See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/280547058

Unconventional seismic attributes for fracture characterization

Article in First Break · March 2015

DOI: 10.3997/1365-2397.33.3.79520

CITATIONS READS

56 5,413

3 authors, including:

Hardeep Jaglan Farrukh Qayyum

3 PUBLICATIONS 62 CITATIONS
Saudi Aramco
38 PUBLICATIONS 396 CITATIONS
SEE PROFILE
SEE PROFILE

All content following this page was uploaded by Hardeep Jaglan on 16 January 2018.

The user has requested enhancement of the downloaded file.

first break volume 33, March 2015 special topic
Modelling/Interpretation

Unconventional seismic attributes for

fracture characterization
Hardeep Jaglan1*, Farrukh Qayyum1 and Hélène Huck1 introduce several new geometrical
attributes that help in characterizing fault networks, fractures density and their connectivity.

F aults and fractures are generally represented as discon-

tinuous reflection patterns on seismic data and mostly
appear as linear, sub-linear or curvilinear features in
three dimensions. In seismic interpretation faults are
routinely detected through multi-trace seismic attributes
recalculated and are further optimized to obtain a fault/frac-
ture cube. The fault cube (such as similarity, curvature and
fault likelihood) is then treated through various new fracture
attributes (fracture proximity, fracture density, fracture
gradient and fracture azimuth) to perform interpretation,
such as coherency, similarity, semblance, variance, curvature, which are the core parts of this paper.
dip, azimuth, etc. These attributes are often grouped as geo- Once an end product is prepared, it is further visualized
metrical attributes because they mostly help in defining the through a voxel connectivity filter to produce the three
geometrical nature of seismic reflections. dimensional nature of the fault network and perform inter-
The objective of this paper is to introduce several new pretation. This is a crucial part of the proposed workflow
geometrical attributes that help in characterizing fault net- that produces an easy to interpret volume with ranked
works, fractures density and their connectivity. The proposed (numbered) geobodies. It opens the ways for interpretation
attributes are derived after specialized workflows based on such as fault connectivity vs. permeability, open vs. close
a precomputed dip-azimuth volume. These attributes are fault networks, fracture density, etc. The following sections
referred as unconventional fracture attributes because they further define and explain the entire workflow in detail.
are new and not routinely applied to fault detection and
characterization. Seismic scale faults/fractures
A dip-azimuth volume forms the heart of the workflow It is important to understand the defining criteria of faults
(Figure 1). It is computed in a sub-volume level to obtain and fractures on the seismic data. A fault is generally defined
local dips of seismic reflectors and associated discontinuities. on seismic data if there is a vertical displacement for a given
This volume is also known as SteeringCube and is computed seismic reflector. Such a vertical displacement can easily
based on various algorithms. Generally, two types of be identified through multi-trace correlation. On the other
algorithms are popular in extracting seismic dip and azimuth hand, seismic dispersion, scattering, diffraction and other
information. (1) A phase-based dip calculation utilizes seis- similar noises may impact on such correlations and final
mic phase attribute. (2) A FFT (Fast Fourier Transformation) results.
based algorithm that utilizes seismic frequency and ampli- It is well-known that the conventional seismic data
tude attributes. We always prefer phase-based dip calculation contains many pitfalls and, due to improper acquisition
for structural interpretation and frequency/amplitude based and/or processing, the imaging of geological faults and their
dip calculation for stratigraphic purposes. patterns may often not reflect the true picture of the subsur-
Once a SteeringCube is prepared, the conventional face. Specialized pre-stack processing steps are required to
attributes can be modified and redefined along the dip and preserve all structural components in the data. For structural
azimuth information. This makes most of the multi-trace interpretation, beam migration in depth domain is considered
seismic attributes dip-steered. The dip-steered attributes as an optimum choice compared to the traditional Kirchhoff
always have an advantage over the non-steered attributes migration algorithm.
because they represent more geological pictures of the sub- This is especially relevant for cases where multiple
surface rather than artefacts. The SteeringCube has further arrivals need to be considered in the presence of steeper dips
applications in data enhancement and in the preparation (Ben-Brahim et al., 2009). Such situations routinely occur in
of dip-steered filters on the seismic data. These filters are subsalt settings at large depths or basement faults. Alai et
applied in a sequential manner to obtain a final cube that al. (2014) showed the significance of seis-mic attributes in
is most representative of faults and fractures rather than characterizing basement fracture reservoirs by presenting
stratigraphy. Following this, the conventional attributes are several case studies. Their work was followed

1
dGB Earth Sciences, The Netherlands.
*
Corresponding author, E-mail: hardeep.jaglan@dgbes.com

© 2015 EAGE www.firstbreak.org 101

 special topic first break volume 33, March 2015

Modelling/Interpretation

Figure 1 A flow chart to create the proposed

unconventional fracture attributes and eventually
3D fracture geobodies from post-stack seismic data.

by testing the cases through wells, which was found positive.

