 30.50 78.10 P axis in degrees: 309.98 12.85
stereographic lower auxiliary plane in degrees: 19.44 42.60 113.99 T axis in degrees: 162.86 74.81
auxiliary plane in degrees: 255.00 5.00 90.00 auxiliary plane in degrees: 124.88 54.52 155.80 auxiliaryPplane
axis in degrees: 282.17
in degrees: 88.24 46.96
60.22 96.93
fault plane in degrees: 230.00 50.00 90.00 P axis in degrees: 231.80 0.64
auxiliary plane in degrees: 50.00 40.00 90.00 T axis in degrees: 278.17 42.97
North
North
T axis in degrees: 56.36 78.81 North
North B
U´
U T
P
T
T U
U´ U U´
P
U´ B
T P
B
P
U B
stereographic lower
stereographic lower fault plane in degrees: 209.30 32.90 75.20
fault plane in degrees: 190.10 88.00 88.00 auxiliary plane in degrees: 46.77 58.32 99.38 stereographic lower
auxiliary plane in degrees: 325.08 2.83 134.98 fault plane in degrees: 97.50 82.00 9.00
P axis in degrees: 56.29
auxiliary plane 68.89 6.24 81.09 171.90
in degrees:
stereographic lower T axis in degrees: 284.21 14.50
North
U
P
U´ T
stereographic lower
P axis in degrees: 199.66 71.76 auxiliary plane in degrees: 1.75 61.15 107.18
North
North
T U´
P T
U
P
B
B
U´
stereographic lower
stereographic lower
P auxiliary
axis in degrees: 220.83
plane 32.26107.06 24.74
in degrees: 104.54
North
P axis in degrees: 204.30 33.00 U´

North
P
U´
B U
B
T
T
U P axis in degrees: 341.26 64.19

stereographic lower
P fault plane North 92.20 84.70
in degrees: 41.70

T axis
stereographic lowerin degrees: 150.55 19.12 T
auxiliary plane in North
degrees: 294.30 12.00 90.00
U´
B
U´
P
P
U
U
B stereographic lower
T
stereographic lower
North
T
North
U´
U´
P B
B T
U
P
stereographic lower
P stereographic
axis lower
in degrees: 1.83 77.04
T fault
axis plane in degrees:in293.47
degrees:4.85
46.90 78.60 71.60
P axis in degrees: 279.81 3.54 P axis in degrees: 13.34 29.47 P axis in degrees: 18.83 42.56 North
T axis in degrees: 186.17 45.79 P axis in degrees: 316.02 8.16 T axis in degrees: 215.87 58.54 T axis in degrees: 249.55 34.59
T axis in degrees: 208.55 64.48 P axis in degrees: 5.70 57.00
T axis in degrees: 172.85 10.00 T axis in degrees: 5.70 33.00
North North North
North
North B
North
U T
T
U U B U
T T
B T P
U´ T
T U´ U´
B U
P U´ U´
U´ P
B
B P
P
B P
U
P U stereographic lower
U´ auxiliary plane in degrees: 10.62 41.51 108.26
stereographic lower stereographic lower stereographic lower
fault plane in degrees: 153.70 56.60 33.70 stereographic lower fault plane in degrees: 256.10 18.00 55.70 fault plane in degrees: 217.50 28.40 9.30
auxiliary plane in degrees: 43.54 62.41 141.60 fault plane in degrees: 200.50 42.40 52.90 stereographic lower stereographic lower auxiliary plane in degrees: 111.75 75.21 100.38 auxiliary plane in degrees: 315.70 85.59 118.08
auxiliary plane in degrees: 66.18 57.47 118.85 fault plane in degrees: 109.40 40.80 51.50 fault plane in degrees: 275.70 78.00 90.00
auxiliary plane in degrees: 242.98 59.24 118.25 auxiliary plane in degrees: 95.70 12.00 90.00
Fig. 6 Locations of events of used for drawing Fault Plane Solutions around Tehri dam
the data set of about 7000 events with magnitude zero and above (i.e., 0 ≤ M L ≤
5.2) recorded during 1995–2021 is shown in Fig. 7. Based on this data set, a- and
b-values have been estimated for the Garhwal Himalaya region as 4.86 and 0.76,
respectively, with magnitude of completeness as 2.1.
The variation of a- and b-values from 1995 to 2021 is shown in Fig. 8a, b. The
variation shows the accumulation of stresses and its release in cyclic form. However,
Fig. 7 G-R relationship for

