IOP PUBLISHING JOURNAL OF GEOPHYSICS AND ENGINEERING

J. Geophys. Eng. 8 (2011) 123–141 doi:10.1088/1742-2132/8/2/001

Seismic reservoir characterization: how

can multicomponent data help?

Downloaded from [Link] by National Geophysical Research Institute user on 05 September 2023
Xiang-Yang Li1,2 and Yong-Gang Zhang3
1
British Geological Survey, Murchison House, West Mains Road, Edinburgh EH9 3LA, UK
2
CNPC Geophysical Key Laboratory, China University of Petroleum-Beijing, No 18, Fuxue Road,
Changping, Beijing 100049, People’s Republic of China
3
Science and Technology Institute, China Petroleum and Chemical Corporation, Beijing 100086,
People’s Republic of China

Received 27 June 2010

Accepted for publication 21 December 2010
Published 10 March 2011
Online at [Link]/JGE/8/123

Abstract
This paper discusses the concepts of multicomponent seismology and how it can be applied to
characterize hydrocarbon reservoirs, illustrated using a 3D three-component real-data example
from southwest China. Hydrocarbon reservoirs formed from subtle lithological changes, such
as stratigraphic traps, may be delineated from changes in P- and S-wave velocities and
impedances, whilst hydrocarbon reservoirs containing aligned fractures are anisotropic.
Examination of the resultant split shear waves can give us a better definition of their internal
structures. Furthermore, frequency-dependent variations in seismic attributes derived from
multicomponent data can provide us with vital information about fluid type and distribution.
Current practice and various examples have demonstrated the undoubted potential of
multicomponent seismic in reservoir characterization. Despite all this, there are still
substantial challenges ahead. In particular, the improvement and interpretation of
converted-wave imaging are major hurdles that need to be overcome before multicomponent
seismic becomes a mainstream technology.
Keywords: multicomponent seismic, reservoir characterizations, anisotropy, shear-wave
splitting, frequency-dependent attribute

Introduction a complementary approach and can provide added values to

these circumstances.
There is an ever-increasing drive within the hydrocarbons Multicomponent seismic is a term to describe the seismic
industry to maximize extraction from existing oilfields, as the method for recording and analysing multiple modes of
finding of new oilfields has become difficult and expensive. the particle motion of a seismic wave, such as pressure
Few existing fields are structurally simple, and most consist and displacement measured in terms of either velocity or
of complex formations in which many heterogeneities exist, acceleration. Whilst pressure is a scalar, displacement can
of which only a small proportion are reservoirs. Thus a more be a vector consisting of vertical and horizontal components.
complete understanding of the distribution and geometry of To record all these modes requires more than one component
the reservoir heterogeneity is the key to sustained recovery of geophone sensors. Hence, the term ‘multicomponent’
and growth of reserves. The conventional approach to this (Tatham and McCormack 1991, Sheriff 2006). Three-
problem relies on the application of high definition 3D P-wave component (3C) recording uses three orthogonal geophones:
seismic, which has been widely accepted by the hydrocarbon one vertical and two horizontals; up to nine-component
industry. However, for matured oilfields, where there is a need (9C) data can be recorded using three orthogonally polarized
to understand the fluid pathways characterized by small-scale sources and three orthogonal receivers (z, x and y, as shown in
heterogeneities, such as cracks and fractures, or to distinguish figure 1(a)). Compared with the single-component P-
fluid type and distribution for optimizing in-fill drilling during wave data, it is more critical to understand how these new
production and development, multicomponent seismic offers components are related to the vector wave field during both

1742-2132/11/020123+19$33.00 © 2011 Nanjing Geophysical Research Institute Printed in the UK 123

X-Y Li and Y-G Zhang

Receiver r

Source
t z

Downloaded from [Link] by National Geophysical Research Institute user on 05 September 2023
y
z

Survey line
(a) (b)

Figure 1. (a) Data matrix for nine-component recording geometry. There are three orthogonal sources (X, Y and Z) represented by the
columns, and three orthogonal receivers (x, y and z) by the rows. The shaded area represents the conventional three-component
converted-wave geometry studied in this paper. (b) The acquisition coordinate system, x, y and z, and the processing coordinate system,
r, t, z.

the acquisition and processing stages, and this requires a until the mid-1980s, when the main drive was to improve
clear definition of the coordinate system used. Following seismic resolution using shear waves (Anno 1987, Tatham
Gaiser (1999), we adopt the acquisition coordinate system and McCormack 1991) due to their lower velocity, hence
as specified by the Cartesian coordinates x, y and z, where potentially shorter wavelength than the conventional P-wave.
x is the inline direction, y is the crossline direction and z is However, this did not materialize due to the lower frequency
the vertical direction, forming a right-hand coordinate system content and poorer data quality than P-waves. The third
(figure 1(b)); we also adopt the source centred cylindrical phase occurred in the mid-1990s due to the advent of marine
system during the processing stage specified by r, t and z, 4C technology, and the main drive was to image beneath
where r is the radial direction point away from the source, t gas clouds using converted waves (C-waves). This proved
is the transverse direction and z is still the vertical direction to be very successful, and heralded more than a decade of
(figure 1(b)). research and development into multicomponent technology,
Starting in the 1960s with the development of shear wave which has led into the current phase, starting in the new
sources, multicomponent seismic has been used to distinguish millennium. Owing to the development of MEMS (micro-
lithologic and subtle diagenetic changes in the subsurface for electro-mechanical system) digital sensors, it is now feasible
characterizing stratigraphic traps. Multicomponent seismic to economically record multicomponent seismic data in both
has also been commonly used to study seismic anisotropy, land and marine (sea bottom) settings.
which allows us not only to produce better imaging and This paper will summarize and discuss the potential
better definition of the reservoir, but also to determine the applications of multicomponent seismic data, focusing on
directional characteristics of the reservoir heterogeneities, the study of seismic anisotropy for better imaging, improved
such as aggregate alignments and cluster distribution. One fracture (heterogeneity) characterization and better fluid
major application in this respect is the study of fractured prediction. 3D3C multicomponent data from southwest China
reservoirs. will be presented for this purpose.
In recent years, multicomponent seismic has been
used to detect reservoir fluid directly, the ‘holy grail’ of
exploration geophysics. Using single component data, we are Reservoir heterogeneities at different scales
unable to distinguish between low- and high-saturation gas
reservoirs, so-called ‘fizz gas’, and we are also prevented from Many methods have been used to study the structure of
distinguishing between oil and water-saturated reservoirs. reservoirs, with the combination of 3D seismics with well
However, we are now able to tackle these problems using logs possibly being the most popular so far. However, seismic
multicomponent data (Zhu et al 2000, Davis 2007, Qian et al data often lack enough detail to accurately represent some
2007). Furthermore, multicomponent time-lapse (4D) seismic significant reservoir heterogeneities, and higher resolution
offers great potential in monitoring reservoir changes during is indispensable in gaining better definition of small-scale
production for improving recovery efficiency and finding by- features.
passed oil (Angerer et al 2002). It is possible to gain some insights into the distribution
The development of the multicomponent technology in of reservoir heterogeneities utilizing the concept of scattering
seismic exploration has gone through four phases. The first from sub-wavelength heterogeneities, which sufficiently alter
phase started in the late 1960s, when shear-wave source the transmitted wave field to carry forward an impression of the
reflection data were collected with vertical seismic profiles group characteristics. The seismic effects of the scattering can
(VSPs) to assist interpretation (Cherry and Waters 1968, be simulated by an ‘equivalent homogeneous medium’ which
Erickson et al 1968), and this continued to the second phase replaces the heterogeneous medium and acts as a seismic

124
 Seismic reservoir characterization: how can multicomponent data help?

Downloaded from [Link] by National Geophysical Research Institute user on 05 September 2023
(a) Reservoir pattern (b) Stratigraphy

Figure 2. (a) A typical geological section showing the gas reservoir pattern, and (b) the stratigraphic column, where T3x2 marks the target
formation.