A relatively modern processing technique of 5D interpolation
is also emerging (Trad, 2009) and helps to remove artefacts
(e.g., acquisition footprints) caused by suboptimally sampled Aiso, Biso, and Ciso are isotropic terms;
data during geometrical attribute computations (Chopra and
Marfurt, 2007). 5D interpolation also addresses the need of and
regularly sampled azimuths and offsets required by AVAZ
analysis and pre-stack inversion. are the HTI anisotropy terms;
In theory, a stress field is defined by three principal stress
components, which are oriented perpendicular to each other. θ = incidence angle, Φ= azimuth angle, δV = Thomsen’s δ
The magnitudes and orientations of these three principal (delta) parameter relative to vertical, εV = Thomsen’s ε (epsilon)
stresses are governed by the tectonic regime in the region and parameter relative to vertical, γ = Thomsen’s gamma parameter
by depth, pore pressure and rock properties. These parameters Another sophisticated method requires multi-component
control how stress is transferred and distributed in the sub seismic data to analyse shear wave polarization that occurs in
surface, and consequently the orientation and propagation HTI media. It can be a powerful method of characterizing ani-
direction of fractures. sotropy, fracture orientation and intensity (Bale et al., 2009).
In order to perform quantitative fracture interpretation In most cases, geoscientists are left with the conventional
using seismic data, one has to understand that conven- seismic data, particularly in exploration projects. The require-
tional seismic data do not contain such information unless it is ments of rarely acquired multi-azimuth or multi-component
adequately acquired in various azimuth and processed accord- seismic data, limits the usage of the two aforementioned
ingly. Faults and fractures are conceptualized as horizontal methods. This is where a number of geometrical seismic attrib-
transversely isotropic (HTI) media (Sengupta and Bachrach, utes come into play that can be explored for fault network
2006) which are mathematically expressed by Ruger (1997). delineation and fracture analysis.
Ruger suggested linearized approximations to the Zoeppritz
equation for estimating HTI anisotropy parameters from Data conditioning
azimuthal seismic P-wave data. This AVAZ (Amplitude vs. Post-stack seismic data is typically a combination of noise
Azimuth) analysis eventually enables quantification of frac- and signal. Broadly speaking, the seismic signal may cor-
ture density and their orientation. respond to continuous or discontinuous reflectors; both may