the Garhwal Himalaya region
based on 1995–2021 data
Fig. 8 Variation of a- and b-values for the Garhwal Himalaya region based on 1995–2021 data
the low magnitudes occurring in this region reveal the complexity of this region
which is not allowing accumulation of higher stresses which may be considered as
a sign of release of energy in low magnitude and lesser probability of occurrence of
great magnitude earthquakes in the region [29]. Such interpretations encourage the
scientists to look for other reasons like slow or quite earthquake occurrence in this
region.
6 Reservoir-Induced Seismicity (RIS)
When a dam is built and the reservoir filled with water, the amount of pressure
exerted on the area changes drastically. When the water level of a reservoir is raised,
pressure on the underlying ground increases, when the water level is lowered, the
pressure decreases. This fluctuation can stress the delicate balance between tectonic
plates beneath the surface, possibly causing them to shift. Another factor is the water
itself. When the water pressure increases, more of it is forced into the ground, filling
cracks and crevices. This water pressure can expand those cracks and even create
new, tiny ones in the rock, causing greater instability below ground. As the water
sinks deeper, it can act as sort of a lubricant for rock plates that are being held in place
by friction alone. The lubrication can cause those plates to slip. The phenomenon
of dam-triggered earthquakes is known as reservoir-induced seismicity (RIS). RIS
usually starts in the period between the beginning of the impoundment of water
and the time when full impoundment is first achieved. The activity subsequently
decreases to the normal level of seismicity for a while. The water level of a reservoir
changes seasonally after full impoundment. Ibenbrahim et al. [30], suggested that
RIS correlates significantly with the seasonal variation in water level.
Gupta et al. [31] suggested that prevailing fault environment plays an important
role in causing RIS. In an environment of normal faulting, the maximum effective
stress, σ1, is vertical and remains unaffected by the filling of an infinite reservoir
of height, h. However, the horizontal effective stress is reduced by 0.57ρgh, thus
driving the Mohr circle toward failure. In a strike-slip environment, the stresses σ1
and σ3 are both in a horizontal plane and are reduced by 0.57ρgh on filling the
reservoir. However, in a thrust environment, the maximum effective stress, σ1, is in
a horizontal plane. Filling of an infinite reservoir does not affect the vertical stress,
σ3, whereas σ1 is reduced by 0.57ρgh, thus, diminishing the diameter of Mohr circle
and consequently driving it away from the failure envelop and thus bringing stability
[32].
The daily frequency of occurrence of earthquakes within 30 km of Tehri dam
along with dam reservoir level has been monitored from Oct. 2005 to Dec. 2019
in Fig. 9. Barring few peaks of maximum two events in a day at initial level of
reservoir filling, the level of earthquake activity remains stationary. In view of the
above, it can be said that during last fifteen years after dam reservoir filling no
noticeable change in the seismicity level within 30 km of Tehri dam with respect
to dam reservoir cycles has been observed. Therefore, non-occurrence of reservoir-
induced earthquakes after a number of cycles of impoundment and depletion of dam
reservoir could be comprehended primarily due to thrust-fault environment, generally
prevailing in Garhwal Himalaya, which is not conducive for RIS.
Fig. 9 Daily frequency of local earthquake within 30 km of Tehri dam along with level of Tehri
dam reservoir since Oct. 2005 to Dec. 2019
7 Conclusions
The seismological network deployed by IIT Roorkee and funded by THDCIL is one
of the state-of-the-art network of seismological studies in the Garhwal Himalaya.
The network includes reception of real-time data at IIT Roorkee for speedy analysis
of data. The continuous acquisition of local earthquake data for more than two and