(a) Well X565 (b) Well X853

Figure 3. Core samples in wells from the study area, revealing heavy fracturing after Tang et al (2008).

pseudo-function for reservoir complexity. Using this concept of the type of data used, it is not sufficient to describe reservoirs
gives us an opportunity to observe a diversity of reservoir in terms of bedding planes, fault distributions or depositional
features such as faulting, cross-bedding, channels and facies, systems alone; successful reservoir characterization depends
whilst retaining the original complexity of the wave field. It upon the effective integration of representative data measured
has been particularly successful in estimating the predominant at different scales. This can be accomplished through the
fracture direction and density from both P- and shear-wave application of equivalent medium theory, which is a valuable
seismic data. concept for conserving a detailed level of characterization for
Definitions of the term reservoir heterogeneity cover a reservoir heterogeneity and may also assist in correlating data
wide spectrum and depend very much upon the particular from one scale to another.
geological features under observation. Nevertheless, it is
generally taken to refer to an irregular spatial uniformity of Figure 2 shows a typical geological section from the study
diverse physical constituents, whether structural, lithological, area in southwest China. Reservoir heterogeneities at a range
petrophysical or relating to fluid flow, evident at various scales. of scales can be observed in this section, from lithological
Scales of reservoir heterogeneities can vary from a large interfaces (‘macroscale’), through cross-bedding and ripple
basin (‘megascale’), and the distribution of stratigraphic and laminations (‘mesoscale’), and fractures can be observed from
lithologic intervals (‘macroscale’), through cross-bedding and the core samples at the smallest ‘microscale’ level (figure 3).
ripple laminations (‘mesoscale’) to the smallest ‘microscale’, Multicomponent seismic data have been acquired in this field
which includes grains, pores or pore throats (MacBeth 1995). to characterize the reservoir heterogeneities, and the results
Although heterogeneities still tend to be viewed from the basis will be presented in this paper.

125
X-Y Li and Y-G Zhang

Downloaded from [Link] by National Geophysical Research Institute user on 05 September 2023
(a) Layer-induced anisotropy (b) Fracture-induced anisotropy

Figure 4. Two common causes of seismic anisotropy: (a) layering- and (b) fracturing-induced anisotropy.

Seismic anisotropy subsequent reservoir modelling and development, since the

reservoir is going to be misplaced both vertically and laterally.
Seismic anisotropy refers to the wave phenomenon in which The amount of mispositioning is dependent on the amount
seismic velocity changes with the direction of measurement. of anisotropy and the degree of dip. For horizontal beds,
There are two common causes of seismic anisotropy due to depth error is the main side effect, which can be corrected
reservoir heterogeneity. One is due to the presence of shales or by adjusting the vertical velocity. However, complications
shaly sediments, as well as sequences of thin layers, which give can arise in the presence of both anisotropy and dip. For
rise to transverse isotropy with a vertical axis of symmetry, the minor dips of 15–20◦ , with a mild anisotropy of 5–10%, lateral
so-called vertical transverse isotropy (VTI). Another common misposition can be up to two to three wavelengths, according
type of anisotropy is horizontal transverse isotropy (HTI) to Alkhalifah and Larner (1994). This cannot be corrected
due to vertical fracturing, or other reservoir heterogeneities. simply by adjusting the velocity; anisotropic migration is the
Figure 4 illustrates the geological origins of these types of only solution. For HTI, or azimuthal anisotropy, misfocusing
anisotropy, where the seismic waves will travel faster along is the main effect on imaging, particularly in the azimuthal
the bedding planes as in figure 4(a), than along any other plane. The consequence of misfocusing is blurred images of
direction. In the cases of fracturing in figure 4(b), the waves the subsurface, which in turn led to misinterpreted reservoir
will travel faster along the fracture planes than other directions. structures. Unfortunately, seismic imaging for wide-azimuth
These velocity differences lead to seismic anisotropy. 3D seismic data is a highly challenging problem and is still an
There are also observations of anisotropy with lower area of intensive research.
symmetry than TI (transverse isotropy), such as TTI (TI Early attempts to take account of anisotropy during
seismic imaging focused on elliptical anisotropy, that is an
with a tilted symmetry axis), and orthorhombic symmetry,
elliptical wavefront and slowness surface, as published in
which can be found in the literature. Note that the most
Levin (1978), Byun (1982), Helbig (1983) and VerWest
convincing observations of seismic anisotropy within the
(1989). Elliptical anisotropy turns out to be a very special case
context of exploration seismology are documented in VSPs
of VTI, as pointed out by Thomsen (1986) and Dellinger and
(e.g. MacBeth 2001), and publications of reflection data
Muir (1988). According to Thomsen (1986), four parameters
that emphatically prove the existence of anisotropy are also
can be used to describe the anisotropic wave field in TI media
restricted to shale-induced VTI or TTI (e.g. Ball 1995 and
for P- and SV-wave propagations, which replace the four
Winterstein and De 2001), or fracture-induced HTI (e.g. elastic constants (c1l, c33, c44 and c13). These four parameters
Li 1997, Gaiser et al 2001, Gaiser and Van Dok 2001). are the P- and S-wave vertical velocities, Vp 0 and Vs0, plus
Observations of other types of anisotropy using reflection data two dimensionless anisotropic parameters, ε and δ, where ε is
are rarely reported (Grechka et al 2005). In our study area related to the fractional difference between the P-wave vertical
the formation is relatively flat with gentle dips up to 10◦ , as and horizontal velocities, or the amount of P-wave velocity
shown in figure 1(a). For dips 10◦ or less, the effect of the anisotropy, and δ is related to the fractional difference between
dip is small and cannot be accurately resolved (Alkhalifah and the P-wave vertical velocity and the short-spread moveout
Larner 1994, Khuenel and Li 1996). Therefore, TTI is not velocity, controlling the amount of vertical mispositioning
applicable; VTI anisotropy dominates in the overburden and during seismic imaging. For isotropic media, both ε and
HTI in the target formation. δ become zero, whilst for elliptical anisotropy, ε equals δ.
There are two major applications of the study of seismic With the advent of Thomsen parameters for TI media, the
anisotropy: one is to improve the quality of seismic imaging, focus of anisotropic imaging shifted to VTI. Tsvankin (1996)
and the other is to determine the distribution and geometry of gave a comprehensive summary of the effects of VTI and its
the reservoir heterogeneities. The main effects of anisotropy parameterization. Since then, seismic imaging accounting for
on seismic imaging are misfocusing and mispositioning. VTI was gradually accepted by the hydrocarbon industry.
For VTI, both vertical and lateral mispositioning can be Nowadays, the Thomsen parameters are commonly used
introduced, and this can have catastrophic consequences for for describing seismic wave behaviour in TI media. For

126
 Seismic reservoir characterization: how can multicomponent data help?

for mapping geological structure, two- or three-component

surveys for mapping geological structure and lithology (Anno
1987, Lynn and Thomsen 1990).
For investigating shear-wave splitting, a configuration
of two-by-two shear-wave surveys (xx-, xy-, yx- and yy-
component; the crossed-area, figure 1(a)) has been used in
the past (Alford 1986, Li 1997). Early efforts were focused
on the development of shear-wave sources (e.g. Cherry and

