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

net/publication/252625646

One-way wave-equation migration of compressional and converted waves in a

VTI medium

Article  in  Geophysics · November 2010

DOI: 10.1190/1.3509466

CITATIONS READS
5 35

3 authors, including:

Ørjan Pedersen Bjørn Ursin

15 PUBLICATIONS   32 CITATIONS   
Norwegian University of Science and Technology
298 PUBLICATIONS   2,882 CITATIONS   
SEE PROFILE
SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Pore pressure from seismiç View project

Full Wave Seismic-Acoustic Propagation modelling: Horizontl/Vertical Arrays View project

All content following this page was uploaded by Bjørn Ursin on 23 October 2015.

The user has requested enhancement of the downloaded file.

 GEOPHYSICS, VOL. 75, NO. 6 共NOVEMBER-DECEMBER 2010兲; P. S237–S248, 19 FIGS.
10.1190/1.3509466
Downloaded 10/14/14 to 129.241.69.108. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
One-way wave-equation migration of compressional
and converted waves in a VTI medium
Ørjan Pedersen1, Bjørn Ursin1, and Hans-Kristian Helgesen2
ABSTRACT
In seismic reflection surveying, by recording both pressure
and shear-wave reflections, one can increase the amount of
information obtained about the subsurface rather than by re-
cording pressure waves alone. Geologic structures that are
not visible by using conventional pressure-data may possibly
be imaged using shear waves, thus mitigating the risk in oil
and gas exploration and production. Horizontally layered
sedimentary rocks exhibit anisotropy that can be approximat-
ed by an effective transverse isotropic medium with a vertical
axis of symmetry. Taking into account a vertically transverse
isotropic earth, we derive phase-slowness expressions for
quasi-P and quasi-SV waves that are used in a one-way wave-
equation migration scheme. We derive simplified slowness-
expressions that are useful for processing of conventional
pressure data. Numerical examples demonstrate that the
slowness approximations are valid for wide-angle propaga-
tion, and the resulting one-way propagators are validated on a
series of synthetic tests and applied on a field ocean-bottom
seismic data set. The results show that the method accurately
images both compressional and converted waves in OBS data
over a vertically transverse isotropic medium.
INTRODUCTION
In seismic reflection surveying, by deploying both hydrophones
and geophones at the seafloor as in ocean-bottom seismic 共OBS兲 sur-
veying, it is possible to record both pressure and shear-wave reflec-
tions from the subsurface. By combining both pressure and shear-
wave reflections, one can increase the amount of information about
the subsurface than obtained with pressure waves alone 共Hokstad,
2000兲. A dominant part of the seismic energy recorded on the hori-
zontal geophone components is energy that has experienced conver-
Manuscript received by the Editor 4 November 2009; revised manuscript received 15 March 2010; published online 2 December 2010.
1
Norwegian University of Science and Technology 共NTNU兲, Department of Petroleum, Engineering andApplied Geophysics, Trondheim, Norway and Statoil,
Trondheim, Norway. E-mail: orjan.pedersen@ntnu.no; bjorn.ursin@ntnu.no.
2
Statoil, Trondheim, Norway. E-mail: hkhe@statoil.com.
© 2010 Society of Exploration Geophysicists. All rights reserved.
S237
sion from pressure waves to shear waves at subsurface reflectors.
Converted shear-wave data can possibly be used to image subsur-
face reflectors that are weak using pressure data alone, especially in
gas-charged formations 共Granli et al., 1999; MacLeod et al., 1999;
Barkved et al., 2004兲, hence, reducing the risk in hydrocarbon explo-
ration and production. Shear-wave information may also help im-
prove reservoir characterization by providing further constraints on
rock properties, lithology, and fracture density and orientation.
The earth is anisotropic in nature and in particular sedimentary
rocks exhibit anisotropy. These sedimentary rocks may often be de-
scribed as being transverse isotropic with a symmetry axis perpen-
dicular to the bedding plane. A homogeneous medium that is frac-
tured may also be described as being transverse isotropic, now with a
symmetry axis perpendicular to the fractures. A transversely isotro-
pic media with a vertical symmetry axis 共VTI兲 often presents a good
model for the description of real rocks. The characteristics of wave-
propagation in a VTI medium can be described by the dispersion re-
lation, relating the vertical and horizontal phase-slowness. Seismic
waves, and shear waves in particular, are sensitive to anisotropy.
There are several migration methods used in subsurface imaging.
Depending on their underlying assumptions, these methods can usu-
ally be classified as either Kirchhoff 共Schneider, 1978; Sun et al.,
2000兲 or wavefield extrapolation 共Bleistein, 1987; Gazdag, 1978;
Stolt, 1978兲 methods. Kirchhoff methods explicitly introduce a
high-frequency approximation of the wave-equation. In areas with
complex geology where multipathing occurs, Kirchhoff methods
may not provide reliable subsurface images 共Biondi, 2006兲. In
wavefield extrapolation methods, multipathing is handled in a natu-
ral way. Two-way wave extrapolation methods, provide an accurate
description of wave propagation; however, the computational cost
and challenges in noise when creating subsurface images can be-
come a limitation. One-way wavefield extrapolation methods are
routinely used in 3D-depth migration of seismic data. One-way
methods separate the wavefields into upgoing and downgoing wave-
field constituents, so they are not valid for horizontally traveling
wavefield components. Because of their efficient computer imple-
 S238 Pedersen et al.

