1.

INTRODUCTION

1.1. Background

Accurate measurement of rainfall on global and regional scale are critical for understanding the
climate and hydrological cycle, simulating land surface hydrological processes, water resources
management and security (Stage et al., 2017; Tan et al., 2016) and it is also an imperative input
for hydrological models (Habib et al., 2014).

Variations in rainfall often influenced by non-linear relations between numerous factors like local
variations of topography, the orientation of mountains and aspects (Haile et al., 2009) which clues
directly to flexible discharge output (Biemans, 2012) and fluctuations in water storage (Narjary
and Kamra, 2013). Finally, this leads to incorrect simulation results and sometimes even in wrong
conclusions (Vrgut et al., 2005). Precipitation data can be collected directly through ground-based
observations using rain gauges and weather radars (Zhang et al., 2014), and indirectly through
satellites (Lopez et al., 2017). Still, ground rainfall stations in many parts of the world and most of
parts of Ethiopia are very sparse and unevenly distributed (Ayehu et al., 2017).

Thus, rainfall estimation is unavoidably subject to error due to different factors depending on the
type of measurement (Alemohammed et al., 2015). To overcome some of the above-mentioned
restrictions of ground-based precipitation measurements, space-based (satellite) estimations of
precipitations provide a hopeful alternative source (Bigiarini et al., 2017). Satellite-based
precipitation estimates may offer plentiful information with spatio-temporal (near-real-time) high-
resolution over wide-ranging regions wherever conventional rainfall data are unusual or absent
(Kidd and Huffman, 2011; Kidd et al., 2009).

Since combined precipitation products mostly have more improved quality than individual data
set (Xie and Xiong, 2011), different emerged precipitation products (reanalysis-gauge and
satellite-gauge) are gradually being developed (Duan et al., 2017).It is principally worth nothing
that the lately released Climate Hazards Group Infra-Red Precipitation with Station data
(CHIRPS_2) provides rainfall at the finest spatial resolution of 0.05o and 0.25o (Funk et al., 2015).

At different times many scholars have been carried out a study to assess the performance of satellite
precipitation estimates for hydrological modeling in different regions of the world including
Ethiopia and indicated that high resolution satellite precipitation products have potential use for
hydrological modeling based on their product types.

Hence, this study was intended to assess the suitability of CHIRPS_2 and Tropical Rainfall
Measuring Mission (TRMM) Multisatellite Precipitation Analysis,TMPA_3B42v7 satellite
precipitation products besides in-situ precipitation products for stream flow simulation in Dabus
watershed, Upper Blue Nile basin, Ethiopia in a more comprehensive way in-terms of evaluated
aspects, temporal and spatial scales.

Subsequently measurement of all parameters that affect watershed’s stream flow simulation is
difficult, choosing a suitable model of simple structure, minimum input data requests and
reasonable precision is essential (Sharifi et al., 2004).

One of the hydrological models that meet the listed criteria around the globe is Hydrological
Engineering Center-Hydrologic Modeling System (HEC-HMS). Due to the above reasons and its
confirmed competence in the Upper Blue Nile basin (Fekadu et al., 2017; Onyutha, 2016; Gebre,
2015), this study used HEC-HMS hydrological model to evaluate the highly resolution satellite
precipitation products for stream flow simulations.

1.2. Statement of the Problem

Spatial distribution and extent of rainfall are imperative for establishing a concrete flood
prediction, warning services and drought monitoring. But, in many regions of the world the number
of ground measuring stations is very scarce and randomly distributed which make countries to
have poor (humble) water resources assessment and flood prediction (Shrestha et al, 2011; Yong
et al., 2010).

In order to challenge the above doubts, a large number of academics have been exploring that
satellite rainfall assessments have potential use for hydrological modeling, but model performance
and uncertainty depends on way of the model developed, physical condition of application,
external factors (data), satellite product type, watershed size, hydro climatic regimes ( Liu, 2015;
Gebregiorgis and Hossain, 2013) and some behaviors of models are still difficult to totally
understand and can hide surprising properties (Berthet et al., 2010a).