Downloaded from [Link] by National Geophysical Research Institute user on 05 September 2023
Waters 1968, Erickson et al 1968, Won and Clough 1981,
amongst others), and Gaiser and Hilterman (2010) gave a good
account on these developments. The main application of such
surveys is for fractured reservoir characterization. Fractures
lower the overall seismic velocities of both the two shear
waves by different degrees. The two shear waves excited
by multicomponent acquisition thus have different impedance
contrasts. Such surveys may therefore detect fracture clusters
by identifying dim spots in the top reservoir event when
observed on the slower shear wave section. This is useful when
the fracture porosity is too low to significantly alter the P-wave
impedance contrast. This approach has been successfully
applied to locating fracture swarms in the Austin Chalk in
Texas (Mueller 1991, Li 1997) using 2D lines, and guiding
horizontal drilling into productive zones. The approach has
also been employed in large 3D surveys (Lewis et al 1991), and
Figure 5. Shear-wave splitting in fracture-induced HTI media. for a variety of reservoirs with fractured dolomites, sandstones
and coals. In the latter, the high/low impedance contrast now
produces a brightening of the slow shear-wave horizon (Shuck
orthorhombic media, similar Thomsen-type parameters can et al 1996). Despite many successes, these two-by-two shear-
also be defined (Tsvankin 1997). Note that there are five wave source surveys are not shot on a routine basis due to their
Thomsen parameters for VTI, corresponding to the five elastic expensive setup. In addition, especially with land data, the
constants. However, for P and P-SV wave propagation in VTI quality of pure-shear-wave data is often poor.
media, only four parameters are required. Over recent years, it has become popular for both practical
In the study of the effects of HTI and azimuthal anisotropy, and economic reasons to use three- or four-component (3C or
the early effort focused on shear waves (Crampin 1981, 1985, 4C) receivers with a compressional P-wave source. After
Crampin and Lovell 1991). In fact, azimuthal anisotropy decades of experimenting, it turned out that the most effective
arising from fractures (HTI) was once inappropriately coined way to acquire shear wave is to use a P-wave source
shear wave anisotropy. This over-emphasis is principally due (Gaiser and Hilterman 2010). The three-component receiver
to the ease with which shear wave propagation in fractured measures displacement of the ground, usually in two horizontal
media can be recognized, and the simplicity of its resultant directions and one vertical direction (the shaded area in
interpretation. For near vertical propagation, when a shear figure 1(a)). The fourth component is a measurement
wave enters a fracture-induced HTI medium, it necessarily of pressure, which is used in marine sea-bottom surveys.
splits into two: a fast and a slow wave. The fast wave Garotta and Granger (1988) first reported the acquisition and
is polarized along the fracture strike and the slow wave is processing of 3D3C data, and this was followed by several
polarized perpendicular to it. Furthermore, the time delay studies in the late 1990s and early 2000s (e.g. Van Dok
between the fast and slow waves is proportional to the fracture et al 1997, Gaiser 1997, 1999, Gaiser et al 2001 and Gaiser
density (figure 5). This simple and direct approach was and Van Dok 2001). Stewart et al (2002, 2003) reviewed
exploited in the first generation methods of multicomponent these developments with more details. More recently, with
seismic survey, which relied upon direct excitation and the advent of the MEMS sensor, the acquisition cost is
recording of shear waves in land multicomponent acquisitions substantially reduced, and such acquisitions are often shot
(Alford 1986), as discussed in the following section. in a wide azimuthal fashion, utilizing the azimuthal variations
in seismic attributes (Roche et al 2005).
Multicomponent seismic surveys If wide azimuth data are available, azimuthal anisotropy
can be quantified either from P-waves through analysis of the
As shown in figure 1(a), a full nine-component survey is azimuthal variation in P-wave attributes including travel time,
required to describe the vector wave field accurately, but in velocity, and amplitudes (e.g. Mallick et al 1996, Rueger 1996
practice several configurations of sources and receivers have and Jenner 2001, amongst others), or from P-SV converted
been used depending on the purpose of the surveys. These waves through the analysis of shear-wave splitting (e.g. Gaiser
include conventional one-component P-wave seismic surveys 1997 and Li 1998). Azimuthal anisotropy may be used

127
X-Y Li and Y-G Zhang

to invert for fracture parameters during fractured reservoir

characterization. Estimation of the fracture-induced azimuthal
anisotropy can determine the aggregate alignment and porosity
of the fracture ensemble if the underlying fracture model is
known. Note that one needs to make careful calibration for
using the results derived from azimuthal anisotropy, because
this approach is not generally applicable, and the relationship
between the stress field and the fracture model requires

Downloaded from [Link] by National Geophysical Research Institute user on 05 September 2023
calibration and verification.

Information content of multicomponent surveys

Figure 6. Vp /Vs ratio and lithology.
Depending on the usage, there are several types of information
that can be extracted from the multicomponent data. The
commonly used attributes are: (a) P- and S-wave reflection
the P-wave behaviour in VTI media. Similarly, the SV-
moveout signatures for anisotropic imaging and parameter
reflection moveout is controlled by the short-spread SV-
estimation; (b) P- and S-wave velocity ratio (Vp /Vs ratio) and
wave NMO velocity Vs 2 and the anisotropic parameter ζ .
their corresponding P- and S-wave impedance for lithological
Another parameter σ was introduced to describe the fractional
identification; (c) azimuthal P-wave attributes and shear-
difference between the SV vertical and moveout velocities, and
wave splitting for characterizing reservoir fractures and
σ controls the depth conversion of the SV-wave. Therefore,
heterogeneities; and (d) frequency-dependent P- and S-wave
three parameters, Vs 2, ζ and σ , are sufficient to describe the
attributes for fluid prediction.
S-wave behaviour in VTI media. The following equation
summarizes the relationships among these parameters:
Reflection moveout for anisotropic imaging and parameter
ε−δ
estimation VP22 = VP20 (1 + 2δ), η =
1 + 2δ
ε−δ (1)
Anisotropic imaging and parameter estimation have become
VS2 = VS0 (1 + 2σ ), ζ =
2 2
.
increasingly popular in exploration geophysics. Ignoring 1 + 2σ
anisotropy during seismic imaging may obscure subsurface For PS-converted waves (C-waves), four parameters are
structures and in some cases lead to serious errors in required to describe the converted-wave reflection moveout
locating hydrocarbon prospects. These subjects have also (appendix A). These four parameters are Vc2, γ 0, γ eff and χ ,
been intensively researched and there are a vast amount of where Vc2 is the short-spread converted-wave NMO velocity,
publications in the literature (e.g. Tsvankin 1996, Thomsen γ 0 and γ eff are the vertical and effective velocity ratios (see
1999, Gaiser et al 2001 and Li and Yuan 2003). Generally appendices A and B), and χ is the converted-wave anisotropic
speaking, modelling and migration algorithms are relatively parameter which is defined as a combination of ε and ζ , χ =
well understood, whilst parameter estimation and model γ 0γ 2eff η−ζ . These four parameters (Vc2, γ 0, γ eff and χ ) may
building are less well defined and require further research. be referred to as the converted-wave stacking velocity model.
Regardless of the symmetry of the anisotropic class, VTI However, for converted-wave time migration in VTI media, it
symmetry is the starting point for parameter estimation, and is more convenient to use the parameters Vp 2, Vs 2, η and ζ ,
reflection moveout signatures in VTI media are the main source which are sometimes referred to as the converted-wave time
of information, as recorded by multicomponent surveys, for migration velocity model. The migration parameters can be
anisotropic parameter estimation and model building. calculated from the stacking velocity parameters (appendix B).
Traditionally, P-wave data are the driving force for To sum up, the ability to relate between different
structural imaging. However, in areas with gas clouds or parameters for different wave behaviours is the major
weak P-wave impedance, PS converted waves may be used advantage of using the Thomsen parameters, and different
to improve the reservoir imaging, as demonstrated in Gaiser wave types have different sensitivities to the presence
et al (2001). Recent years have seen an increase in the use of anisotropy, which makes the estimation of anisotropy
of P-SV converted waves driven by the economic use of the parameters realistic and practical (Li et al 2007).
P-wave source to obtain S-wave information. Therefore, here
we will review the use of P- and PS converted-wave reflection
Vp /Vs ratio and impedance analysis
moveout for estimating VTI parameters.
As shown in appendix A, for time processing, the P- These are the commonly used attributes derived from
wave reflection moveout is controlled only by two parameters, multicomponent seismic data. Variations in P and S-wave
the short-spread P-wave NMO velocity Vp 2 and anisotropic velocity ratio (Vp /Vs ratio) are diagnostic of lithological
parameter η. Anisotropic parameter δ controls the error in changes in the subsurface (Tatham 1982). Typically, different
P-wave depth conversion since it is related to the fractional lithologies may give rise to different velocity ratios, as shown
difference between the P-wave vertical and moveout velocities. in figure 6. Hydrocarbon bearing rocks are often characterized
Thus, three parameters, Vp 2, η and δ, are sufficient to describe by a low velocity ratio.

128
 Seismic reservoir characterization: how can multicomponent data help?

The average velocity ratio may be determined directly P-wave attributes for fracture characterization
by correlating the P- and S-wave data as discussed by Anno
Assuming that the fracture population consists of
(11987). Gaiser (1996) introduced a robust technique for
predominantly one major orientation, the azimuthal variation
determining long-wavelength Vp /Vs ratios by correlating the
of P-wave seismic attributes, such as travel time, stacking
P- and PS converted-wave data. Fomel and Backus (2003)
velocity, reflected wave amplitudes, impedance, etc can be
applied an automatic data registration (warping) algorithm to approximately described by an ellipse (Mallick et al 1996,
correlate the P- and PS-wave migrated data and obtained the Rueger 1996, MacBeth and Li 1999). The long axis of the
Vp /Vs ratios. ellipse indicates the fracture orientation, and the relative ratio