102 www.firstbreak.org © 2015 EAGE

first break volume 33, March 2015 special topic
Modelling/Interpretation

or may not reflect geology. In our experience, it is found that dip trends of seismic reflectors. This volume has more applica-
conventionally processed seismic data is mostly far less than tions in data enhancement and attribute calculation. A mildly
perfect for seismic interpretation. Hence, mostly a set of data filtered one is informally referred to as a detailed SteeringCube
conditioning techniques are used and some of these (also used because it helps in defining local dips of seismic reflectors. Such
in this paper) are explained below. a volume is mostly used for detailed attribute analysis and is
not recommended for data enhancement.
Frequency filtering
Prior to any step, the first goal is to prepare input seismic data Structural oriented filtering
such that they do not show frequency related noise. This can It is essential to start with seismic data on which the faults and
be achieved during pre-stack or post-stack seismic processing. fracture features can be sharply defined. To obtain such a goal,
A common approach is to apply a frequency (e.g., a band-pass) the seismic data is routinely pre-processed and conditioned.
filter to adjust the bandwidth of the data by minimizing the fre- There are three main steps performed to condition the seis-
quency related noises. Such a filter helps to minimize the noise mic data and produce a final volume for faults and fractures
related to higher frequencies, multiples, or energy dissipation. interpretation.
In other cases, often the higher frequencies are boosted-up 1. The smoothing of seismic reflectors through dip-steered
by applying a specialized technique as explained by Blache- median filtering.
Fraser and Neep (2004); and Kazemeini et al. (2008). Such a 2. The enhancement of fault positions through dip-steered
technique matches the reflectivity spectra obtained from well diffusion filtering.
and seismic data. Only those frequencies are enhanced that lie 3. The merging of the first two steps through a logical state-
within the bandwidth of the seismic data. As a result a con- ment to produce a fault enhanced seismic data.
volution operator is designed that is convolved with the input
seismic data to enhance the frequency contents. This approach These filters are further described below:
often improves the faults definition. Dip-steered Median Filter (DSMF): This is a statistical filter
that is applied on the seismic data using a pre-processed
Dip extraction SteeringCube. It uses median statistics on the seismic ampli-
Dip computation is a crucial part of the workflow as it influ- tudes by following the seismic dips. It results in a smoothed
ences both the post-stack seismic filtering as well as the seismic seismic volume such that the continuity of seismic reflectors
attribute results. Seismic dips can be computed using various is improved by removing the background random noise. It is
methods: amplitude-based, phase-based, amplitude-frequency often considered as an edge preserving smoothing filter if a
based etc. (Tingdahl and de Rooij, 2005; Chopra and Marfurt, fault zone is larger than the median filter size. However, if a
2007). fault zone is smaller than the size of a median filter, the data
These methods operate on post-stack seismic data. Phase- will be smoothed out. Hence, additional filters may be required
based dip computation is used in this paper. The calculation to sharpen the fault zones as described below.
and filtering parameters for the dip estimation are optimized Dip-steered Diffusion Filter (DSDF): Mostly, the seismic
through a series of experiments. The final dip volume is data show diffused character close to a fault zone. If pre-stack
processed in the entire 3D seismic at every sample position. processing is not optimized for the faults, a fault zone may
This volume is used as an input for structural oriented not be sharply defined on the seismic data. To improve the
filtering and dip-steered seismic attributes and is hereinafter sharpness of the faults, one may apply an intermediate filter
referred to as the ‘SteeringCube’. The tool can be filtered either that is referred to as a DSDF. This filter evaluates the quality of
locally or regionally according to the objective of the study and the seismic data in a dip-steered circle. The central amplitude is
data quality. A heavily filtered one is informally referred as a replaced by the amplitude where the quality is deemed best. In
background SteeringCube because it helps in defining overall the vicinity of a fault, the effect is such that good quality seismic

Figure 2 Inline from F3 block of the Dutch sector of North Sea showing (a) raw seismic data, (b) dip-steered median filtered (DSMF) seismic data and (c) seismic
data after application of fault enhancement filter (FEF).

© 2015 EAGE www.firstbreak.org 103

 special topic first break volume 33, March 2015

Modelling/Interpretation

Figure 3 Examples of various conventional fracture attributes are shown using the F3 dataset from North Sea: (a) Base Tertiary horizon with seismic time (ms)
contours, (b) similarity computed using raw seismic, (c) similarity computed using FEF seismic data which shows sharp fault patterns. (d) Polar dip, (e) apparent
dip along 45 degree azimuthal direction and (f) azimuth attributes are similar to the similarity attribute with an added dip/azimuth information. The curvature
attributes such as (g) maximum (h) most positive and (i) most negative curvatures show the sense of fault throw (upthrown and downthrown sides). The maxi-
mum curvature attribute has an edge over others and it shows simultaneously both directions with different colours (signs). In our experience, the attributes
such as ‘c’ and ‘g’ are superior for fracture detection and candidate inputs for unconventional fracture attributes.