half decades in high seismically active region like Himalayas has made this data set
as one of the best data sets to understand the complex geotectonic process of the
region as well as its effect on reservoir-induced seismicity. Following are the broad
conclusions drawn from the study.
• The distribution of local seismic activity observed during last 27 years from 1995
to 2021 is confined in a segment of 400 km × 50 km extending 180 km NW to
200 km SE of Tehri dam.
• The majority of events follows the trend of surface trace of the MCT.
• The majority of events occurred at focal depths ranging between 10 and 18 km.
This seems to be the depth of upper part of the Indian Plate below the Lesser
Himalaya and the Higher Himalaya. However, the activity between 30 and 50 km
signify the depth of the brittle-ductile transition zone.
• Based on the entire data set, the b-value of the region around Tehri has been
obtained as 0.76.
• During last fifteen years from the year 2005 after the dam reservoir filling no
noticeable change in the seismicity level with respect to dam reservoir cycles has
been observed. Hence, no trigger seismicity is associated with Tehri dam reservoir.
• The style of faulting of seismic activity associated with the MCT demonstrated
both reverse and normal faulting, while seismic activity south of dam site
associated with Srinagar thrust demonstrated mainly normal faulting.
Acknowledgements We acknowledge with thanks to the THDCIL, Rishikesh for funding the study
in continuously since July 1995. We are also thankful to the Apex Advisory Committee of THDCIL
for their guidance and continuous encouragement specially by Mr. Rajeev Vishnoi and Mr. Gopala
Krishana. We also express our gratitude to the OMS and Tehri Dam-Stage1 officials for providing
help and cooperation for the operation and maintenance of CRS, NTT and seismological field
stations.
References
1. Gupta, S.C., Sachdeva, R., Sharma, M.L., Vishnoi, R.K., Mani, M.: Strong motion network in
Tehri dam and Koeshwar dam in Uttarakhand Himalaya (2021). In: 17th World Conference on
Earthquake Engineering , Sendai, Japan, Paper No. C002514, Sept. 29–Oct. 2 (2021)
2. Sen, A., Gupta, S.C., Sharma, M.L., Kumar, A.: Seismotectonic prospective adjacent to main
frontal thrust and Ganga Foredeep in the Garhwal Himalaya, India. In: 17th World Conference
on Earthquake Engineering , Sendai, Japan, Paper No. C002865, Sept. 29–Oct. 2 (2021)
3. Jain, S.K., Gupta, S.C., Kumar, A.: Attenuation of coda waves in the Garhwal Lesser Himalaya,
India. J. Seismol. 355–369(19), 2015 (2015)
4. Joytima, K., Kumar, A., Gupta, S.C.: 3-D velocity structure around the Tehri region of the
Garhwal Lesser Himalaya: constraints on geometry of the underthrusting Indian plate. Geophys.
J. Int. 900–914, 205(2) (2016)
5. Geological Survey of India Report: Geology of the Himalaya appraisal of status and definition
of problems (1976)
6. Karunakaran, C., Rao, A.R.: Status of exploration for hydrocarbons. In: The Himalayan Region-
contributions to Stratigraphy and Structure Himalayan Geology Seminar, New Delhi, Section
III (1976)
7. Raiverman, V., Srivastava, A.K., Prasad, D.N.: On the foothill thrust of northwestern Himalaya.
J. Himal. Geol. 4, 237–256 (1993)
8. Geological Survey of India: Seismological Atlas of India and its environs (2000)
9. IS 1893–2002 (Part 1): Indian Standard Criteria for Earthquake Resistant Design of Structures,
Part 1—General Provisions and Buildings. Bureau of Indian Standards, New Delhi
10. Gaur, V.K., Chander, R., Sarkar, I., Khattri, K.N., Sinvhal, H.: Seismicity and the state of stress
from investigations of local earthquakes in the Kumaon Himalaya. Tectonophys. 118, 243–252
(1985)
11. Kumar, S., Chander, R., Khattri, K.N.: Compressional wave speed in the second crustal layer
in Garhwal Himalaya. J. Assoc. Expl. Geophys 8, 219–225 (1987)
12. Kumar, A.: Microearthquakes—Recording and study of their parameters for selected regions,
Ph.D. Thesis, University of Roorkee, India (1988)