Downloaded from [Link] by National Geophysical Research Institute user on 05 September 2023
Originating from the velocity ratio, P- and S-wave of the long to short axes of this ellipse is proportional to the
impedance attributes are used more often, and these can be fracture density of the rock concerned. At least three data
derived from inversion techniques. It is possible to obtain P- points are required to define an ellipse in azimuthal planes.
and S-wave impedance attributes from P-wave data alone using Thus fracture orientation and intensity maps can be built from
pre-stack AVO inversion. However, the inversion problem can 3D P-wave data if there is sufficient azimuthal coverage. We
be better constrained using multicomponent data, and there shall call this technique azimuthal attribute analysis or the
are numerous examples in the literature (e.g. Blott et al 1999, 3A technique. This technique was also widely studied in the
Garotta et al 2002 and Ikwuakor 2006). literature, e.g. Corrigan et al (1996), Hall et al (2000) and
One common AVO inversion technique is the weighted- Lynn (2007), amongst others.
stacking method (Smith and Gidlow 1987, Stewart 1990). The In the practical application of the 3A technique,
method works by least-squares inversion of Aki–Richards’ two methods are often employed to extract the fracture
approximations (Aki and Richards 1980) to recover P- information: full-azimuth surface fitting and narrow-azimuth
and S-impedance reflectivities ( Ip /Ip , Is /Is ) and pseudo stacking. The first method fits an elliptical surface to the
Poisson’s ratio reflectivity ( σ /σ ). The PP and PS-reflection data from all available azimuths and offsets by a least-squares
coefficients can be reformulated as fitting technique. The second method divides the data into a
number of narrow-azimuth volumes. Corresponding to these
1 + tan2 θ Ip β2 Is two methods, there are mainly four seismic attributes which
Rpp(θ ) ≈ − 4 2 sin2 θ
2 Ip α Is may be used to extract the fracture information, including
 2  velocity, travel times/interval travel times, amplitude, and
1 β ρ
− tan2 θ − 2 2 sin2 θ , (2) AVO (amplitude versus offset) gradient. The surface fitting
2 α ρ
method is applicable to the amplitude and travel time attributes,
and whilst the narrow-azimuth stacking method is applicable to the
  velocity and AVO gradient attributes. Past experiences reveal
α tan ϕ 2β
Rps(θ, ϕ) ≈ − 1 + 2 sin ϕ −
2
cos θ cos ϕ that during data analysis one should avoid the commonly used
2β α
   narrow-azimuthal stacking technique, which may enhance
ρ 4β Is the acquisition footprint. Instead, surface-fitting across all
× − 4 sin ϕ −
2
cos θ cos ϕ ,
ρ α Is azimuths and offsets is recommended.
(3)
Shear-wave splitting
where α, β and ρ are the average P- and S-wave velocities Vp ,
Vs and density across the interface, Ip and Is are the average P- It has been seen that shear-wave splitting is the most diagnostic
I
and S-wave impedances across the interface, Ipp and IIs s are, effect of wave propagation in an anisotropic medium, and
respectively, the relative P- and S-impedance contrasts across information about reservoir structure can be gathered by
the interface, and ρρ is the relative density contrast. θ is an observation of a number of shear-wave attributes (Martin and
average of the P-wave angles of incidence and transmission Davis 1987, Garotta and Granger 1988, Lynn and Thomsen
across the interface, and ϕ is the average of the shear wave 1990, Lynn et al 1995, Li 1997, Gaiser 1997, 1999). Of
angles of reflection and transmission across the interface. The these, the orientation and polarization of the first arrival,
the time delay between the two split shear waves, and the
above equations are accurate up to angles of incidence of
differential reflectivity at normal incidence are the three most
around 50◦ for typical velocity and density contrasts (Smith
used attributes.
and Gidlow 1987).
Fracture strike and the probable direction of fluid flow
We assume that the PP- and PS reflectivities are known within a reservoir can be inferred from the polarization
from multicomponent seismic data, and that the density direction of the fast split shear wave. The time delay
follows Gardner’s relation (ρ = 0.23V 1/4 ). For a given between the two shear waves can give information about
I
velocity model, there are only two unknowns, Ipp and IIs s , in fracture and crack density, while differential reflectivity can
equations (2) and (3), which can then be solved help in identifying more intensely fractured zones in the
deterministically from the two equations. This process is often seismic section, and is thus a very useful indicator for fracture
referred to as joint PP- and PS-wave AVO inversion. The delineation.
inverted attributes can then be used directly for lithological Note that some complications may occur when analysing
interpretation (Zhang and Li 2005). shear-wave splitting. For instance, on entering an overburden

129
X-Y Li and Y-G Zhang

with a different fracture orientation, both the fast and slow frequency is 10 Hz, crack density is 0.1, and time-scale value
components may split again giving four shear waves with 2 × 10−5 s. There is a decrease in P-wave velocity when gas
new polarizations and velocities. This means that the fracture replaces water, which is often referred to as the Gassmann
orientation of the reservoir is concealed by the anisotropic effect. P-wave velocity with gas saturation shows a larger
effects of the overburden, which can be a major problem in variation from high to low frequency and higher attenuation
the use of polarization information. For a thin or weakly than the water saturation case, which proves that gas saturation
anisotropic reservoir, the fast and slow split shear waves are can lead to abnormally high attenuation. Figure 7(b) shows
likely to converge, and consequently the time delay may be there is no Gassmann effect for the S-wave case because at low

Downloaded from [Link] by National Geophysical Research Institute user on 05 September 2023
too small to determine accurately. In addition, time delays are frequency the shear modulus is decoupled from the saturating
relatively insensitive to fracture orientations. fluid. S-wave attenuation is very small for gas saturation
compared to the P-wave case. Note that the bulk modulus of
Frequency-dependent attributes and spectra decomposition the water is slightly higher here, but it should not affect the
modelling results (Chapman et al 2006), and the magnitude of
Recent years have seen the development of technologies the frequency dependence is influenced strongly by the extent
which exploit the phenomenon of frequency-related seismic
of equant porosity (Chapman 2003).
attributes, such as dispersion and attenuation, for predicting
As the elastic constants of dispersive materials are
reservoir fluid, and most of these are empirical understandings
frequency dependent, P-wave reflection coefficients vary with
based on extrapolation of laboratory measurements.
frequency. This variation depends on the AVO behaviour at the
Chapman (2003) has developed a dynamic poroelastic
theory for the frequency-dependent properties of fractured interface. For class III sand with low impedance, the absolute
rock. The model contains two scale lengths of reservoir value of impedance contrast will decrease with frequency
heterogeneities: microscale and mesoscale. For mesoscale where dispersion occurs, so reflection shifts towards lower
fluid-filled open fractures (longer than the grain scale but still frequency values. For class I AVO with high impedance,
shorter than the wavelength, on the metre scale for typical rock the absolute value of impedance contrast will be increased
properties) the theory predicts that the anisotropy becomes with frequency in the case of dispersion, so reflection tends
frequency dependent, with the frequency dependence being to shift towards higher frequency values. AVO classes were
related directly to the scale lengths of the fractures, as well as introduced by Rutherford and Williams (1989) and discussed
the types of fluid within the fractures. In particular, shear-wave in more details in their publications.
splitting is predicted to depend on frequency, while P-wave To derive frequency-related attributes, it is necessary
attenuation and dispersion vary azimuthally. This provides to use spectral decomposition techniques. Spectral
a basis for relating frequency-dependent seismic attributes decomposition has been used for accurate structure mapping
directly to the properties of interest, principally the dominant over the past few years. Recently, it has been used as a direct
orientation and scale length of the open fractures, and the hydrocarbon indicator (Partyka et al 1999). The traditional
reservoir fluid. spectrum generated by the Fourier transform gives the spectral
In the case of reflection from a single interface, amplitudes at each frequency over the entire record length
Gassmann’s theory predicts that changing the fluid saturation and contains no information about the evolution of frequency
in one of the layers will change the impedance contrast and content over time. Spectral decomposition overcomes this
therefore the amplitude of the reflection. Chapman et al (2006) problem and provides an instantaneous frequency analysis for
have presented the mathematical procedure for calculating each time sample.
the frequency-dependent seismic reflection and transmission To perform spectral decomposition, there are several
responses. In this work, the introduction of gas results in a techniques available such as Gabor transform (Margrave and
marked increase in attenuation and associated dispersion. The Lamoureux 1998, 2001), short time Fourier transform (STFT,
existence of strong dispersion in a hydrocarbon-saturated layer Partyka et al 1999), the maximum entropy method (Penland
leads to the frequency dependence of the impedance contrast at et al 1991), Stockwell transform (Stockwell et al 1996),
the interface and so makes the reflection coefficient frequency
continuous wavelet transform (Sinha et al 2005), Wigner-Ville
dependent. This tends to markedly shift the reflections to
transform (Rauch-Davies and Ralston 2005) and matching
higher or lower frequency compared to the background trend,
pursuit decomposition (Castagna et al 2003). Castagna et al
with the direction of the shift depending on the AVO class of
(2003) compared the different techniques and pointed out that
the reflection. This reflection response does not depend on the
the resultant time–frequency analysis is not unique; many
thickness of the layer and is the main focus of our study.
different time–frequency decompositions can result from the
Figure 7 shows an example of the velocity and attenuation
same seismogram.
variation with frequency. Figure 7(a) shows the P-wave
velocity variation drift from high frequency to low frequency We have successfully applied the new theoretical rock
for both gas and water saturations for a sand rock. The physics model: (1) to determine fracture scale length and
parameters in sand are Vp = 2790 m s−1, Vs = 1463 m s−1, to distinguish between open or closed fractures in VSP data
and ρ = 2.08 g cc−1 with water saturation, 2.06 g cc−1 with (Maultzsch et al 2003), (2) to predict gas anomalies in
gas saturation, porosity = 30%. The gas bulk modulus is reflection data (Odebeatu et al 2006), and (3) to distinguish oil–
taken as 400 MPa and water as 2000 MPa. The reference water using multicomponent seismic data (Qian et al 2007).