data is moved from either side of the fault in the direction of when the waveforms are offsets because of vertical offsets
the fault plane, hence the fault becomes sharper. due to fault throws. This attribute normally shows correlation
Fault Enhancement Filter (FEF): This filter is produced by strength between a number of waveforms in three dimensions.
applying a logical expression on the earlier produced volumes If the strength is 100%, it suggests that the traces are quite
(DSMF and DSDF). It also requires a seismic attribute that similar in their response. If the strength is below 60%, then
highlights discontinuities e.g., similarity (recommended) or the traces do not correlate properly, suggesting dissimilar
curvature. Based on a cut-off value, a logical expression is waveforms, discontinuities or in other words faults and
prepared such that shows reflectors and faults in one seismic fractures. In our presented workflow, this attribute is also used
volume. For instance, if the similarity attribute is used, the as a defining criteria to differentiate faults from reflectors (as
values below a cut-off may correspond to discontinuities and explained in FEF). Furthermore, it is notable that similarity
above a cut-off may correspond to continuities. The resultant attribute calculations made from fault enhanced seismic are
fault enhanced seismic data is thereafter used as an input better than the raw seismic.
for seismic attributes. The attribute mostly shows sharper
definition of faults and hence improves the visualization and Dip-Azimuths:
interpretation. The seismic dip and azimuth attributes natively represent a
Figure 2 shows the comparisons between raw seismic, SteeringCube.
DSMF seismic and FEF seismic on an example inline from the Polar Dip: This attribute converts extracted inline and
F3 block of the Dutch sector of the North Sea. This F3 dataset crossline dips to a true geological dip. The polar dip is the
is used throughout the paper to create various example figures. square root of the sum of (inline dip)2 and (crossline dip)2 .
The polar dip is thus larger or equal to zero. Dips are given
Dip steered attributes in μseconds/metres in time surveys, since they are a ratio
between a vertical length and a horizontal distance. Polar dip
Conventional fracture attributes is a valuable attribute to explain how dipping is a fault plane
when transformed in degrees using a seismic velocity volume.
Similarity and its equivalents: Azimuth: This attribute returns the azimuth of the dip
The lateral correlation between waveforms along a given direction of seismic reflectors. It is often represented in
reflection can be measured by computing the euclidean degrees ranging from -180 to +180. Azimuth is typically
distance between the amplitude vectors representing the wave- defined relative to the seismic survey geometry/orientation.
form. This operator is often referred to as a similarity Positive azimuth is defined from the inline in the direction of
measurement. Its computation provides a direct measurement increasing crossline numbers. Azimuth = 0 indicates that the
of lateral discontinuities in the waveform. The attribute is seismic reflector at evaluation point is dipping in the direc-
very sensitive to phase changes, which makes it very useful tion of increasing cross-line numbers. Azimuth = 90 indicates

104 www.firstbreak.org © 2015 EAGE

first break volume 33, March 2015 special topic
Modelling/Interpretation

that the seismic reflector at evaluation point, is dipping in Most Negative Curvature: This attribute returns the most
the direction of increasing in-line numbers. Therefore, this negative curvature at various points along the interpreted
attribute becomes useful to explain the orientation of fault horizon, from the infinite number of Normal Curvatures
networks. It can also be used to fine tune a volume by clip- that exist. Similar to Most Positive Curvature, this attribute
ping the fault network to a specific azimuth range when used also reveals faulting and lineaments. Again, the magnitude of
with logical expressions. the lineaments is preserved but the shape information is lost.
This attribute can also be compared to the first derivative-
Curvatures: based attributes (e.g. dip, edge, and azimuth). It primarily
Curvature analysis has been used for many years in the oil amplifies synclines and downthrown side of faults.
industry to detect geological features from the shape of the
seismic reflections. Mathematical calculation of curvature
is usually done by the least-square fitting of a quadratic
surface: Maximum Curvature: From the infinite number of Normal
Curvatures there exists one curve, which defines the largest
absolute curvature of the surface. This is called the maximum
curvature. The plane in which maximum curvature is calcu-
A large variety of curvature attributes are described in the lated is orthogonal to the plane of the Minimum Curvature.
literature, e.g. Gaussian curvature, Mean curvature etc. This attribute is computed at various points along the
(Roberts, 2001). However, the three most frequently used interpreted horizon and is very effective at delineating faults
attributes for fracture detection are described here. and fault geometries. Maximum curvature is derived from
Most Positive Curvature: This attribute returns the most Gaussian and mean curvatures. Figure 3 shows the examples
positive curvature at various points along the interpreted of aforementioned attributes.
horizon, from the infinite number of normal curvatures
that exist. The attribute reveals faulting and lineaments. Unconventional fracture attributes
The magnitude of the lineaments is preserved but the Fault Likelihood and Sharpness: Hale (2013) defines the
shape information is lost. This attribute can be compared fault likelihood as a power (for instance n= 8) of semblance
to the first derivative-based attributes (e.g., dip, edge, and (1–Semblancen). However, this results into faults that have
azimuth). It primarily amplifies anticlines and upthrow side non-geological dips and strikes. Such information is obtained
of faults. when the result is scanned for a given range of positive and
negative dips to maximize the fault likelihood. This produces
faults with true dips and strikes. Further filtering is performed
to highlight only the local maxima within the fault likelihood
volume to achieve a sharper fault plane. Fault likelihood has
a value range from 0 to 1. We believe that Hale’s contribution
for fault extraction is a great attempt towards unconventional
fracture attributes. We illustrate one example as a starting
point towards fracture characterization (Figure 4).
Fracture Proximity: This attribute improves the visualiza-
tion of potential fracture anomalies by revealing the con-
nectivity of fracture networks, as well as the distance between
them. It computes the lateral distance (i.e., along a Z-slice)
of any trace in consideration from a trace location classified
as a fracture. Whether a particular trace can be defined as
being part of a potential fracture anomaly, is determined by
a user-specified threshold, on various discontinuity attributes
such as similarity or curvature. This attribute generally is a
measure of a distance of hundreds of metres. For interpreta-
tion it is vital to understand the proximity to (and between)
Figure 4 A comparison of (a) similarity attribute computed using the FEF fracture zones; the lower attribute values suggest closeness to
seismic against the (b) fault likelihood attribute. Clearly the fault likeli- the centre of the fracture zone.
hood attribute results in sharper and correctly outlined faults identification. Fracture Density: This attribute is useful in pinpointing
Furthermore, note that the dip of the fault plane is also adequately captured
by the fault likelihood attribute when compared with the similarity attribute locations of maximum number of fractures within a given
result. radius. This radius can theoretically be linked to a fracking