Subsequently, all the above evaluations did not include Dabus watershed of Ethiopia which is
rugged in topography and have more sparsely in-situ rain gauge networks. Additionally, most of
the studies were limited to satellite precipitation products of early version and only event-based
analysis for examples, in (Tesfaye et al., 2017). Moreover, in Dabus watershed
the influences of water withdrawal, land cover and climate change have not been quantified yet
due to scarce rainfall data for rainfall-runoff modeling.

As a result, this study concentrates on evaluating satellite rainfall products of CHIRPS_2 and
TMPA_3B42v7 estimate by using in-situ rain gauge products based on their hydrologic simulation
result as well as identification of sensitive parameters of model out puts when calibrated and
validated with input specific satellite precipitation estimates. The HEC-HMS hydrologic model
was used to simulate the stream flow for both rain gauge and satellite rainfall products.

1.3. Objectives

1.3.1. General Objective

The main objective of this research is to evaluate CHIRPS_2 and TMPA_3B42v7 satellite rainfall
products for hydrologic simulation in Dabus watershed, Abbay basin, Ethiopia.

1.3.2. Specific Objectives

The specific objectives of this research are;

 To evaluate the ability of satellite rainfall products to characterize rainfall patterns and
capture the magnitude of rainfall over Dabus watershed.
 To assess the skill of satellite rainfall products as an input into a hydrological model for
stream flow simulation.
 To evaluate the performance of HEC-HMS model using satellite and ground based
precipitation products.
1.4. Research Questions

 Is CHIRPS_2 or TMPA_3B42v7 satellite precipitation product is relatively suitable for

stream flow simulation using HEC-HMS models in the study area?
 Is high resolution satellite or in-situ rain gauge rainfall data’s being comparatively suitable
for hydrological simulation in the study area?
 Does bias correction of satellite precipitation products has effects on stream flow
simulation in the study area?
1.5. Significance of the Study

Even if, in-situ gauging stations exist in the study area the distribution is very inadequate, random
and less reliable. Therefore, to overcome such type of problems this study tried to evaluate high
spatial and temporal resolution of satellite rainfall products for hydrological simulation in order to
add confidence on reality of satellite products in place of inaccessible (data scarcity) area. The
study also may be helpful for hydrologists, climate scientists and researchers in real life (design)
applications mainly, for the study area.

1.6. Scope and limitation of the Study

This study is limited on evaluating CHIRPS_2 and TMPA_3B42v7 satellite precipitation products
by using in-situ rain gauge precipitation data’s with HEC-HMS hydrologic model for Dabus
watershed, Abbay basin, Ethiopia. Primarily, simulation of observed rainfall data was takes place;
secondly, simulation of satellite precipitation data’s was performed.

Finally, comparison of in-situ rainfall data’s with satellite rainfall products and comparison
between the two satellite precipitation products were achieved. This study doesn’t take in to
account stream flow changes caused due to model inputs other than rainfall.

1.7. Structure of the Thesis

This thesis work is organized in five chapters, references and appendices. The first chapter consist
of introduction, background, statement of the problem, objectives of the study, research questions,
significance and scope of the study. Chapter two review the work done earlier on related topics.
Chapter three describes about location and background of the study area, methodology, data source
and methods for data collection, analysis systems and conceptual framework.

Chapter four presents the outcomes and discussions based on the objectives of the study. Chapter
five contain the conclusions and recommendations of the study depending on the results, whereas
the list of references includes the references of different works cited in the manuscript. Finally, list
of the appendices in the form of tables and figures are presented as a supporting document to this
study.
 2. LITERATURE REVIEW

2.1. Rainfall

Rainfall is a main component of the climate system and plays a critical role in the Earth’s
hydrological cycle and energy balance. Conventionally, the rain gauge is the main source of
precipitation data, which has been the most accurate and reliable approach for rainfall data
measurement (Ayehu et al., 2017). However, ground rainfall stations in many parts of the world
and most parts of Ethiopia especially, in relatively remote and mountainous areas are very spars
and unevenly distributed.