130
 Seismic reservoir characterization: how can multicomponent data help?

Downloaded from [Link] by National Geophysical Research Institute user on 05 September 2023
(a) P-Velocity (b) P-wave attenuation

(c) S-Velocity (d ) S-wave attenuation

Figure 7. (a) Predicted P-velocity and attenuation as a function of the non-dimensional frequency ωτ under gas and water saturation (solid
line—gas, dashed line—water) (Chapman et al 2006). (b) Predicted S-velocity and attenuation as a function of the non-dimensional
frequency ωτ under gas and water saturation (solid line—gas, dashed line—water) (Chapman et al 2006).

(a) Survey location (b) Surface topography

Figure 8. (a) Location of Xinchang 3D3C survey, and (b) topography in Xinchang 3D3C survey.

Field data example in southwest China. As shown in figure 2, the reservoir

formation is a tight sandstone and buried at around 5000 m
A good example of using multicomponent reflection data in depth from the surface. Due to the very low porosity and
hydrocarbon exploration comes from the Xinchang gas field, matrix permeability, these reservoirs are classified as deep

131
X-Y Li and Y-G Zhang

Table 1. Xinchang 3D3C acquisition parameters.

Template parameters
• Template type: 12 receiver lines, 264 receivers in each line, 16 shots, brick layout.
• Bin size: 25 m × 25 m
• Fold number: 11(v) × 6(h)
• Total traces in single component: 3168
• Receiver interval: 50 m
• Shot interval: 50 m
• Receiver line interval: 400 m
• Shot line interval: 300 m

Downloaded from [Link] by National Geophysical Research Institute user on 05 September 2023
• Offset (min): 25 m
• Offset (max): 8010 m
• Aspect ratio: 0.7
• Rolling gap: 800 m (two lines)
• Spread in one line: 6575-25-50-25-6575
Explosion parameters
• Source type: dynamite
• Source size: 10 kg
• Explosion depth: average 14 m
Receiving parameters
• Receiver type: VectorSeis I/O company
• Number of sensor: single digital 3C
• Array: N/A
Instrument factors
• Acquisition equipment: SYSTEM FOUR
• Sampling rate: 1 ms
• Recorder length: 7 s
• Anti-Nyquist: 3/4 Nyquist
• Record format: SEG-Y
• Normal polarity: negative at first break

buried tight sandstone reservoirs with extremely low porosity crossline (Y) component, indicating the presence of shear-wave
and permeability (Tang et al 2008). Therefore, production is splitting due to anisotropy.
entirely dependent on fractures and fracture intensity. In 2005, Due to the good data quality, the workflow to perform
a 3D3C multicomponent survey was acquired to characterize anisotropic processing is relatively straightforward, but the
the fracture reservoirs, and the final results are presented here workload is intensive. The whole dataset consists of 29
to illustrate the information content discussed previously. swaths, and each swath consists of 12 receiver lines, 264
3C receivers in each line and 16 shots as specified in
Data acquisition and characteristics table 1. The data are divided into 18 azimuthal sectors in
order to handle the azimuthal anisotropy. The following
Xinchang gas field is located near Deyang City, 60 km north of
nine steps are applied to each azimuthal sector of the three
Chengdu, the capital city of the Sichuan province (figure 8(a)).
components: (1) geometry loading; (2) static correction;
The survey area is relatively flat with a gentle dip towards the
(3) noise attenuation (ground roll); (4) CMP bin for PP, but
west. The eastern part of the survey area is mostly foothills,
ACP (asymptotic conversion point) bin for PS with a binning
occupying about 20% of the survey area, and the rest of the area
is a plain, as shown in figure 8(b). The near surface consists velocity ratio γ = 2.0; (5) velocity analysis, (6) moveout
largely of quaternary soft depositions, including hard and soft correction; (7) stack; (8) migration velocity analysis and (9)
sands, gravel, and channel pebble overlaid by 20 to 30 cm of final prestack time migration.
thin clay from east to west. However, the deep subsurface Figure 10 shows the prestack time-migrated CIP gathers
structure is simple with gentle dips of 2–4◦ . The acquisition for some selected azimuthal sectors, and the data are
parameters are documented in table 1. displayed in the source-centred cylindrical coordinate system
The use of digital MEMS sensors has led to the acquisition (figure 1(b)). The radial components (figure 10(a)) show
of very high quality shear-wave data (figure 9), and major the travel time variation with azimuth of the fast (S1) and
events can be clearly identified in the original shot records, slow (S2) split shear waves; the transverse components
although some ground roll can be observed. In general, most (figure 10(b)) exhibit the polarity reversal; azimuthal variation
of the original records have strong reflected and converted of residual moveout in the P-wave gathers can also be observed
energy and high S/N ratio. Continuity is good in the three (figure 10(c)). After the initial data processing, the
components on account of 80% of shot points and sensors following analyses have been applied to the prestack gathers:
being operational in this survey. The dominant frequency is (1) reflection moveout analysis for anisotropic parameter
about 40 Hz for the P-wave and about 20 Hz for the converted estimation and velocity model building; (2) joint PP- and
wave. There is a significant amount of energy present in the PS-impedance inversion; (3) azimuthal analysis of P-wave

132
 Seismic reservoir characterization: how can multicomponent data help?

Downloaded from [Link] by National Geophysical Research Institute user on 05 September 2023
(a) Original horizontal X-component

(a) Radial components

(b) Original horizontal Y-component

(b) Transverse component

(c) Original horizontal Z-component

Figure 9. Original sample shot records, displayed in the acquisition

coordinate system of x (in line), y (crossline) and z (vertical).

attributes; (4) shear-wave splitting analysis; and (5) frequency-

dependent AVO analysis. The results of these analyses are
presented below.
(c) Vertical component

Final results Figure 10. Azimuth-sorted CIP gathers for the horizontal and
vertical components.
Figure 11 shows the anisotropic parameters estimated from
the PP and PS reflection moveout for inline 1540 and
azimuth sector 9, illustrating the four converted-wave stacking
parameters (Vc2, γ 0, γ eff and χ ). These parameters are then and transverse sections are very different despite there being
used to perform anisotropic prestack time migration for the one-to-one correlation in the major horizons. Since the signal
radial, transverse and vertical components of azimuth sector in the transverse component is very weak, a slightly heavy
9, inline 1540. Figures 12 and 13 show comparisons of the trace mixing is employed to improve the signal–noise ratio,
final migrated sections of the vertical, radial and transverse and this was also likely to have smoothed out the structural
components for the selected line and azimuth sector. Note features.
that there is a good correlation between the sections of vertical Figure 14(a) shows the inverted impedance sections for
and radial components. The target horizon is at 2.4 s PP time, the target horizon, and the attribute displayed is the ratio of the
and 3.6 PS time at well location W851, giving a Vp /Vs ratio P- and S-wave impedance. The impedance ratio is equivalent
of 2. However, the details of the time structure in the radial to the Vp /Vs ratio, and the warm colour represents the area

133
X-Y Li and Y-G Zhang

Downloaded from [Link] by National Geophysical Research Institute user on 05 September 2023
(a) VC2 (b) γ0

(c) γeff (d ) χeff

Figure 11. The four parameters: (a) VC 2, (b) γ 0, (c) γ eff and (d) χ eff estimated from PP and PS reflection moveout of the vertical and radial
components, azimuth sector 9 and inline 1540.