mentations and robustness, one-way methods have become increas- In terms of density and stiffness coefficients, the vertical qP velocity
ingly popular and a wide variety of methods has been introduced is given in equation 2 by
共Gazdag, 1978; Stolt, 1978; Gazdag and Sguazzero, 1984; Stoffa et
al., 1990; Wu and Huang, 1992; Ristow and Rühl, 1994兲.
Utilizing an isotropic propagator for migration in a VTI medium ␣0 ⳱ 冑 c33
␳
, 共2兲
Downloaded 10/14/14 to 129.241.69.108. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

may lead to position errors of subsurface structures 共Larner and Co-

hen, 1993; Alkhalifah and Larner, 1994; Vestrum et al., 1999兲. In a
VTI medium with no lateral variations where all medium parameters and the vertical quasi-SV 共qSV兲 velocity is given in equation 3 by
are known, the dispersion relation can be used directly in phase-shift
migration methods 共Gazdag, 1978; Alkhalifah, 2000b兲, and pro-
vides an accurate description of the kinematics of wave-propaga-
tion. If we allow the medium to contain small lateral variations, a
␤0⳱ 冑 c44
␳
, 共3兲

split-step Fourier approach 共Stoffa et al., 1990兲 accounts for the ze-
ro-order lateral perturbations in the medium parameters. To exploit where ␳ denotes the density of the medium. The Thomsen 共1986兲 pa-
the phase-shift approach efficiently with a fast Fourier transform rameters ␧ and ␦ are defined in equation 4 in terms of the stiffness
and, in addition, provide an accurate description of wave-propaga- coefficients by
tion in an isotropic laterally varying media, the scalar generalized- c11 ⳮ c33
screen 共Wu and Huang, 1992; Rousseau and de Hoop, 2001a兲 or ␧⳱ ,
2c33
Fourier finite-difference 共FFD兲 method 共Ristow and Rühl, 1994兲 can
be used. 共c13 Ⳮ c44兲2 ⳮ 共c33 ⳮ c44兲2
␦⳱ , 共4兲
The isotropic generalized-screen approach was extended to VTI 2c44共c33 ⳮ c44兲
media by Rousseau and de Hoop 共2001b兲.Asimplified dispersion re-
lation for quasi-P 共qP兲-waves was developed by Alkhalifah 共1998, and allows us to describe a VTI medium by the parameters ␣ 0, ␤ 0, ␦ ,
2000a兲. Based on this dispersion relation, Han and Wu 共2005兲 devel- and ␧. For notational convenience, in equation 5 we introduce the pa-
op a one-way propagator using a screen approximation with a finite rameters
difference correction term. Xie and Wu 共2005兲 present a multicom-
ponent prestack depth migration method utilizing the elastic screen ␣0
␥ 0⳱ ,
method. Nolte 共2005兲 describes a technique for converted-wave 共C- ␤0
wave兲 migration in anisotropic media with VTI symmetry using
␨ ⳱␧ⳮ␦ ,
FFD methods through least-squares fitting of finite difference coeffi-
cients to an anisotropic dispersion relation. ␴ ⳱ ␥ 20␨ ,
We derive one-way propagators for compressional and converted
␹ ⳱ 1 Ⳮ 2␦ 共5兲
waves through a Fourier finite-difference approach based on a repre-
sentation of the 共squared兲 vertical slowness as a function of horizon- The dispersion relation relates the vertical and the horizontal
tal slowness. We provide a simplified vertical-slowness approxima- slowness and can be found by inserting the stiffness coefficients in
tion involving fewer parameters that can be used in processing of matrix 1 into the Christoffel equation 共see, e.g., Červený 共2001兲兲.
compressional waves where not all medium parameters are avail- This relation can be expressed in terms of the Thomsen parameters ␧,
able. The accuracy of the derived one-way propagators is demon- ␦ , and the vertical qP, and the vertical qSV phase velocities ␣ 0 and
strated in 2D on synthetic data and in 3D on OBS field data from the ␤ 0. It can be solved for the squared vertical slowness q2 as a function
Volve field in the central North Sea. of horizontal-slowness p, both for qP- and qSV-wave modes.
From the dispersion relation, the squared vertical slowness for qP-
and qSV-waves in a VTI medium can be expressed in equation 6 by
共Stovas and Ursin, 2003; Ursin and Stovas, 2006兲
THEORY

We consider a VTI medium, which is described by its stiffness co- 1

q␣2 ,␤ ⳱ 共q␣2 Ⳮ q␤2 ⳮ 2p2共␴ Ⳮ ␦ 兲兲
efficients in the Voigt notation cij. For a VTI medium the only non- 2 0 0

冋
zero stiffness coefficients are seen in equation 1 共Thomsen, 1986兲
1 p2
⫿ 共q␤2 ⳮ q␣2 兲2 ⳮ 4 2 共␥ 20 ⳮ 1兲共␴ ⳮ ␦ 兲