13. Khattri, K.N., Chander, R., Gaur, V.K., Sarkar, I., Kumar, S.: New seismological results on the
tectonics of the Garhwal Himalaya. Proc. Indian Acad. Sci. (Earth Planet. Sci.) 98, 91–109
(1989)
14. Kumar, A., et al.: Processing and preliminary Interpretation of digital data obtained from
telemetred seismic array in the Garhwal Himalaya (1994)
15. Sharma, M.L., Gupta, S.C., Narayan, J.P., Das, J., Jindal, A.K., Jain, S.K., Sen, A.: Report on
processing and interpretation of seismological data collected from January 2019 to December
2019 through a local seismological network around Tehri region, submitted to THDCIL,
Rishikesh, EQ: 2020-09, Earthquake Engineering Studies IIT Roorkee (2020)
16. Lee, W.H.K., Lahr, J.C.: Hypo71: A computer program for determining hypocenter, magnitude
and first motion pattern of local earthquakes. USGS Open File Rep. 75–311, 1–116 (1975)
17. Havskov, J., Ottemöller, L.: Seisan: the earthquake analysis software. For Windows, Solaris,
Linux and Macosx Version, 8 (2008)
18. Chander, R., Sarkar, I., Khattri, K.N., Gaur, V.K.: Upper crustal compressional wave velocity
in the Garhwal Himalaya. Tectonophys. 124(1–2), 133–140 (1986)
19. Kumar, A., Pandey, A.D., Sharma, M.L., Gupta, S.C., Kamal, A.K.: Report on processing and
interpretation of seismological data collected from January 1995 to December 1995 through
a local seismological network around Tehri region, submitted to THDCIL, Rishikesh, EQ:
1996–06, Earthquake Engineering Studies IIT Roorkee (1996)
20. Kanaujia, J., Mitra, S., Gupta, S.C., Sharma, M.L.: Crustal anisotropy from shear wave spilliting
of local earthquakes in the Garhwal lesser Himalaya. Geophys. J. Int. 219(3), 2013–2033 (2019)
21. Sharma, M.L., Gupta, S.C., Jindal, A.K., Jain, S.K., Sen, A., Kumari, N., Arora, H.L., Saxena,
P., Vyas, A.P., Singh, R., Pal, I.: Seismicity studies for HE projects–a case study of Tehri Dam,
International Dam Safety Conference 2018, Jan. 23–24, Thiruvananathpuram, Kerala (2018)
22. Sharma, M.L.: Seismic hazard in Northern India region seismological research letters, 74(2),
March/April 2003, 140–146 (Impact factor: 1.826) (2003)
23. Bajaj, S., Sharma, M.L.: Modeling Earthquake Recurrence in the Himalayan Seismic Belt
Using Time-Dependent Stochastic Models: Implications for Future Seismic Hazards. Pure
Appl. Geophys. 176(12), 5261–5278 (2019)
24. Goswami, N., Gupta, S.C.: Determination of characteristics of the microearthquakes occurring
in Lesser and Higher Himalaya India, Act. Geophys. 70(1):1–13. https://doi.org/10.1007/s11
600-021-00717-z
25. Choudhary, C., Sharma, M.L.: Global strain rates in western to central Himalayas and their
implications in seismic hazard assessment. Nat. Hazards 94(3), 1211–1224 (2018)
26. Sharma, M.L., Gupta, S.C., Narayan, J.P., Das, J., Jindal, A.K., Jain, S.K., Sen, A.: Report on
processing and interpretation of seismological data collected from January 2020 to December
2020 through a local seismological network around Tehri region, submitted to THDCIL,
Rishikesh, EQ: 2021-04, Earthquake Engineering Studies, IIT Roorkee (2021)
27. Coldwell, J.G., Forohlich, C.: Microearthquake study of the Alpline fault zone near Heast,
South Island, New Zealand. Bull. Seism. Soc. Am. 65, 1097–1104 (1975)
28. Gutenberg, B., Richter, C.F.: Seismicity of the Earth, 1st edn., 273 p; 2d edn., 310 p. (Princeton
University Press, Housner, G. W., 1954)
29. Sharma, M.L., Wason, H.R.: Occurrence of low stress drop earthquakes in Garhwal Himalaya
region. Phys. Earth Planet. Int. 85, 265–272 (1994)
30. Ibenbrahim, A., Ni, J., Salyards, S., Ali, I.M.: Induced seismicity of the Tarbela reservoir,
Pakistan. Seismol. Res. Lett. 60(4), 185–197 (1989)
31. Gupta, H.K., Rastogi, B.K., Narain, H.: Common features of reservoir-associated seismic
activities. Bull. Seismol. Soc. Am. 62, 481 (1972)
32. Gupta, H.K., Rajendran, K.: Large artificial water reservoirs in the vicinity of the Himalayan
foothills and reservoir-induced seismicity. Bull. Seismol. Soc. Am. 76(1), 205–215 (1986)