© 2015 EAGE www.firstbreak.org 105

 special topic first break volume 33, March 2015

Modelling/Interpretation

Figure 5 Fracture proximity computed by defining a fracture threshold of 0.2 on fault likelihood attribute in (a), 0.4 in (b) and 0.6 in (c). In addition to these
threshold values, a radius of 250 m is used to calculate fracture density in (d), (e) and (f) respectively. Using the higher threshold of 0.6 results in the identifica-
tion of the strongest fracture patterns while supressing the stratigraphic expressions (arrow).

radius for drilling a fractured reservoir. It computes the ratio Directional Attributes: The majority of multi-trace
of a ‘number of seismic traces classified as being fractures’ seismic attributes can be treated through a simple equa-
to the ‘total number of traces present’ in a given radius. This tion to obtain a (pseudo)-response in a given azimuthal
attribute directly highlights the regions with higher fracture direction. For such purposes, two gradients are pre-defined.
density as sweet-spots for drilling. For other cases, such as One along the inline direction and the other one along
geothermal energy or hydrocarbon traps, this attribute high- the crossline direction. Once these gradients are extracted,
lights the regions of higher risk or leakage for drilling. Figure 5 we can use the following formula to derive directional
presents examples of fracture proximity and density attributes attributes:
calculated using different thresholds on fault likelihood.
Fracture Gradient: It is another new dip-steered seismic BΦ = Attribute GradientILcos(Φ) + Attribute
attribute and is defined as a spatial derivative of fault cube GradientXLsin(Φ),
using a SteeringCube. It is computed along all azimuths Φ= azimuth angle (referenced from true geographic north).
(i.e., 0 to 360 degrees) and the result is outputted along the
direction where it is maximum. It requires a discontinuity This technique provides new insights, as directional varia-
attribute e.g., fault cube: dip-steered similarity using FEF tions in certain attributes, could in theory, be linked with
data or max curvature as an input. The main advantage of HTI anisotropy caused by fracture networks. An example
this attribute is that a fault/fracture becomes prominent at of this attribute is presented in Figure 7.
the centre of the attribute response.
Fracture Azimuth: It is another dip-steered attribute Voxel connectivity filter (VCF)
and outputs the azimuth along the maximum value of a Fault networks and fracture zones need to be visualized in
fracture gradient. It is a supplementary result to understand three dimensions as geobodies. To understand their con-
the fracture gradient and its orientation. Figure 6 shows nectivity in 3D, one has to apply some advanced filters to
fracture gradient and azimuth attributes applied on similarity produce bodies of faults and fractures. Voxel connectivity
from FEF seismic, maximum curvature and fault likelihood filter is one such filter. It produces continuous geo-bodies
cubes. based on a fault cube. A ‘voxel’ is defined as the small 3D