冤 冥
c11 c11 ⳮ 2c66 c13
2 0 0 ␣0

冉 冊册
1
c11 ⳮ 2c66 c11 c13 共␥ 20 ⳮ 1兲 2
Ⳮ 4p 2 4
␴ Ⳮ 共␴ Ⳮ ␦ 兲 2
, 共6兲
c13 c13 c33
. 共1兲 ␥ 20
c44
c44 where q2␣ 0 ⳱ 1 / ␣ 20 ⳮ p2 and q2␤ 0 ⳱ 1 / ␤ 20 ⳮ p2. With no horizontal on-
axis shear-wave triplication 共Musgrave, 1970; Thomsen and Del-
c66
linger, 2003兲, the negative sign in front of the radical corresponds to
 One-way VTI propagator S239

the qP slowness and the positive sign corresponds to the qSV slow-
ness. In case of a horizontal on-axis shear-wave triplication, q␤ is
multivalued for some values of p 共Pedersen et al., 2007兲. This means
that q2␤ has a concave shape near the horizontal axis, that is, when p is
close to 1 / ␤ 0 and is described by both the positive and negative sign
Downloaded 10/14/14 to 129.241.69.108. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
Following the FFD approach 共Ristow and Rühl, 1994兲 we trun-
cate the series expansion in equation 9. These expressions will pro-
vide a vertical-slowness approximation that can be used in con-
structing one-way propagators in an FFD approach after cascading
by a continued-fraction approximation.

in front of the radical. To get the vertical wavenumber kz, the vertical
slowness q can be written as q ⳱ kz / ␻ , and the horizontal-slowness
p can be written p ⳱ kx / ␻ where kx is the lateral wave-number. SLOWNESS EXPRESSIONS FOR
For simplicity in our derivations, we assume that we consider a 2D QP- AND QSV-WAVES
VTI medium where we can describe the parameters ␣ 0 and ␤ 0, and
By a series representation of the square-root term in equation 6,
the anisotropy parameters ␧ and ␦ , as functions of both depth z and
the squared qP slowness can be written as
spatial position x. An extension to 3D is straightforward. The medi-
um is divided into thin depth-slabs of thickness ⌬z, where the param-
eters are assumed constant in depth within each slab, and the lateral
medium variations are smooth. Given a wavefield ⌿ 共z,:兲 at some
q␣2 ⳱
1
␣ 20冉1 ⳮ 兺 a j共p␣ 0兲2jⳭ2
jⱖ0
冊 共10兲

depth z, the solution of the one-way wave equation provides the thin-
slab propagator 共Claerbout, 1985兲 with
a0 ⳱ ␹ ,

冉 冊
⌿ 共z Ⳮ ⌬z,:兲 ⳱ eⳲi⌬z␻ q共p,x兲⌿ 共z,:兲, 共7兲
2␴ 2␥ 20␦
a1 ⳱ 1Ⳮ ,
where the superscript sign corresponds to backward 共⫺兲 and forward ␥ 20 ␥ 20 ⳮ 1
共⫹兲 propagation. Extrapolating the wavefield using equation 7 is ex-
pensive 共Holberg, 1988兲. More efficient one-way propagators can be
constructed by an approximation of the thin-slab propagator given in
a2 ⳱
ⳮ4␴
␥ 20共␥ 20 ⳮ 1兲
共␦ ⳮ ␴ 兲 1 Ⳮ 冉 2␥ 20␦
␥ 20 ⳮ 1
冊 . 共11兲

equation 7.
We will focus on an approximation of the thin-slab propagator
similar to the FFD approach, as described by Ristow and Rühl
共1994兲, where the propagator is separated into a phase-shift propaga-
tor in a background medium and a spatial finite-difference correction
accounting for the varying model components.
By introducing a constant background medium for each slab in a
VTI medium, described by the parameters ␣ 00, ␤ 00, ␧0, and ␦ 0, we can
The higher-order coefficients of the square-root expansion can be
found in Ursin and Stovas 共2006, equation B-9兲. To express the verti-
cal slowness for qP-waves on polynomial form as given in equation
9 共and hence being able to separate the spatial and wave-number de-
pendencies兲, we take the square-root of equation 10.
By a Taylor expansion of the square root of equation 10 around p
⳱ 0, we obtain

冉 冊
represent the vertical slowness as 1
q␣ ⳱ 1 ⳮ 兺 ã j共p␣ 0兲2jⳭ2 共12兲
␣0 jⱖ0
q␣ ,␤ 共p,x兲 ⳱ q␣0 ,␤ 共p兲 Ⳮ ⌬q␣ ,␤ 共p,x兲, 共8兲
where the first three coefficients ã j in terms of a j are
where q0␣ ,␤ 共p兲 denotes the vertical slowness in the background medi-
1
um and ⌬q␣ ,␤ 共p,x兲 denotes the phase-correcting term. Within a ã0 ⳱ a0,
depth slab, each background medium parameter can be chosen as the 2
minimum within each slab. 1 1
The wavefield is globally propagated in the background medium ã1 ⳱ a1 Ⳮ a20, 共13兲
in vertical slowness domain using q0␣ ,␤ 共p兲, and locally corrected for
2 8
the vertical-slowness perturbations with an approximation of the 1 1 1
phase-correcting term ⌬q␣ ,␤ 共p,x兲 through a finite-difference scheme ã2 ⳱ a2 Ⳮ a0a1 Ⳮ a30,
2 4 16
in space. The vertical slowness is independent of lateral position in
the background medium; hence, we can accurately propagate the thus q␣ is expressed in the desired form as represented in equation 9.
wavefield in the background medium utilizing a phase-shift operator For qSV-waves, we find a series expansion of the squared vertical
using the exact expression in equation 6. slowness as:
When we consider the phase-correcting term, we notice that it is a

冉 冊
nonlinear function of both lateral position x and horizontal-slowness 1
p; hence, we need to decouple the spatial and horizontal-slowness q␤2 ⳱ 1 ⳮ 兺 c j共p␤ 0兲2jⳭ2 , 共14兲
dependency. Thus, for the phase-correcting term, we represent the ␤ 20 jⱖ0
vertical-slowness in the following form:
where

q␣ ,␤ ⳱ 兺 k␣j ,␤ 共x兲ᐉ␣j ,␤ 共p兲. 共9兲 c0 ⳱ 1 Ⳮ 2␴ ,

jⱖ0
c j ⳱ ⳮa j␥ 2j
0, j ⱖ 1. 共15兲
 S240 Pedersen et al.