Correspondingly, rainfall data with very fine resolution and high quality are important for
hydrological computations (Zhou et al., 2017), in turn hydrological model output is highly affected
by precipitation input uncertainty ( Kamali et al., 2017; Haile et al., 2010). In spite of its
significance role in our lives and global climate, the measurement of precipitation is very difficult
due to its high spatial and temporal variability’s (Rodriguez et al., 2015b).

In practice, the density and configuration of rain gauge network is determined based on the
availability of funds, accessibility of site and the purpose of the network. As a result, most of the
time in-situ rain gauges instruments in developing countries are universally installed in towns that
located nearby main roads to provide accessibility.

In contrast, satellite remote sensing can make available the spatial precipitation data over large
areas in a temporally continuous way for various hydrological studies since 1970s’ (Liu et al.,
2014). However, due to the lack of quantitative confidence in the accuracy, these satellite rainfall
products are not often used for different operations (Burkea et al., 2017).

Over the past decades, countless efforts have been made to generate gridded precipitation products,
thus leading to the increasing accessibility of rainfall dataset at different spatial and temporal
resolution over the global or quasi-global scale (Schneider et al., 2015) and can be broadly
categorized into four classes (Duan et al., 2016).

1). Gauge-only products that build only on observations from in-situ rain gauge stations using
diverse interpolation methods which are often accessible at coarser spatial resolution (> 0.5º) 2).
Rainfall products from numerical weather forecasts or atmospheric models that uses a blend of
satellite and in-situ observations of several atmospheric properties as inputs (Ebert et al., 2007).
3). Satellite remote sensing-only products. 4). Satellite-gauge products that combine two
individual gauge-only and satellite-only data’s together through different bias correction or
blending technique and which products are often existing at the spatial resolution of 0.25º or finer.

2.2. Satellite Rainfall Products

Satellite is an object which has been purposely placed into an orbit to do specific work. Satellite
system that provide weather (precipitation) related observations are separated in to geostationary
GEO satellites and low earth orbiting LEO satellites (Aoki, 2017). Imagery from geostationary
satellite systems, when compared to polar-orbiting satellite information, normally, results in a
reduction of the sampling errors at all temporal scales.

Consequently, at the present time, estimates of global, tropical and subtropical precipitation at
short timescales (from few hours to perhaps a few weeks) are most accurately obtained from
geostationary satellites (Tang et al., 2016). Algorithms to estimate precipitation data from satellite
observations are based either on thermal infrared (TIR) bands or on passive microwave (PMW)
sensors (Duan et al., 2016), which have their own specific strengths and limitations.

Thermal infrared sensors on geostationary satellite produce high spatial coverage with continuous
temporal coverage, covering nearly the complete globe every one hour or less. Thermal infrared
information based on the cloud-top brightness is frequently valuable for the distinction between
raining and non-raining time (Yang et al., 2017). Several studies displayed that TIR-based
precipitation evaluations may have a large uncertainty when some types of cloud or warm clouds
are present (Huang et al., 2017).

On the other hand, the passive microwave sensors on polar-orbiting satellite recognize the rainfalls
particles by scattering due to the large ice particles present in the clouds. PMW is better at
approximating the precipitation quantity owing to the increased direct physical connection
between the sensor signal and precipitation, but runs at a much lower temporal frequency and
coarser spatial resolution (Bergsted et al., 2017).

To deal with these technical restrictions, the more recently satellite-based rainfall products, such
as CHIRPS_2 and TMPA_3B42v7 combine multiple data sources as TIR-PMW based rainfall data
sets.
2.2.1. CHIRPS_2

The Climate Hazards Group Infra-Red Precipitation with Station data (CHIRPS version 2) is a
global daily, pentadal (5), monthly precipitation product and a third generation rainfall procedures
clearly designed for monitoring agricultural drought and global environmental change over land
(Funk et al., 2015), which was developed by the U.S Geological Survey and the Climate Hazards
Group at the University of California, Santa Barbara in 2015.

CHIRPS_2 rainfall products mainly organized in support of drought-related issue in Africa

(Climate Hazard Group, 2016), principally for Ethiopia, Afghanistan or Haiti (Funk et al., 2015).