17 TH Symposium

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

17 TH Symposium

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

17 TH Symposium

Uploaded by

Copyright:

Available Formats

Lecture Notes in Civil Engineering

To submit a proposal or request further information, please contact the appropriate

ISSN 2366-2557 ISSN 2366-2565 (electronic)

Probabilistic Arias Intensity Maps of Uttarakhand State (India) . . . . . . . 1

Assessment of Double Resonance from Microtremor Observations

Seismic Landslide Hazard Assessment of Mandi Town, Himachal

A Variational Mode Decomposition Approach for Modal

Characteristics of Strong Ground Motions for Delhi National

Dr. Manish Shrikhande is Professor and Head of the Department of Earthquake

Dr. Pankaj Agarwal is Professor at the Department of Earthquake Engineering,

Dr. P. C. Ashwin Kumar is Assistant Professor at the Department of Earthquake

Kunal Gupta and Neelima Satyam

Abstract Uttarakhand, an Indian state in the Western Himalayan region, is highly

Keywords Arias intensity · PGA · Uttarakhand · PSHA · Logic tree · Attenuation

K. Gupta (B) · N. Satyam

The Indian subcontinent’s northern state, Uttarakhand (Western Himalayas), has a

hazard estimates. The spatial distribution of hazards due to earthquake shaking in

2 The Geological and Seismo-tectonic Framework

Fig. 1 Geographical location of the study area: a India b Uttarakhand state

Alpine-Himalayan seismic belt, this underthrusting has resulted in continual defor-

3 Seismic Source Zone Delineation

4 Development of Earthquake Catalogue

Mw = 0.762M M I + 0.865 (1)

where MMI is Modified Mercalli Intensity.

5 Estimation of Recurrence Parameters

5.1 Maximum Magnitude (m max )

and fault characteristics. The non-parametric Gaussian estimator proposed by [30]

6 Ground Motion Prediction Models for Arias Intensity

log Ia = −4.1 + M − 2 log R ± 0.44 (4)

log Ia = −4.9 + 0.98M − 1.35 log R (5)

Travasarou et al. [34] discussed about the usefulness of employing Ia to evaluate

Table 1 FPA model

7 Logic Tree Structure

The choice of a seismic hazard model is challenging because of the uncertainty

8 Seismic Hazard Assessment

Fig. 5 Logic tree framework adopted in the study

9 Results and Discussions

Fig. 6 Arias intensity map for 475 years return period

Fig. 7 Arias intensity map for 2475 years return period

1. Rajendran, K., Parameswaran, R.M., Rajendran, C.P.: Seismotectonic perspectives on the

M. P. Hari Padmanabhan, Siddhardha R, Sreevalsa Kolathayar,

Abstract In the present study, seismicity data analysis of Sree Padmanabhaswamy

Keywords Gutenberg · Richter · Statistical analysis · Foreshocks · Seismicity

Sree Padmanabhaswamy Temple is located in the center of the circle-shaped research

Fig. 1 Geographical extent of the study region

2 General Analysis of Earthquake Catalog

Making sure a recurrence connection is complete; it is essential to confirm that all

4.1 Visual Cumulative Method

Table 1 List of major faults

5 Gutenberg–Richter Recurrence Relationship

Fig. 2 A Histogram of seismic events in the research area

Fig. 3 Map showing the research region’s seismicity

Cumulative number of Earthquakes 2

Fig. 4 (a–e) CUVI method for catalog completeness

(d) Mw = 4.5 to 4.99

Gutenberg Richter plot

-1 log λM = -0.5321Mw + 1.2349