with a low Vp /Vs ratio indicating gas sand distribution. The Integrated interpretation
warm colours correlate well with the high productive wells
Figure 16 shows the final composite drilling targets interpreted
such as W851. The estimated azimuthal anisotropy for the
from the above attributes. The targets are superimposed in
target horizon from P- and converted waves is shown in
the fracture maps predicated by the previous analysis. All
figures 15(a) and (b), respectively. Figure 15(a) shows
the targets are at least indicated on two attribute maps in
the azimuthal variation of the interval velocity, the colour figures 12–15. Type I targets are indicated by two attributes,
representing the ratio of the short to the long axis of the type II are by three attributes, and type III are by all attributes.
velocity ellipse, and the line showing the direction of the For example, targets around well W851 are indicated in all
fracture strike, whilst in figure 15(b) the colour displays the four attribute maps, and these are the most consistent targets.
time delay between the fast and slow shear waves, and the short Among the four types of attributes, the most reliable is the
line represents the direction of fast wave polarization. W851 Vp /Vs impedance attribute characterizing the gas sands by a
is a highly productive well, which is characterized by a higher low velocity ratio, correlating with wells most consistently.
amount of anisotropy. The attenuation attribute estimated The PP-azimuthal and shear-wave splitting attributes are also
from frequency-dependent analysis is shown in figure 14(b), very reliable in characterizing the fractured gas sand. In
which marks the area of gas concentration (the zones of warm contrast, the attenuation attributes are relatively unreliable.
colours). The integrated analysis and the final results confirm that we

134
 Seismic reservoir characterization: how can multicomponent data help?

Downloaded from [Link] by National Geophysical Research Institute user on 05 September 2023
(a) Radial component (PS-wave) (b) Vertical component (PP-wave)

Figure 12. Comparison of the radial (PS-wave) and vertical (PP-wave) sections for the selected data.

(a) Radial component (b) Transverse component

Figure 13. Comparison of the radial and transverse sections for the selected data.

can characterize the reservoir heterogeneity from lithology to in the title, we know that hydrocarbon reserves will remain
fractures as well as fluid distribution in the study area using healthy well into the next 50 years; nevertheless, finite reserves
3D3C multicomponent seismic data. combined with increasing demand means that better reservoir
characterization is needed.
Discussions and conclusions Hydrocarbon reservoirs containing aligned fractures are
anisotropic, and therefore examination of the resultant split
This paper has discussed the concept of multicomponent shear waves can give us vital information about reservoir
reflection surveys and data processing, and has also structure. Hydrocarbon reservoirs become increasingly
described several types of information, deduced from the complex, and examining the resulting frequency-dependent
multicomponent data. In discussing what we can obtain information can provide vital information about fluid
from multicomponent data, it was established that anisotropy properties. The example presented here is of particular
can play a major role in reservoir characterization, in terms importance, as it demonstrates not only proven applications,
of providing both better imaging and better characterization but also the undoubted potential of multicomponent seismic
of the internal structure. In answer to the question posed in reservoir characterization.

135
X-Y Li and Y-G Zhang

Downloaded from [Link] by National Geophysical Research Institute user on 05 September 2023
(a) Joint PP−PS impedance inversion

(b) Attenuation gradient from the PP

Figure 14. Comparison of impedance inversion with frequency-dependent analysis, (a) the inverted ratio of the P- and shear-wave
impedance from the vertical and radial components, with warm colours indicating the sand distribution. (b) Inverted attenuation gradient
from the PP data by frequency-dependent analysis.

Finally, although we have made substantial progress in • Accounting for the effects of the near surface is still
the research and development of multicomponent seismology, a challenge. Shear-wave velocities vary sharply in the
near surface, and degrade the data quality. A full
there are still big challenges ahead, in particular, the quality of and comprehensive near surface survey may have to be
converted-wave imaging and interpretation, as detailed here. considered. Another option is to bury the receivers in the

136
 Seismic reservoir characterization: how can multicomponent data help?

Downloaded from [Link] by National Geophysical Research Institute user on 05 September 2023
(a) Azimuthal variation in P-wave interval velocity

(b) Time-delay of the fast and slow shear-waves

Figure 15. Comparison of the azimuthal anisotropy estimated from (a) P-wave interval velocity and (b) converted-wave splitting for the
target horizon.

ground, although both of these measures will increase the with sufficient fold in both positive and negative offsets
cost of acquisition. (for one limited-azimuth sector) so that the positive and
• Accounting for the characteristics of converted-wave (C- negative offset images can be properly formed.
wave) propagation is still difficult. Due to overburden • The interpretation of C-waves is at the discovery stage,
velocity variations and an imperfect velocity model in despite the availability of several standard industrial
imaging, reconciling images acquired with positive and software programs. The multiplicity of volumes can
negative offsets in the absence of sufficient fold coverage complicate the life of the interpreter, particularly when
can be frustrating. Therefore, one must acquire PS data the P and C volumes are not on the same time scale (the

137
X-Y Li and Y-G Zhang

of the TI medium. Equation (A.1) shows that the P-wave

reflection moveout in a VTI medium is controlled by only two
parameters, the short-spread P-wave NMO velocity VP 2 and
anisotropic parameter η.
Similarly, S-wave reflection moveout (tS ) in a VTI
medium can be written as (Li and Yuan 2003)
x2 2ζ x 4
tS2 = tS0
2
+ 2
− 22 2 , (A.2)
VS2 VS2 tS0 VS2 + x 2

Downloaded from [Link] by National Geophysical Research Institute user on 05 September 2023
where tS0 is the vertical S-wave two-way, VS 2 is the short-
spread S-wave NMO velocity, and ζ = γ 2eff η. Note that γ eff
is the effective velocity ratio (γ eff = γ 22/γ 0), first introduced
by Chung and Corrigan (1985), and further developed by
Thomsen (1999), γ 2 is the P- and SV-wave stacking velocity
ratio (γ 2 = VP 2/VS 2), and γ 0 is the vertical velocity ratio (γ 0 =
VP 0/VS 0). Equation (A.2) shows that only two parameters are
required for describing the SV-wave behaviour in VTI media,
Figure 16. Final composite drilling targets integrated from the the SV-reflection moveout is controlled by the short-spread
previous results. Roman numerals indicate the new targets SV-wave NMO velocity VS 2 and the anisotropic parameter ζ .
identified, the red circles indicate the new well positions, white Based on the above parameterization, Li and Yuan (2003)
curves mark the possible boundaries of the targets, and the blue
shapes in the background mark the possible fracture zone as derived an improved four-parameter equation for quantifying
predicted by the PP- and PS-wave attributes. the PS converted-wave (C-wave) reflection moveout in a VTI
medium:
x2 A4 x 4
registration problem) and in the absence of calibrating tC2 = tC0
2
+ 2 + , (A.3)
VC2 1 + A5 x 2
VSPs.
• Slow processing and interpretation turnaround are the where
main reasons why C-wave imaging does not yet impact (γ0 γeff − 1)2 + 8(1 + γ0 )χ
A4 = − 2 4
, and
drilling decisions and leads to operators questioning the 4tC0 VC2 γ0 (1 + γeff )2
added value of shear information.  
2
A4 VC2 (1 + γ0 )γeff (γ0 − 1)γeff
2
+ 2χ
A5 = . (A.4)
(γ0 − 1)γeff
2
(1 − γ0 γeff ) − 2(1 + γ0 )γeff χ
Acknowledgments
VC 2 is the short spread C-wave stacking velocity, γ 0 and γ eff
We thank SinoPec for permission to show the data, and the are the vertical and effective velocity ratios, and χ is the C-
contractor GXT for providing some of the processed results. wave anisotropic coefficient. For the layered media, the first
We thank Hengchang Dai, Mark Chapman, Ma Zhaojun, three parameters (VC 2, γ 0 and γ eff) are defined as
Li Xiangui and Qian Zhongping for pre-processing the data 2
2tC0 VC2 = tP 0 VP22 + tS0 VS2
2
; 2tC0 = tP 0 + tS0 ;
and discussing the results. We thank Tang Jianming for γ0 = tS0 /tP 0 ; (A.5)
various discussions and David Booth for his comments on
the manuscript. This work is supported by the SinoPec γeff = γ22 /γ0 ; γ2 = VP 2 /VS2 ,
international collaboration programme through the Edinburgh where tC 0 is the corresponding two-way vertical C-wave travel
Anisotropy Project (EAP) of the British Geological Survey time, and the anisotropic coefficient χ is defined as
(BGS), and is published with the approval of all project
χ = ηγeff
2
γ0 − ζ = ηγeff
2
(γ0 − 1). (A.6)
partners and the executive director of BGS (NERC). This work
is also partially supported by the National Natural Science Equations (A.3) and (A.4) are controlled by four parameters
Foundation of China (grant no. 41074080). (VC 2, γ 0, γ eff and χ ), and they are accurate for an offset-
depth ratio of 2.0 (x/<2.0) (Li and Yuan 2003). These four
parameters are referred to as the C-wave stacking velocity
Appendix A. P-, S- and PS converted-wave reflection
model.
moveout signatures in VTI