CHIRPS product which belongs to the “satellite-gauge” rainfall source category combines
remotely sensed precipitation of geo-synchronous and polar orbiting satellites, from five different
most recommended satellite precipitation products, with more than 2000 station records to
calibrate global Cold Cloud Duration (CCD) precipitation estimates (Funk et al., 2015). The
product structures at a spatial resolution of 0.05º and 0.25ºfrom 50º S to 50º N across all longitudes
with more than 30-year final monthly rainfall records (1981-present).

The station statistics from the near-real time World Meteorological Organization’s Global
Telecommunication System (GTS) are constantly used to update roughly every two days, and
validate the remote sensors using extra evidence (physiographic and remotely sensed earth energy
emissions that agree to the location and intensity of rainfall) through moving window geo
statistical regression (Funk et al., 2015).The main used datasets for the creation of CHIRPS_2
product includes: -

1). The monthly Climate Hazard Precipitation climatology (CHPclim) that is molded with in-situ
rain gauge stations.

CHPclim is composed from the Agro-met Group of the Food and Agriculture Organization of the
United Nations (FAO), the Global Historical Climate Network (GHCN) and the Cold Cloud
Duration (CCD) information based on thermal infrared data archived from Climate Prediction
Center (CPC) and National Oceanographic and Atmospheric Administration (NOAA) National
Climate Data Center (NCDC).
2). The Version_7 TRMM 3B42 data. 3). The Version_2 atmospheric model precipitation data
field from the NOAA Climate Forecast System (CFS) and 4). The rain gauge stations data from
several sources. Primarily, the CCD records are calibrated with TRMM-3B42 to generate the five
daily CCD-based precipitation estimates which are further reformed to the fractions of the long-
term mean precipitation estimates.

The fractions are then multiplied with CHPclim data to eliminate the systematic bias and the
derived product is called CHIRP product. Finally, the CHIRP rainfall product is merged with rain
gauge stations information using a modified Inverse Distance Weighting (IDW) algorithm to
produce the CHIRPS_2.

All the process indicated above are accomplished at the five-daily (pentadal) time scales. The daily
CCD measurements and daily CFS measurements are finally used to disaggregate the 5-daily
products to daily rainfall estimates using a simple redistribution method which data is available at:
http://chg.geog.ucsb.edu/data/chirps/ with resolution of 0.05ºand 0.25º.

2.2.2. TMPA_3B42v7

The Tropical Rainfall Measuring Mission (TRMM) Multisatellite Precipitation Analysis (TMPA)
is a joint US and Japan satellite mission, which was launched in 1997 to monitor tropical and
subtropical precipitation data and to estimate its related latent heating covering the latitude band
50º N – 50º S.

It also intended to provide a best estimate of quasi-global precipitation data from a wide variety of
modern satellite-born rainfall related sensors. Its rainfall estimates are provided at relatively high
spatial resolution (0.25º×0.25º) and 3-hourly time stapes, in both real and post-real time to meet a
wide range of research needs (Huffman et al., 2010).

This satellite rainfall estimates combines IR data’s from geosynchronous earth orbit (GEO)
satellites with four passive microwave (PMW) sensors, namely TRMM Microwave Imager (TMI),
Special Sensor Microwave/Imager (SSM/I), Advanced Microwave Sounding Unit-B (AMSUB),
and the Advanced Microwave Scanning Radiometer – EOS (AMSR-E).

The TMPA_3B42v7 satellite precipitation estimates include the version (6 and 7) real-time
products 3B42RT (3B42RTv6 and 3B42RTv7), the 3-hourly research data 3B42 (3B42v6 and
3B42v7), and the daily accumulated precipitation products 3B42_daily (v6 and v7).
The 3B42 algorithm is executed in four steps:1). PMW rainfall estimates are calibrated and
combined; 2). TIR precipitation estimates are generated using the calibrated PMW data; 3). Both
TIR and PMW data are then combined and 4). Rescaled on a monthly basis using two sources of
rain gauge records: such as,

i). The Global Precipitation Climatology Center (GPCC) monthly rain gauge analysis and ii). The
Climate Assessment and Monitoring System (CAMS) monthly rain gauge analysis developed by
CPC (Bigiarini et al., 2017).