By a similar approach as for the slowness approximation of SIMPLIFIED SLOWNESS EXPRESSIONS

qP-waves, we find a slowness approximation for qSV-waves given FOR QP-WAVES
by
For conventional towed streamer data, only qP-waves are record-
ed. In this case, it is convenient to consider simplified expressions of
1
冉
1 ⳮ 兺 c̃ j共p␤ 0兲2jⳭ2 冊
Downloaded 10/14/14 to 129.241.69.108. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

the vertical slowness for qP-waves where the qSV-wave velocity is

q␤ ⳱ 共16兲
␤0 jⱖ0 disregarded, i.e., a quasi-acoustic approximation 共Alkhalifah,
1998兲. In the quasi-acoustic approximation, we assume that ␥ 20
1,
and a simplified slowness expression for qP waves can be provided
where the first three coefficients c̃ j in terms of c j are by 共Alkhalifah, 1998; Pedersen et al., 2007兲
1

冑
c̃0 ⳱ c0,
2 1 1 ⳮ 共1 Ⳮ 2␧兲共p␣ 0兲2
q␣s ⳱ , 共22兲
1 1 ␣0 1 ⳮ 2␨ 共p␣ 0兲2
c̃1 ⳱ c1 Ⳮ c20,
2 8
where the superscript s denotes a quasi-acoustic approximation.
1 1 1 Equation 22 can be found by a continuous-fraction approximation
c̃2 ⳱ c2 Ⳮ c0c1 Ⳮ c30 . 共17兲 that matches the Taylor expansion of q2␣ given in equation 10 up to
2 4 16
共p␣ 兲6 and setting ␥ 20
1.
From the quasi-acoustic approximation and equations 11, we find
The representation of the qP- and qSV-slowness can now be cascad-
ed by continued-fraction approximations. In equations 18 and 19, by as0 ⳱ ␹ ,
cascading the expression in 12, we find that the qP slowness can be
as1 ⳱ 2␹ ␨ ,
approximated by
as2 ⳱ 4␨ 2␹ , 共23兲

q␣ ⳱
1
␣0
冉
1 Ⳮ ␬ 2␣ p2 Ⳮ
␬ 1␣ p2
1 ⳮ ␬ 0␣ p2
冊 , 共18兲 and the Taylor-expanded vertical-slowness can be approximated by
the truncated series

where

␬ 0␣ ⳱
ã2
␣ 0,
q␣s ⳱
1
␣0
冉 j⳱0
2
1 ⳮ 兺 asj 共p␣ 0兲2jⳭ2 . 冊 共24兲
ã1
ã21 By substituting the simplified parameters asj into equations 19 we
␬ 1␣ ⳱ ␣ 20, find
ã2

冉 冊
1 1
ã21 2␨ 2 Ⳮ ␨ ␹ Ⳮ ␹ 2
␬ 2␣ ⳱ ã0 ⳮ ␣0. 共19兲 2 16
ã2 ␬ 0␣ ,s ⳱ ␣ 0,
1
␨Ⳮ ␹
8
Similar for the qSV-waves, by cascading the slowness approxima-
tion for qSV-waves in equation 16, we find 1 1
␨ 2␹ 2 Ⳮ ␨ ␹ 3 Ⳮ
4 64 2
␬ 1␣ ,s ⳱ ␣ 0,

冉 冊
1 1
1 ␬ 1␤ p2 2 ␨ 2␹ Ⳮ ␨ ␹ 2 Ⳮ ␹ 3
q␤ ⳱ 1 Ⳮ ␬ 2␤ p2 Ⳮ , 共20兲 2 16
␤0 1 ⳮ ␬ 0␤ p2

冢 冣
1 1
␨ 2␹ 2 Ⳮ ␨ ␹ 3 Ⳮ
1 4 64
where ␬ 2␣ ,s ⳱ ␹ⳮ ␣0. 共25兲
2 1 2 1 3
c̃2 2␨ ␹ Ⳮ ␨ ␹ Ⳮ ␹
2
␬ 0␤ ⳱ ␤ 0, 2 16
c̃1
c̃21 Thus, a simplified cascaded qP-slowness expression can be found by
␬ 1␤ ⳱ ␤ 20, inserting the simplified coefficients ␬ i␣ ,s into equation 18 that is
c̃2

冉
␬ 2␤ ⳱ c̃0 ⳮ
c̃21
c̃2
冊 ␤ 0. 共21兲 q␣s ⳱
1
␣0
冉
1 Ⳮ ␬ 2␣ ,s p2 Ⳮ
␬ 1␣ ,s p2
1 ⳮ ␬ 0␣ ,s p2
冊
. 共26兲
 One-way VTI propagator S241