The P-wave reflection moveout (tp ) in a VTI medium can be Appendix B. Parameter dependence and anisotropic
written as (Alkhalifah and Tsvankin 1995) links
x2 2ηx 4
tP2 = tP2 0 + −  , (A.1) Various parameters are used for describing P and PS-converted
VP22 VP22 tP2 0 VP22 + (1 + 2η)x 2 waves in VTI media. These parameters may be divided into
where tP 0 is the vertical two-way time, x is the source– various four-parameter sets, and each set consists of four
receiver offset, VP 2 is the short-spread P-wave NMO velocity, independent parameters, controls different wave behaviours,
and η = (ε − δ)/(1 + 2δ). η represents the anellipticity and has different usages. There are four common sets. The first

138
 Seismic reservoir characterization: how can multicomponent data help?

set is (Vp 0, Vs 0, ε and δ), from which all the other parameter Alford R M 1986 Shear data in the presence of azimuthal
sets can be derived. This is the original Thomsen (1986)’s anisotropy: Dilley, Texas 56th Int. Annu. Meeting Soc. Expl.
Geophys. Expanded Abstracts pp 476–9
parameter set, and it is often used for forward modelling
Alkhalifah T and Larner K 1994 Migration error in transversely
and depth imaging. The second four-parameter set is Vp 0, isotropic media Geophysics 59 1405–18
Vs 0, δ and σ ; this set controls the wave behaviours at near Alkhalifah T and Tsvankin I 1995 Velocity analysis for transversely
vertical propagation, and it is essential for depth processing isotropic media Geophysics 60 1550–66
and conversion. However, this set is not as widely used as the Angerer E, Crampin S, Li X-Y and David T L 2002 Processing,
modelling, and predicting time-lapse effects of overpressured
first set. The third set is Vp 2, Vs 2, η and ζ , and this set controls fluid-injection in a fractured reservoir Geophys. J.

Downloaded from [Link] by National Geophysical Research Institute user on 05 September 2023
the P and SV-wave reflection moveout and time-migration Int. 149 267–80
process: (Vp 2, η) for P-wave and (Vs 2, ζ ) for SV-wave, Anno P D 1987 Two critical aspects of shear-wave analysis: statics
respectively. Consequently, this set also controls the time- solutions and reflection correlations Shear-Wave Exploration:
migration processing of the P-SV converted waves since the SEG Geophysical Development Series 1 3–18
ed S H Danbom and S N Domenico
converted wave consists of a down-going P-wave leg and an up- Ball G 1995 Estimation of anisotropy and anisotropic 3-D prestack
going SV-wave leg. The fourth set is Vc2, γ 0, γ eff and χ , which depth migration, offshore Zaire Geophysics 60 1495–513
controls the reflection moveout of the P-SV converted waves Blott J E, Davis T L and Benson R D 1999 Morrow sandstone
and the stacking process. In theory, these four four-parameter reservoir characterization: a 3-D multicomponent seismic
success Leading Edge 18 394–7
sets can be calculated from each other and there is a one-to-one Byun B S 1982 Seismic parameters for media with elliptical
correspondence between each of them. The ability to relate velocity dependencies Geophysics 47 1621–6
between different parameters for different wave behaviours is Castagna J P, Sun S and Siegfried R W 2003 Instantaneous spectral
the major advantage of using the Thomsen parameters, which analysis: detection of low-frequency shadows associated with
makes the estimation of anisotropy parameters realistic and hydrocarbons Leading Edge 22 120–7
Chapman M 2003 Frequency dependent anisotropy due to
practical. meso-scale fractures in the presence of equant porosity
Since Vc2, γ 0, γ eff and χ can often be estimated from the Geophys. Prospect. 51 369–79
P- and C-wave reflection moveout, there is a need to calculate Chapman M, Liu E and Li X-Y 2006 The influence of fluid-sensitive
the other parameters from this set. Based on the definition in dispersion and attenuation on AVO analysis Geophys. J.
Int. 167 89–105
equation (A.5), we have
Cherry J T and Waters K H 1968 Shear-wave recording using
continuous signal methods (part I) Geophysics 33 229–39
γeff (1 + γ0 ) (1 + γ0 )
VP22 = VC2
2
, 2
VS2 = VC2
2
. Chung W Y and Corrigan D 1985 Gathering mode-converted shear
1 + γeff γ0 (1 + γeff ) waves: a model study 55th Int. Ann. Meeting Soc. Expl.
(B.1) Geophys. Expanded Abstracts pp 602–4
Corrigan D, Withers R, Darnall J and Skopinski T 1996 Fracture
mapping from azimuthal velocity analysis using 3-D surface
This defines the link between (VP 2, VS 2) and (VC 2, γ 0, γ eff). seismic data 66th Int. Ann. Meeting Soc. Expl. Geophys.
The links between the anisotropic parameters are as Expanded Abstracts pp 1834–7
follows: Crampin S 1981 A review of wave motion in anisotropic and
χ χ cracked elastic media Wave Motion 3 343–91
η= ; ζ = . (B.2) Crampin S 1985 Evaluation of anisotropy by shear-wave splitting
(γ0 − 1)γeff
2 (γ0 − 1) Geophysics 50 142–52
Crampin S and Lovell J H 1991 A decade of shear-wave splitting in
Therefore, there is a one-to-one analytical relationship the Earth’s crust: what does it mean? What use can we make of
between the parameter sets (Vp 2, Vs 2, η and ζ ) and (Vc2, γ 0, it? And what should we do next? Geophys. J. Int. 107 387–407
γ eff and χ ). The other two parameter sets can be calculated as Davis T L 2007 Let’s get cracking Leading Edge 26 160–1
Dellinger J and Muir F 1988 Imaging reflections in elliptically
  anisotropic media (short note) Geophysics 53 1616–8
ηγ0 γeff 1 γeff
σ = ; δ= − γ0 (B.3) Erickson E L, Miller D E and Waters K H 1968 Shear-wave
1 − 2ηγ0 γeff 2γ0 1 − 2ηγ0 γeff recording using continuous signal methods (part II)
  Geophysics 33 240–54
1 γeff η 1 γeff
ε= + − γ0 (B.4) Fomel S and Backus M 2003 Multicomponent seismic data
γ0 1 − 2ηγ0 γeff 2γ0 1 − 2ηγ0 γeff registration by least squares 73rd Int. Annu. Meeting Soc. Expl.
Geophys. Expanded Abstracts, MC2.1 pp 781–4
Gaiser J E 1996 Multicomponent Vp /Vs correlation analysis
and Geophysics 61 1137–49
 
γ0 Gaiser J E 1997 3-D converted shear wave rotation with layer
VP20 = VP22 − 2ηγ0 ,
2 2
VS0 = VS2
2
(1 − 2ηγ0 γeff ). stripping: US Patent 5610875
γeff
Gaiser J E 1999 Applications for vector coordinate systems of 3-D
(B.5) converted-wave data Leading Edge 18 1290–300
Gaiser J E, Moldoveanu N, MacBeth C, Michelena R and Spitz S
2001 Multicomponent technology: the players, problems,
applications, and trends Leading Edge 20 974–7
Gaiser J and Hilterman F 2010 Why the delay in S-wave
References interpretation? Geophysical Society of Houston,
Robinson-Treitel Spring Symposium
Aki K and Richards P G 1980 Quantitative Seismology: Theory and Gaiser J E and Van Dok R 2001 Green River basin 3D/3C case
Methods vol 1 (San Francisco: Freeman) study for fracture characterization: analysis of PS-wave

139
X-Y Li and Y-G Zhang

birefringence 71st Ann. Meeting Soc. Expl. Geophys. Expanded Mattocks B, Li J and Roche S L 2005 Converted-wave azimuthal
Abstracts, MC1.1 pp 764–7 anisotropy in a carbonate foreland basin 75th Int. Annu.
Garotta R and Granger P Y 1988 Acquisition and processing of Meeting Soc. Expl. Geophys. Expanded Abstracts 897–900
3C× 3-D data using converted waves 58th Int. Annu. Meeting Maultzsch S, Chapman M, Liu E and Li X-Y 2003 Modelling
Soc. Expl. Geophys. Expanded Abstracts pp 995–7 frequency dependent seismic anisotropy in fluid-saturated rock
Garotta R, Granger P Y and Dariu H 2002 Combined interpretation with aligned fractures: implication of fracture size estimation
of PP and PS data provides direct access to elastic rock from anisotropic measurements Geophys. Prospect.
properties Leading Edge 21 532–5 51 381–92
Grechka V, Bakulin A and Tsvankin I 2005 Parameter estimation in Mueller M C 1991 Prediction of lateral variability in fracture
orthorhombic media using multicomponent wide-azimuth intensity using multicomponent shear wave surface seismic as