The latest version of TMPA_3B42v7 rainfall product was released in June 2012, and recent studies
indicated that 3B42v7 estimates improve upon 3B42v6 (Bigiarini et al., 2017; Chen et al., 2013).
Nowadays, after more than 17 years of data collection, the instruments turned off on 8, April, 2015,
after the spacecraft exhausted its fuel reserves and replaced by Global Precipitation Measurement
GPM mission. For this thesis work TMPA_3B42v7 daily data’s was used and can be easily
accessed from (https://pmm.nasa.gov/).

2.3. Previous Studies around the World and Ethiopia

Several studies have been carried out to evaluate a single or multiple satellite precipitation products
on stream flow simulation capability with and without using hydrological models in scale from the
global or quasi-global (Massari et al., 2017), continental scale (Maidment et al.,2017), Country-
wide (Maidment et al., 2017), regional scale (Haile et al., 2016), basin scales (Duan et al., 2016)
and watershed scales (Liu et al., 2014; Stage et al., 2014).

However, an insufficient studies have been conducted in Ethiopia, for instances, (Bhatti et al.,
2016) evaluated CMORPH satellite rainfall products for Upper Gilgel Abay catchment. Habib et
al., (2014) assessed effect of bias correction on CMORPH satellite product. Haile et al., (2012)
evaluated CMORPH satellite rainfall products for Gilgel Abay watershed.

Gebremichaiel and Romilly, (2011) evaluated three satellite rainfall products (CMORPH,
PERSIAN and TRMM) for six Ethiopian river basins. The condition is primarily remarking that
because of latest availability of the (CHIRPS_2 and TMPA_3B42v7) high resolution satellite
rainfall products, very few investigations have been done to evaluate and compare the products.
Table 2.1 below show earlier studies about satellite precipitation products, the method they used
and their main gaps.
 Table 2.1: Earlier Studies around the World and Ethiopia

Evaluated satellite Method Gaps of the study Main output/result/ References

rainfall products used for
evaluation
CHIRPS, SWAT The assessment did CHIRPS performed best (Duan et
TRMM_3B42, model not considered with BIAS of < 0.05 and al.,2016).
CMORPH_RAW, effect of elevation CMORPH_RAW products
CMORPH_CRT, difference and the performed the worst. All
CMORPH_BLD, spatial resolution of products showed large error
PCDR, the compared during low rainfall seasons.
GSMaP_MVK, products were not
and PGF the same.
CHIRPS, Categorical, The assessment of CHIRPS exhibited higher (Ayehu et
TAMSAT_2, Volumetric the products was skill and less affected by al., 2017).
TAMSAT_3 and indices and dependent only on elevation variation, while
ARC_2 Statistical event based ARC-2 products performed
measures analysis, which the weakest value which
may lead to wrong performance dependent on
conclusion. elevations.