ACCURACY OF SLOWNESS EXPRESSIONS ure 6a and b for model 1 and model 2, respectively. The results are
similar to the qP approximations.
We compare the accuracy of the Taylor-expanded slowness ex-
pressions in equations 12 and 16 and the cascaded slowness expres-
sions in equations 18 and 20 for qP and qSV waves, respectively. The
ONE-WAY PROPAGATORS
Downloaded 10/14/14 to 129.241.69.108. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

performance of the slowness approximations is illustrated using two

models, where the Taylor-expanded series are truncated at j ⳱ 2. We Introducing a laterally invariant background medium allows us to
also show the accuracy of the quasi-acoustic qP slowness expression define a vertical slowness perturbation ⌬q␣ using equation 12. Fur-
given by equation 24 on polynomial form and equation 26 on cas- ther, by cascading ⌬q␣ by a continued fraction approximation, we
caded form.
The medium parameters in model 1 is defined by ␣ 0 ⳱ 2.00 km/ s,
␤ 0 ⳱ 1.00 km/ s, ␧ ⳱ 0.10 and ␦ ⳱ 0.05 and the medium parameters a) b)
0.5 1.0
in model 2 is defined by ␣ 0 ⳱ 2.00 km/ s, ␤ 0 ⳱ 1.00 km/ s, ␧ ⳱ 0.10

qα (ms/m)

qβ (ms/m)
0.4 0.8
and ␦ ⳱ 0.15. In Figure 1a and b, we plot the derived slowness 0.3 0.6
curves for qP- and qSV-waves in model 1, respectively. The qP ap- 0.2 0.4
proximations is accurate up to about p ⳱ 0.4 ms/ m and the qSV ap- 0.1 0.2
proximations up to about p ⳱ 0.7 ms/ m. For both wavemodes, the 0.0 0.0
0 0.1 0.2 0.3 0.4 0.5 0 0.2 0.4 0.6 0.8 1
cascaded approximation is more accurate than the Taylor-expanded
p(ms/m) p(ms/m)
approximation. Similar results are observed for model 2, as illustrat-
ed in Figure 2a and b. Figure 2. Slowness surfaces for ␣ 0 ⳱ 2.0 km/ s, ␤ 0 ⳱ 1.0 km/ s, ␧
The derived quasi-acoustic slowness curves for models 1 and 2 ⳱ 0.10, and ␦ ⳱ 0.15. 共a兲 The exact qP slowness curve in solid line;
are shown in Figure 3a and b. The accuracy of the quasi-acoustic ap- the Taylor-expanded qP slowness curve in dotted line; and the cas-
caded qP slowness curve in dashed line. 共b兲 The exact qSV slowness
proximations are similar to that of the qPⳮ qP approximations. curve in solid line; the Taylor-expanded qSV slowness curve in dot-
To quantify the accuracy, it is useful to show the relative error of ted line; and the cascaded qSV slowness curve in dashed line.
the group-velocity as a function of group-angle corresponding to the
exact and approximated dispersion relations. Berryman 共1979兲, for
example, provides a description of how the group-velocity and a) b)
0.5 0.5
group-angle can be found.
qα (ms/m)

qα (ms/m)
0.4 0.4
Figure 4 shows the relative error in group-velocity as a function of 0.3 0.3
group-angle for model 1, where the dashed and solid line corre- 0.2 0.2
sponds to the Taylor-expanded and the cascaded approximation, re- 0.1 0.1
spectively. Figure 4a shows the qP and Figure 4b shows the qSV ap- 0.0 0.0
proximations. If one allows the relative error to be less than 1%, the 0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5
p(ms/m) p(ms/m)
cascaded qP and qSV approximations are valid up to about 60° in
model 1. In model 2, the cascaded qP approximation has less than
1% relative error up to about 70° while the cascaded qSV approxi- Figure 3. 共a兲 Slowness surfaces for ␣ 0 ⳱ 2.0 km/ s, ␤ 0 ⳱ 1.0 km/ s,
␧ ⳱ 0.10, and ␦ ⳱ 0.05. 共b兲 Slowness surfaces for ␣ 0 ⳱ 2.0 km/ s,
mation is valid up to about 40° and is shown in Figure 5. The group- ␤ 0 ⳱ 1.0 km/ s, ␧ ⳱ 0.10, and ␦ ⳱ 0.15. The exact qP slowness
velocity errors for the quasi-acoustic approximations are shown in curve in solid line; the quasi-acoustic Taylor-expanded qP slowness
Fig- curve in dotted line; and the quasi-acoustic cascaded qP slowness
curve in dashed line.

a) b)
a) b) 6 6
Group-velocity error (%)
Group-velocity error (%)

0.5 1.0
5 5
qα (ms/m)

qβ (ms/m)

0.4 0.8
4 4
0.3 0.6
3 3
0.2 0.4
2 2
0.1 0.2
1 1
0.0 0.0 0 0
0 0.1 0.2 0.3 0.4 0.5 0 0.2 0.4 0.6 0.8 1
0 20 40 60 80 0 20 40 60 80
p(ms/m) p(ms/m)
Group angle (deg) Group angle (deg)