106 www.firstbreak.org © 2015 EAGE

first break volume 33, March 2015 special topic
Modelling/Interpretation

Figure 6 Fracture gradient applied on (a) similarity from FEF seismic, (b) maximum curvature and (c) fault likelihood attributes. Corresponding fracture azimuth
results are shown in (d), (e) and (f) respectively. In this example, it is evident that the fracture gradient and azimuth attributes from similarity and curvature are
noisier; however, the fault likelihood attribute produces reasonable results.

Figure 7 A comparison of directional (amplitude) attributes: (a) along 135 degrees azimuth highlighting stratigraphic features, while (b) along 45 degrees
azimuth primarily amplifies fault lineaments.

volume corresponding to one seismic sample, and is thus body. The neighbours are searched on the basis of all
linked to the bin size and sampling rate of a seismic survey. common faces, or all common faces + edges, or all com-
This filter works as follows: mon points. This can be understood by imagining the
1. Use the voxel value to make it binary (true or false) voxel in question, at the centre of a 3x3x3 Rubik’s cube
based on criteria such as cut-off/threshold. (Figure 8e). If all the 26 cubes (i.e., voxels) around the
2. The neighbourhood around each voxel with a ‘true’ central cube are having the ‘true’ value, it corresponds
value is searched for other voxels with ‘true’ values. If to all common points. If only those cubes that share faces
such voxels are found, the voxels are joined into one with the central cube have ‘true’ value, it corresponds to

© 2015 EAGE www.firstbreak.org 107

 special topic first break volume 33, March 2015

Modelling/Interpretation

Figure 8 Voxel connectivity filter applied on (a) similarity attribute from FEF seismic, (b) fault likelihood (0-1) attribute with low connectivity threshold of 0.2
covering more faults, (c) fault likelihood (0-1) attribute with low connectivity threshold of 0.6 covering only the biggest faults and (d) fracture density attribute
pinpointing the ranked fractured zones with highest density, e.g. ‘1’ and ‘2’. Rubik’s cube is shown in (e) to understand the neighbourhood searching criteria
of the voxel connectivity filter.

common faces. Finally, if in addition to the faces, the cubes Seismic data conditioning is a vital step of the workflow
that share edges with the central cube are having a ‘true’ as it aims to clean the data for all unwanted elements and
value, it corresponds to all common faces + edges. focuses on enhancing faults. The fracture attributes – which
3. Store the connected geo-bodies. At this stage, a seismic are applied on conventional seismic attributes derived from
attribute volume is written out where the values inside the dip-steered and fault enhanced datasets – further enhance the
connected bodies can be either: strength, pattern and orientation of fractures.
a. The attribute value itself, We identify two promising attributes that can directly
b. The body’s size, i.e., number of voxels forming the characterize the fracture proximity and their density in a
body (often considered as an important result in faults/ given area. These attributes lead to the identification of
fracture studies), potential sweet spots. However, fracture gradient/azimuth
c. A unique ranking of bodies based on their size. attributes are found less useful because of their noisier
appearance when applied on seismic data. Nevertheless, the
In this paper, the VCF is applied on various fault attribute directional attributes can alternatively be used for analysing
volumes: fault likelihood, similarity from FEF seismic and fracture orientation.
fracture density volume derived from the fault likelihood The remaining part of the workflow uses the power of
attribute (Figure 8). Clearly, the extracted geobodies from visualization to understand fault networks and their connec-
the fault likelihood attribute provide crucial insights into the tivity in a three-dimensional space using the voxel connectiv-
3D fracture patterns and their connectivity. In addition, the ity filter. This produces a ranked fracture volume that can be
ranked (see 3c above) geobody extracted from the fracture used for hydrocarbon exploration, risk assessments, drilling
density attribute help visualize the best zones for hydrocar- through unconventional plays and basement fracture zones.
bon exploration in 3D.
Acknowledgements
Conclusions We are grateful to the Colorado School of Mines for making
This paper presents the popular seismic attributes that are their source code available for fault likelihood attribute; dGB
routinely applied to perform visualization and interpretation Earth Sciences for providing resources; and the government
for faults and fracture detection. These attributes become of the Netherlands for publicly releasing the F3 dataset that
meaningful and useful for interpretation when they are can be downloaded from www.opendtect.org/osr. We also
further processed for fracture detection using the proposed thank Dr Riaz Alai for his exceptional help and fruitful dis-
unconventional fracture attributes. cussions on fracture characterization in basement reservoirs.