NOAA Climate Geospatial The simulated outcome (Bajrachar

The findings did
Prediction Center Stream using modified NOAA RFE ya et al.,
not considered the
Rainfall Estimates Flow data performed well with 2017).
effect of different
Version 2.0 (RFE) Model observed discharge in-terms
rainfall seasons on
(GeoSFM) of timing and magnitude of
the products.
stream flow in the basin
than the un-modified
NOAA RFE data.
CHIRPS, CHIRP, Suite of The evaluated Even if, TAMSAT data (Maidment
TAMSAT-2, binary skill products were achieved best for all skill et al.,
TAMSAT-3, scores, different in terms of measures, except bias it is 2017).
RFE-2, Categorical temporal and not suitable for providing
TMPA_3B42, indices, spatial resolution. information on pluvial
CMORPH and Volumetric flood risk, whereas
ARC-2 indices and CHIRPS scored the
Statistical smallest bias than others
measures satellite rainfall products.
TRMM 3B43V7, Triple The methodology CHIRPS v1.8 products (Hessels et
GPCP V2.2, Collection used to evaluate the showed the best correlation al.,2015).
GPCP 1DD, Method products were not for mountains areas like
CRU TS3.21, satisfactory and did Upper Blue Nile basin,
ERA-Interim, not include all time while ARC-2.0 and RFE
RFE2.0, ARC-2.0, scale (day, time and 2.0 showed the poorest
CHIRPS V1.8, year). correlation in Nile basin but
PERSIANN, in regions with more
CMORPH, convective storms these
GSMaP-MVK, products could be superior.
GPCC and
TAMSAT
ARC-2, Statistical Since the study area Even if, ARC-2 (Novella
GPCP,CMAP, measures was large, the underestimate both daily and Thiaw,
PREC/L, 3B42V6, (RMSE, % number of gauges and monthly validations 2012).
and CMORPH BIAS) used to validate the due to absence of gauge
satellite products reports in real time, the
and the evaluation researchers were suggested
that ARC-2 products might
 system were not be important for
sufficient. understanding climate
change and variability.
ARC-2, CHIRPS, Pairwise The evaluation was Generally, CHIRPS product (Bayissa et
PERSIAN, comparison not considered could be used to develop al., 2017).
TARCAT and statistics different time scale operational drought, flood
TMPA techniques. and the data that monitoring and early
collected to warning system, whereas
validate the satellite TMPA could be used in the
products were not water resources application
enough. due to its best performance
in capturing rainfall
amount.
ARC-2, RFE-2 The 3rd The spatial and Based on their performance, (Dembele
CHIRPS, Algorithm temporal resolution ARC, RFE, and TARCAT and Zwart,
PERSIANN, Inter- of the evaluated suggested for drought 2016 ).
TAMSAT, comparison products were not monitoring and
TARCAT and Project of the same. PERSIANN, CHIRPS and
TRMM GPCP. TRMM at daily scale
assessment selected for
flood monitoring process in
the study area.

CHIRPSv2, Continuous Since full version Generally, all products (Bigiarini

CMORPH, and of PGFv3 were not performed best during the et al.,
TMPA_3B42v7, categorical released it might be wet seasons and all satellite 2017).
PERSIANN_CDR,P indices, cause for wrong rainfall products except
ERSIANN_CCS- including conclusion. PGFv3 which reliant on
ADj, PGFv3, modified elevation.
MSWEPv1.1 Kling-
 Gupta
efficiency

CHIRPSv2 Pair-wise The evaluation CHIRPSv2 showed that an (Trejo et

comparison, system was limited, over-estimation of lower al., 2016).
Categorical only in rainfall monthly rainfall values and
indices and seasons without an under-estimation of
Statistical considering higher values (≥100
measures different time scale mm/month), on the other
resolution. hand it achieved better
during the rainy season for
open and flat areas.
ARC,CHIRPS, Pairwise The validation Both CHIRP and CHIRPS (Dinku,
CHIRP and comparison systems were not achieved better than ARC 2014).
TAMSAT statistics standard (the data and TAMSAT. But
techniques that collected were CHIRPS, which includes
not at the same stations, perform worse
temporal than the satellite only
resolution). (CHIRP) products.
CHIRPS, Pairwise The evaluation CHIRPS products achieved (Tote et al.,
RFEv2.0 and comparison system was not well in all major statistical 2015).
TARCAT v2.0 statistics and fully considering measurements and it
Categorical the effects of recommended for dry spell
indices altitude difference assessments, whereas
in the study area. TARCAT performed best in
detecting rainfall amounts.
TMPA HEC-HMS Bias corrected TRMM- (Haile et
(TRMM-3B42, Hydrologic 3B42RT rainfall estimates al., 2016).
TRMM- model performed better than
3B42RT) and ECMWF based flood
ECMWF forecasts.

CHIRPS, HYMOD_ The spatial and In monthly scales CHIRPS (Pineda et

CMORPH, DS temporal resolution products performed best al., 2016).
PERSIANN and hydrologic of the evaluated result whereas, at all
TRMM-3B42RT model products were temporal scales CMORPH
different. and TRMM-3B42RT
showed grater correlation in
the basin highlands.
CHIRPS and SWAT The evaluation CHIRPS showed its (Le and
CFSR model system was only powerfulness for water Pricope,
limited on daily resources management and 2017).
time scales. stream flow volume
estimation in data scarce
areas and suggested only
for flat, poorly gauged and
small scale watershed.