Figure 1. Slowness surfaces for ␣ 0 ⳱ 2.0 km/ s, ␤ 0 ⳱ 1.0 km/ s, ␧

⳱ 0.10, and ␦ ⳱ 0.05. 共a兲 The exact qP slowness curve in solid line;
the Taylor-expanded qP slowness curve in dotted line; and the cas-
caded qP slowness curve in dashed line. 共b兲 The exact qSV slowness
curve in solid line; the Taylor-expanded qSV slowness curve in dot-
ted line; and the cascaded qSV slowness curve in dashed line.
Figure 4. Relative group-velocity error as a function of group-angle
for ␣ 0 ⳱ 2.0 km/ s, ␤ 0 ⳱ 1.0 km/ s, ␧ ⳱ 0.10, and ␦ ⳱ 0.05. The
Taylor-expanded group-velocity in dashed line and the cascaded
group velocity in solid line corresponding to 共a兲 the qP and 共b兲 the
qSV slowness approximations.
 S242
find from equation 7 that the resulting one-way propagator for
qP-waves can be represented by
0 in space domain by a finite-difference approach in a similar fashion
ei␻ ⌬zq␣ 共p,x兲 ⬇ ei␻ ⌬zq␣ 共p兲ei␻ ⌬z⌬q␣ 共p,x兲 .
Downloaded 10/14/14 to 129.241.69.108. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
In a similar fashion, we find that the qSV propagator can be repre-
sented by
0 position 共Amundsen and Reitan, 1995; Osen et al., 1996兲.
ei␻ ⌬zq␤ 共p,x兲 ⬇ ei␻ ⌬zq␤ 共p兲ei␻ ⌬z⌬q␤ 共p,x兲 .
See the Appendix for a detailed derivation of ⌬q␣ and ⌬q␤ .
a) b)
6 6
Group-velocity error (%)
Group-velocity error (%)
5 5
4 4
3 3
2 2
1 1
0 0
0 20 40 60 80 0 20 40 60 80
Group angle (deg)
Figure 5. Relative group-velocity error as a function of group-angle
for ␣ 0 ⳱ 2.0 km/ s, ␤ 0 ⳱ 1.0 km/ s, ␧ ⳱ 0.10, and ␦ ⳱ 0.15. The
Taylor-expanded group-velocity in dashed line and the cascaded
group velocity in solid line corresponding to 共a兲 the qP and 共b兲 the
qSV slowness approximations.
a) b)
6 6
Group-velocity error (%)
Group-velocity error (%)
5 5
4 4
3 3
2 2
1 1
0 0
0 20 40 60 80 0 20 40 60 80
Group angle (deg)
Figure 6. Group-velocity error as a function of group-angle for the
quasi-acoustic approximations. The Taylor-expanded group-veloci-
ty in dashed line and the cascaded group velocity in solid line for 共a兲
␣ 0 ⳱ 2.0 km/ s, ␤ 0 ⳱ 1.0 km/ s, ␧ ⳱ 0.10, and ␦ ⳱ 0.05 and 共b兲 ␣ 0
⳱ 2.0 km/ s, ␤ 0 ⳱ 1.0 km/ s, ␧ ⳱ 0.10, and ␦ ⳱ 0.15.
a) Distance (km) b)
–2 –1 0 1 2
Depth (km)
Depth (km)
1 1
2 2
3 3
Figure 7. Impulse responses of qPⳮ qSV 共a兲 and qPⳮ qP 共b兲 waves
with analytically computed impulse response overlayed. The medi-
um is given by ␣ 0 ⳱ 2000.0 m / s; ␤ 0 ⳱ 1000.0 m / s; ␦ ⳱ 0.0; and ␧
⳱ 0.1 with a background medium given by ␣ 00 ⳱ 1000.0 m / s; ␤ 00
⳱ 750.0 m / s; ␦ 0 ⳱ 0.0; and ␧0 ⳱ 0.1.
Pedersen et al.
For each frequency, the phase-shift in the background medium is
applied to the wavefield in slowness domain, while the phase-cor-
recting term handling the vertical-slowness perturbation is applied
共27兲 as described by Ristow and Rühl 共1994兲.
In an OBS experiment with a downgoing pressure source field
共DqP兲 in a fluid and by recording pressure and particle velocity at the
seafloor, we can find the upgoing pressure wavefield 共UqP兲 and the
upgoing shear wavefield 共UqSV兲 at the seafloor by wavefield decom-
共28兲 For pre-stack migration, a C-wave subsurface image, denoted
IqP,qSV, can be constructed using the following imaging condition
IqP,qSV共z,:兲 ⳱ 兺 兺 UqSV
k 共z,:兲Dk 共z,:兲
qP * 共29兲
k ␻
and a qPⳮ qP subsurface image, denoted IqP,qP, can be constructed
using the following imaging condition
IqP,qP共z,:兲 ⳱ 兺 兺 UqP
k 共z,:兲Dk 共z,:兲 ,
qP * 共30兲
k ␻
where the superscript * denotes complex conjugate 共Claerbout,
1971兲 and
DqP共z Ⳮ ⌬z,:兲 ⳱ ei␻ ⌬zq␣ 共p,x兲DqP共z,:兲
Group angle (deg)
UqP共z Ⳮ ⌬z,:兲 ⳱ eⳮi␻ ⌬zq␣ 共p,x兲UqP共z,:兲
UqSV共z Ⳮ ⌬z,:兲 ⳱ eⳮi␻ ⌬zq␤ 共p,x兲UqSV共z,:兲. 共31兲
The subscript k denotes shot number for common-shot migration
and receiver number for common-receiver migration.
NUMERICAL RESULTS
In this section we use the derived migration schemes to produce
impulse responses for both compressional and converted waves.
These tests are useful for illustrating the capabilities to image steep-
dips with the proposed scheme in media with lateral medium pertur-
bations. Further, we test the ability of the derived migration scheme
to handle a complex subsurface with a relative high degree of medi-
um perturbations on a 2D synthetic data example.
Group angle (deg)
Impulse responses
To demonstrate the accuracy of the derived one-way propagators,
we first produce a series of impulse response tests. The impulse re-
sponse test involves migrating a single input seismic trace in a given
medium. The input trace contains a single spike located at time 2.0 s.
First, we consider a medium given by ␣ 0 ⳱ 2000.0 m / s, ␤ 0
Distance (km) ⳱ 1000.0 m / s, ␦ ⳱ 0.0 and ␧ ⳱ 0.1. By introducing a background
–2 –1 0 1 2 medium given by ␣ 00 ⳱ 1000.0 m / s, ␤ 00 ⳱ 750.0 m / s, ␦ 0 ⳱ 0.0, and
␧0 ⳱ 0.1 we can illustrate the ability of the derived one-way propaga-
tors to handle the resulting lateral medium variations. The impulse
response for the C-wave migration is shown in Figure 7a. By inspec-
tion, we see that the proposed one-way propagators in this example
are accurate up to about 50°. For larger angles, the error introduced
by the medium perturbations cause the impulse response to be mis-
positioned. This is mainly due to the inaccuracy of the qSV slowness
approximation for large angles, which diverges from the exact qSV
slowness.
 One-way VTI propagator S243