108 www.firstbreak.org © 2015 EAGE

 first break volume 33, March 2015 special topic
Modelling/Interpretation

References Chopra, S. and Marfurt, K.J. [2007] Seismic Attributes for Prospect
Alai, R., Aqrawi, A.A., Mohamed, A.B. and Taha, M.T.A. [2014] Identification and Reservoir Characterization. Geophysical
Fracture characterization in basement reservoirs through seismic Developments Series, SEG, 518.
attributes. First Break, 32 (10), 83–92. Hale, D. [2013] Methods to compute fault images, extract fault surfaces
Bale, R., Gratacos, B., Mattocks, B., Roche, S., Poplavskii, K. and Li, and estimate fault throws from 3D seismic images. Geophysics, 78
X. [2009] Shear wave splitting applications for fracture analysis (2), 33-43, DOI: 10.1190/GEO2012-0331.1.
and improved imaging: some onshore examples. First Break, 27 Kazemeini, S., Fomel, S.,and Juhlin, C. [2008] Prestack spectral blueing:
(9), 73–83. A tool for increasing seismic resolution. 76th Annual International
Ben-Brahim, L., Varley. S., Michaud, O. , Buffet, P., Hollingworth, Meeting, SEG, Expanded Abstracts, 854–858.
S., Woodcock, A. and Purcell, C. [2009] Application of beam Rueger, A. [1997] P-wave reflection coefficients for transversely
migration techniques to the evaluation of an exploration prospect isotropic models with vertical and horizontal axis of symmetry.
(Balvenie): A case history from the UKCS Northern North sea. 71st Geophysics, 62, 713–722.
EAGE Conference & Exhibition, Extended Abstracts. Roberts, A. [2001] Curvature attributes and their application to 3D
Blache-Fraser, G. and Neep, J. [2004] Increasing seismic resolution interpreted horizons. First Break, 19 (2), 85–100, doi:10.1046/
using spectral blueing and colored inversion: Cannonball field, j.0263-5046.2001.00142.x.
Trinidad. 72nd Annual International Meeting, SEG, Expanded Sengupta, M. and Bachrach, R. [2006] High-resolution fracture
Abstracts, 1794–1797. characterization from azimuthal P-wave seismic surveys using rock
Brouwer, F. and Huck, A. [2011] An Integrated Workflow to Optimize physics model-based Bayesian inversion – Pinedale field, Wyoming.
Discontinuity Attributes for the Imaging of Faults. 31st Annual 76th Annual International Meeting, SEG, Expanded Abstracts,
GCSSEPM Foundation Bob F. Perkins Research Conference, 1928–1931.
496–533. Tingdahl, K. and De Rooij, M. [2005] Semi-automated Detection of
Chopra, S. and Marfurt, K. J.[ 2007] Preconditioning seismic data with Faults in 3D Seismic Data. Geophysical Prospecting, 553–542.
5D interpolation for computing geometric attributes. The Leading Trad, D. [2009] Five-dimensional interpolation: Recovering from
Edge, 32 (12), 1456–1460. acquisition constraints. Geophysics, 74 (6), 123–132.

Second EAGE Workshop on

Geomechanics and Energy
The Ground as Energy Source and Storage

13-15 October 2015 – Celle, Germany

Topics
1 Hydrocarbon Exploration 2 Underground Storage
and Production • Nuclear Waste Disposal
• Naturally and Hydraulically • CO2 Sequestration
Fractured Reservoirs • Gas Underground Storage
• Wellbore Stability
• Unconventional Energy Sources 3 Geothermal Energy
(Shale Gas, Methane Hydrate) • Shallow and Deep Geothermal
• Subsidence Energy
• Fault Reactivation • Heat Exchanger Geostructures

www.eage.org
Call for papers deadline:
10 June 2015 ud
ed
l
I nc

GEW15-V2H.indd 1 06-02-15 08:36

© 2015 EAGE www.firstbreak.org 109

View publication stats