In conclusion, based on the studies carried out in different regions of the world as well as in
Ethiopia, the evaluated satellite precipitation estimates showed that satellite precipitation products
have potential use for hydrologic modeling, nevertheless the performance of the
hydrological model for stream flow simulation were depends on the type of satellite rainfall
product, basin area, hydro-climatic regions, length of simulation periods and topography of the
study area.

2.4. Hydrological Model

Hydrological models are picture of the real world system and can be classified as physical,
conceptual and empirical based on the degree of complexity and physical completeness in the
establishment of the structure. Models are further classified as lumped, distributed and semi-
distributed depending on the degree of reorganization when describing the terrain in the basin
(Cunderlik, 2003).

Lumped models:- In this models the hydrologic parameters do not vary spatially within the
watershed and its response is evaluated only at the outlet, without explicitly account for the
response of individual sub-basins (Cunderlic, 2003).

Distributed models are generally require large amount of data in each grid cell. If governing
physical process are modeled in detail and suitably applied, distributed models can provide the
highest degree of accuracy. In semi-distributed model, the parameters of the models are allowed
to vary partially in space by dividing the basin in to a number of smaller sub-basins.

In deterministic models only one outcome is permit from a simulation with one set of inputs and
parameter values, whereas stochastic models allow for some randomness or uncertainty in the
possible results due to uncertainty in input variables, boundary conditions or model parameters
(Beven, 2000).

Conceptual models are constructed on the limited picture of the physical process acting to produce
the hydrological outputs, while physically based models are built more solidly on understanding
of the related process (Ward and Robinson, 2000).

According to Beven, (2000) many rainfall-run of models are carried out purely for research
purposes as a means of improving knowledge about hydrological systems. In addition to this,
models are developed as tools for simulation and prediction aiming ultimately to allow decision
makers to improve decision making about hydrological problems.

2.4.1. Description of HEC-HMS Hydrologic Model

Hydrologic Engineering Center-Hydrologic Modeling System (HEC-HMS) which is well known

as the semi-distributed conceptual model, developed by the United States Army Corps of
Engineers (USACE, 2008).

It is designed to simulate the broad hydrologic process of dendritic watershed system by including
many traditional hydrologic analysis procedures such as event infiltration, unit hydrographs, and
hydrologic routing.
It also includes processes necessary for continuous and gridded runoff simulation including
evapotranspiration, snowmelt and soil moisture accounting. The program features have an
integrated work environment with data entry utilities, computation engines and outcome reporting
tools.

Simulation results are stored in the data storage system HEC-DSS and can be used in union with
others software for studies of water accessibility, urban drainage, flow forecasting, future
urbanization influence, reservoir spillway design, flood damage reduction, flood plain regulation
and system operations (USACE, 2016).

The Geospatial Hydrologic Modeling System extension (HEC-GeoHMS) tools has been
developed as a geospatial hydrology toolkit for engineers and hydrologists with inadequate GIS
understanding (Alazzy et al., 2017). Examining digital terrain data, HEC-GeoHMS transforms the
drainage paths and sub-basin boundaries into a hydrologic data structure that describes the
drainage network.

The model allows user to visualize spatial information, to document watershed features, to perform
spatial investigation, and to delineate sub-basins. The currently free program used for this thesis
was version 4 (HEC-HMS 4.2.1) and focused principally on new computation features which has
an extensive array of capabilities for conducting hydrologic simulations.

HEC-HMS hydrological model is selected on the basis of applicability and restrictions of each
method, availability of data, well recognized, stable, widely acceptable, a lot of scholar
recommendations, simple, powerful tool and its versatility. The model can care to save time and
money in attaining the runoff data rather than measurement of runoff in the watershed.