The impulse response for the qPⳮ qP migration is shown in Fig- a) Distance (km) b) Distance (km)
ure 7b. We see that the proposed propagator is accurate up to about –2 –1 0 1 2 –2 –1 0 1 2
70°.

Depth (km)

Depth (km)
1 1
In our next example, we change the anisotropy parameters such
that ␦ ⳱ 0.05 and ␧ ⳱ 0.10 with and ␦ 0 ⳱ 0.05 and ␧0 ⳱ 0.10. The 2 2
Downloaded 10/14/14 to 129.241.69.108. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

impulse responses are shown in Figure 8. The accuracy of the 3 3

C-wave and qPⳮ qP migration is similar for both wavemodes as in
the previous example. Further, we change the anisotropy parameters Figure 9. Impulse responses of qPⳮ qSV 共a兲 and qPⳮ qP 共b兲 waves
with analytically computed impulse response overlayed. The medi-
such that ␦ ⳱ 0.05 and ␧ ⳱ 0.25 with and ␦ 0 ⳱ 0.05 and ␧0 ⳱ 0.25. um is given by ␣ 0 ⳱ 2000.0 m / s; ␤ 0 ⳱ 1000.0 m / s; ␦ ⳱ 0.05; and
The impulse responses are shown in Figure 9. The qPⳮ qP impulse ␧ ⳱ 0.25 with a background medium given by ␣ 00 ⳱ 1000.0 m / s; ␤ 00
response is reasonably accurate as the previous examples. The ⳱ 750.0 m / s; ␦ 0 ⳱ 0.05; and ␧0 ⳱ 0.25.
C-wave impulse response shows that the C-wave migration is accu-
rate up to about 45°. For angles above 45°, the wavefront is misposi-
tioned.
In general, the accuracy of the derived one-way propagators for
high dips depends on the variation of the medium properties. That is,
for high dips the propagators are more accurate for small medium a) Distance (km)
variations than for larger medium variations. 1 2 3 4 5

3000

Depth (km)
2D Synthetic data example 1 2500
Next, the accuracy of the kinematics of the proposed one-way 2000
propagators is demonstrated on a 2D synthetic data set.
A synthetic common-shot data set was produced using a Born- 2 1500
(m/s)
Kirchhoff-Helmholtz modeling scheme 共Ursin and Tygel, 1997;
Sollid and Ursin, 2003兲 both for qPⳮ qP and qPⳮ qSV waves. The b) Distance (km)
data set model parameters are shown in Figure 10. The model con-
sists of 5 reflecting interfaces and the response from the sea floor is 1 2 3 4 5
not modeled. The sources are located at the sea surface, while the re- 1500
Depth (km)

ceiver cable is located at the seafloor at depth 100 m. Each shot con-
tains 161 receivers with separated by 20 m, and the shots are distrib- 1 1000
uted with a distance of 20 m. Figure 11 depicts selected shot-gathers
500
that are input to migration.
The migrated sections are shown in Figure 12, where Figure 12a 2 0
depicts the migrated qPⳮ qP data and Figure 12b depicts the migrat- (m/s)
ed qPⳮ qSV data. Above the sea floor, ␤ 0 ⳱ ␦ ⳱ ␧ ⳱ 0, thus the
downgoing source pressure wavefield UqP is propagated using an c) Distance (km)
isotropic phase-shift down to the seafloor. From the results, we see 1 2 3 4 5
that both the C-wave and qPⳮ qP-wave migration provide accurate 0.20
results. The difference between the migrated sections is not very dis-
Depth (km)

0.15
tinct, and the reflector interpretation would most likely coincide for
1 0.10
both sections. We notice that the migrated qPⳮ qSV image has
higher resolution than the migrated qPⳮ qP image because the 0.05
shear-wave velocities are slower than the pressure velocities.
2 0.00

d) Distance (km)
a) Distance (km) b) Distance (km) 1 2 3 4 5
–2 –1 0 1 2 –2 –1 0 1 2
0.06
Depth (km)

Depth (km)

Depth (km)

1 1 0.04
2 2 1 0.02
3 3 0.00
–0.02
Figure 8. Impulse responses of qPⳮ qSV 共a兲 and qPⳮ qP 共b兲 waves 2
with analytically computed impulse response overlayed. The medi-
um is given by ␣ 0 ⳱ 2000.0 m / s; ␤ 0 ⳱ 1000.0 m / s; ␦ ⳱ 0.05; and Figure 10. Model parameters for the synthetic data example. 共a兲 Ver-
␧ ⳱ 0.1 with a background medium given by ␣ 00 ⳱ 1000.0 m / s; ␤ 00 tical qP-wave velocity ␣ 0; 共b兲 vertical qSV-wave velocity ␤ 0; 共c兲
⳱ 750.0 m / s; ␦ 0 ⳱ 0.05; and ␧0 ⳱ 0.1. Thomsen parameter ␧; and 共d兲 Thomsen parameter ␦ .
 S244 Pedersen et al.