Downloaded from [Link] by National Geophysical Research Institute user on 05 September 2023
reflection data Geophysics 70 D1–D8 a precursor to horizontal drilling Geophys. J. Int.
Helbig K 1983 Elliptical anisotropy—its significance and meaning 107 409–15
Geophysics 48 825–32 Odebeatu E, Zhang J, Chapman M, Liu E and Li X-Y 2006 The
Ikwuakor K C 2006 The Vp /Vs ratio after 40 years: uses and application of spectral decomposition to detection of dispersion
abuses 76th Int. Annu. Meeting Soc. Expl. Geophys. Expanded anomalies associated with gas saturation Leading
Abstracts pp 1183–6 Edge 25 205–10
Jenner E 2001 Azimuthal anisotropy of 3D compressional wave Partyka G, Gridley J and Lopez J 1999 Interpretational applications
seismic data PhD Thesis Colorado School of Mines of spectral decomposition in reservoir characterization Leading
Kuehnel T and Li X-Y 1996 Anisotropy versus dip—a separation Edge 18 353–60
approach 66th SEG Meeting Expanded Abstracts pp 1866–9 Penland C, Ghil M and Weickmann K 1991 Adaptive filtering and
Levin F K 1978 The reflection refraction and diffraction of waves in maximum entropy spectra with application to changes in
media with an elliptical velocity dependence atmospheric angular momentum J. Geophys.
Geophysics 43 528–37 Res. 96 22659–71
Lewis C, Davis T L and Vuillermoz C 1991 Three-dimensional Qian Z, Li X-Y, Chapman M, Zhang Y and Wang Y 2007 Effects of
multicomponent imaging of reservoir heterogeneity, Silo field, oil-water saturation on shear-wave splitting in multicomponent
Wyoming Geophysics 56 2048–56 seismic data 77th Int. Annu. Meeting Soc. Expl. Geophys.
Li X-Y 1997 Fractured reservoir delineation using multicomponent Expanded Abstracts vol 26 pp 1019–22
seismic data Geophys. Prospect. 45 39–64 Rauch-Davies M and Ralston M 2005 Spectral
Li X-Y 1998 Processing PP and PS waves in multicomponent decomposition–transform methods and fluid and reservoir
sea-floor data for azimuthal anisotropy: theory and overview prediction case study 67th EAGE Conf. and Exhibition
Proc. 8th Int. Workshop on Seismic Anisotropy (Revue De Extended Abstracts: F007
L’institut francia du petrole) 53 607–20 Roche S L, Wagaman M and Watt H J 2005 Analysis of P-wave and
Li X-Y, Dai H and Mancini F 2007 Converted-wave imaging in converted-wave 3D seismic data, Anadarko Basin, Oklahoma,
anisotropic media: theory and case studies Geophys. USA 75th Int. Annu. Meeting Soc. Expl. Geophys. Expanded
Prospect. 55 345–63 Abstracts pp 979–82
Li X-Y and Yuan J 2003 Converted-wave moveout and Rueger A 1996 Reflection coefficients and azimuthal AVO analysis
conversion-point equations in layered VTI media: theory and in anisotropic media PhD Thesis Colorado School of Mines
application J. Appl. Geophys. 54 297–318 CWP-235
Lynn H B, Simon K M, Layman M, Schneider R, Bates C R and Rutherford S R and Williams R H 1989 Amplitude-versus offset
Jones M 1995 Use of anisotropy in P-wave and S-wave data for variations in gas sands Geophysics 54 680–8
fracture characterization in a naturally fractured gas reservoir Sheriff R E 2006 Encyclopedic Dictionary of Exploration
Leading Edge 14 887–93 Geophysics Geophysical References Series vol 1 SEG Special
Lynn H B and Thomsen L 1990 Reflection shear wave data publication
collected near the principal axes of azimuthal anisotropy Shuck E L, David T L and Benson R D 1996 Multicomponent 3-D
Geophysics 55 147–56 characterization of a coalbed methane reservoir
Lynn W 2007 Uncertainty implications in azimuthal velocity Geophysics 61 315–30
analysis 77th Int. Annu. Meeting Soc. Expl. Geophys. Sinha S, Routh P S, Anno P D and Castagna J P 2005 Spectral
Expanded Abstracts pp 84–7 decomposition of seismic data with continuous wavelet
MacBeth C 1995 How can anisotropy be used for reservoir transform Geophysics 70 P19–25
characterization? First Break 13 31–7 Smith G C and Gidlow P M 1987 Weighted stacking for rock
MacBeth C 2001 Multicomponent VSP analysis for applied seismic property estimation and detection of gas Geophys.
anisotropy Handbook of Geophysical Exploration: Seismic Prospect. 35 993–1014
Exploration 26 Elsevier Science Publication 7 Stewart R R 1990 Joint P and P-SV inversion Crewes Res. Rep. 2
MacBeth C and Li X-Y 1999 AVD—an emerging new marine 112–5
technology for reservoir characterization: acquisition and Stewart R R, Gaiser J E, Brown R J and Lawton D C 2002 Converted
application Geophysics 64 1153–9 wave seismic exploration: Methods Geophysics 67 1348–63
Mallick S, Craft K, Meister L and Chambers R 1996 Determination Stewart R R, Gaiser J E, Brown R J and Lawton D C 2003 Converted
of the principal directions of azimuthal anisotropy from P-wave wave seismic exploration: Applications Geophysics 68 40–57
seismic data 58th EAGE Conf. and Exhibition Extended Stockwell R G, Mansinha L and Lowe R P 1996 Localization of the
Abstracts C023 complex spectrum: the S transform IEEE Transactions on
Margrave G F and Lamoureux M P 1998 Gabor deconvolution of Signal Processing 44 998–1001
seismic data for source waveform and Q correction Geophysics Tang J, Zhang S and Li X-Y 2008 PP and PS seismic response from
72 2190–3 fractured tight gas reservoirs: a case study J. Geophys.
Margrave G F and Lamoureux M P 2001 Gabor deconvolution The Eng. 5 92–102
CREWES Project Research Report vol 13 pp 241–76 Tatham R H 1982 Vp /Vs and lithology Geophysics 47 336–44
Martin A M and Davis T L 1987 Shear wave birefringence: a new Tatham R H and McCormack M D 1991 Multicomponent
tool for evaluating fractured reservoirs Leading Edge Seismology in Petroleum Exploration Investigations in
6 22–8 Geophysics 6 SEG special publication

140
 Seismic reservoir characterization: how can multicomponent data help?

Thomsen L 1986 Weak elastic anisotropy Geophysics VerWest B J 1989 Seismic migration in elliptically anisotropic
51 1954–66 media Geophys. Prospect. 37 149–66
Thomsen L 1999 Converted-wave reflection seismology over Winterstein D F and De G S 2001 VTI documented
inhomogeneous, anisotropic media Geophysics 64 678–90 Geophysics 66 237–45
Tsvankin I 1996 P-wave signatures and notation for transversely Won I J and Clough J W 1981 A new torsional shear-wave generator
isotropic media: an overview Geophysics 61 467–83 Geophysics 46 1607
Tsvankin I 1997 Anisotropic parameters and P-wave velocity for Zhang J and Li X-Y 2005 Joint PP- and PS-wave AVO
orthorhombic media Geophysics 62 1292–309 inversion—modelling and case studies from the Ordos Basin
Van Dok R R, Gaiser J E, Jackson A R and Lynn H B 1997 3-D EAP Res. Rep. 12 B1–1 – B1–45
converted-wave processing: Wind River Basin case history Zhu F, Gibson J, R L, Watkins J S and Yuh S H 2000

Downloaded from [Link] by National Geophysical Research Institute user on 05 September 2023
67th Ann. Int. Mtg. Soc. Explor. Geophys. (Dallas, expanded Distinguishing fizz gas from commercial gas reservoirs using
abstract SP5.4) 1206–9 multicomponent seismic data Leading Edge 19 1238–45

141