The HEC-HMS hydrologic model was found precise in spatially and temporally forecasting
watershed response in event based and continuous simulation along with simulating various
scenarios in flood forecasting and early warnings. It is freely accessible for download from the
HEC-HMS website (www.hec.usace.army.mil) and is supported by the US Army Corps of
Engineers. Previous studies used method and range of parameters are presented in below table 2.2
and table 2.3 respectively.
 Table 2.2: Method available in the sub basin and reach hydrologic elements

Hydrologic Calculation Methods Used method and

Elements Type References

Sub basin canopy Dynamic ,Simple (gridded) Dynamic, (Cunderlic. J

and Simonovic. S, 2004).

Surface Simple( gridded)

Deficit and constant (gridded), SCS, for event based

Loss rate Exponential, Green and ampt simulation (Yang et al.,
(gridded), Initial and constant, 2014). Deficit and
SCS curve number (gridded), constant, for continuous
Smith parlange, Soil moisture based simulation (Haile
accounting (gridded). et al., 2016; USACE,
2013).

Transform Clark unit hydrograph SCS unit hydrograph

Kinematic wave Mod Clark (Haile et al., 2016;
SCS unit hydrograph Choudhari et al., 2014;

Snyder unit hydrograph Reshma et al., 2013;

User-specified s-graph Asadi and Boostani,

User –specified unit hydrograph 2013; Oleyiblo et al.,

2010).
 Base flow Bounded recession, Recession, Recession (Haile et al.,
Constant monthly, Linear 2016; Joo et al., 2014;
reservoir ,Non-linear Boussinesq
Choudhari et al., 2014).

Routing Kinematic wave, Lag Muskingum (Haile et al.,

Modified plus, Straddle Stagger 2016; Joo et al., 2014;
Reach Muskingum/ Muskingum-Cunge Asadi and Booftani,
2013; Majidi et al.,
2012).

Gain/Loss Constant, Percolation

Table 2.3: Calibration parameters value of HEC-HMS model used in the past studies

Modeling Method Parameter Allowable Past References

range studies
parameter
range
Routing Muskingum K (hr) 0.1-150 51 - 150 (Haile et al.,
X (-) 0 - 0.5 0.2 2016).
Base flow Recession Recession 0.000011- 0.4 - 0.78 (Joo et al.,
constant 10,000 2014).
0.42 (Haile et al.,
2016).
Ratio to 0–1 0.12 - 0.35 (Haile et al.,
peak 2016; Yang et
al., 2014;
Oleyiblo et
al., 2010).

Runoff SCS CN 0 – 100 51 - 89 (Amini et al.,

volume 2014; Joo et
al., 2014).
 Direct SCS unit Lag 0 – 30,000 3 - 4498 (Verma et al.,
runoff hydrograph time(min) 2009).
transforma Lag time 221- 396 (Haile et al.,
tion (hr) 2016).

Initial 0 - 100,000 0 – 39 (Asadi and

base flow Booftani,
(m3/sec) 2013;
Oleyiblo et
al., 2010).

2.4.1.1. Limitation of HEC-HMS Model

Every simulation system has its own restriction due to the selections prepared in the design and
development of the software (USACE, 2016). Some of the limitation of HEC-HMS models are; -

1. Simplified model formulation - simplifying the model formulation allows to the program to
complete simulation very quickly whereas generating accurate and precise results.

2. Simplified flow representation – simplifying the flow representation aids in keeping the compute
process efficient and reduces repetition of capability in the HEC software suite.

2.5. Simulating Stream Flow for Satellite Rainfall Products

There are two main different calibration methods for simulating stream flow from satellite
precipitation inputs:

1). Calibrating the model using rain gauge rainfall measurements and validating the model with
satellite rainfall products (Elgamal et al., 2017).

2). Calibrating and validating the model with the satellite rainfall products (Bajracharya et al.,
2017; Stisen and Sandholt, 2010).

However, compared to calibrating hydrologic model using in-situ rain gauge precipitation,
calibrating the model with consistent satellite precipitation data increases the performance of
satellite rainfall stream flow simulation but decrease the performances of satellite evapo-
transpiration simulations (Bitew and Gebremichael, 2011a).