3D REAL OCEAN-BOTTOM DATA EXAMPLE data set contains 12 cables of 6 km length, each with 240 receivers.
The shots are separated by 50-by-50 m. In Figure 15, selected shot-
We apply the derived one-way propagators on a 3D field OBS data gathers are shown, where Figure 14a shows a qPⳮ qP gather and
set. The OBS data set was acquired in 2002 in the central North Sea
over the Volve field. The field is located in the Sleipner area in the
a) Crossline number b) Crossline number
southern part of the Viking Graben.
Downloaded 10/14/14 to 129.241.69.108. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

1 200 400 1 200 400

A subset of the entire 3D survey was extracted for input to a com-
mon-receiver migration scheme. A 2D section of the derived model 3500
4500
parameters 共Szydlik et al., 2007兲 are shown in Figure 13, where Fig- 3000
1 1
ure 13a shows the vertical qP-wave velocity; Figure 13b shows the 4000

vertical qSV-wave velocity; Figure 13c shows the Thomsen parame- 2500

Depth (km)
Depth (km)
3500
ter ␦ ; and Figure 13d shows the Thomsen parameter ␧. The extracted 2 2
2000
3000

3 3 1500
2500
1000
2000
4 4
a) b) 500
Offset (km) Offset (km) 1500
(m/s) (m/s)
–1 0 1 –1 0 1
c) Crossline number d) Crossline number
0.5 1 200 400 1 200 400
0.18
0.5 0.09
1.0 0.16
0.08
1 0.14 1
1.5 0.07
1.0
Time (s)

Time (s)

0.12
0.06

Depth (km)
Depth (km)

2.0
2 0.10 2
0.05
1.5 2.5 0.08
0.04
3 0.06 3 0.03
3.0
0.04 0.02
2.0
3.5 4 0.02 4 0.01
0.00 0.00

Figure 11. Selected synthetic shot-gathers — for a source at surface Figure 13. Model parameters for the real data example from offshore
location 2500 m — which are input to migration. 共a兲 qPⳮ qP shot Norway at inline 68. 共a兲 Vertical qP-wave velocity ␣ 0; 共b兲 vertical
gather and 共b兲 qPⳮ qSV shot gather. qSV-wave velocity ␤ 0; 共c兲 Thomsen parameter ␧; and 共d兲 Thomsen
parameter ␦ .

a) Distance (km)
1 2 3 4 5 a) Offset (km) b) Offset (km)
–2 0 2 –2 0 2
Depth (km)

2
2

2
Time (s)

Time (s)

4
b) Distance (km) 4
1 2 3 4 5

6
Depth (km)

6
1
8

2 Figure 14. Selected shot gathers which are sorted into common-re-
ceiver gathers and input to migration for the real data example from
Figure 12. Migrated section of the synthetic data example. 共a兲 qP offshore Norway. 共a兲 Derived qPⳮ qP shot-gathers and 共b兲 derived
ⳮ qP-stack and 共b兲 qPⳮ qSV-stack. qPⳮ qSV shot-gather.
 One-way VTI propagator S245

Figure 14b shows a qPⳮ qSV gather. The data are migrated in a Crossline number
common-receiver migration scheme. 150 200 250 300 350 400
From the stacked migrated data sets IqPⳮqP and IqPⳮqSV we extract
two sections; one along 共inline 68兲 and one across 共crossline 232兲 the
OBS cables indicated by the dashed lines in Figure 15. The sections
Downloaded 10/14/14 to 129.241.69.108. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

IqPⳮqP and IqPⳮqSV for inline 68 are shown in Figures 16 and 17, re-
spectively. The sections IqPⳮqP and IqPⳮqSV for crossline 232 are
1
shown in Figures 18 and 19, respectively. No postprocessing has
been applied after imaging except a gain in depth for displaying pur-
poses.
In general, the migrated images IqPⳮqP and IqPⳮqSV show good
structural focusing. The two images correlate well in depth. We no-
tice that the migrated IqPⳮqP image show better reflector continuity in 2

Depth (km)
the deeper part than the migrated IqPⳮqSV image and thereby seems to
be better focused. Some differences are found between the distinc-
tion of some of the reflectors in the deeper part below about 3 km.
These differences are most probably due to differences in the reflec-
tivity for the converted waves. 3

493
273 4
232
68
1 1
Inline Crossline
Figure 17. Migrated qPⳮ qSV section from inline 68 of the real data
Figure 15. Representation of the coverage used in migration of the example from offshore Norway.
3D field data example. Extracted IqPⳮqP and IqPⳮqSV sections are indi-
cated by dashed lines and displayed in Figures 16–19.

Inline number
Crossline number
50 100 150 200
150 200 250 300 350 400

1
1

2
Depth (km)

2
Depth (km)

3 3

4 4

Figure 16. Migrated qPⳮ qP section from inline 68 of the real data Figure 18. Migrated qPⳮ qP section from crossline 232 of the real
example from offshore Norway. data example from offshore Norway.
 S246 Pedersen et al.

Inline number structural definitions, we also see good correlation in depth in the im-
50 100 150 200 ages.

ACKNOWLEDGMENTS
Downloaded 10/14/14 to 129.241.69.108. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

This research was supported by the Norwegian Research Council

via the ROSE project. We would like to thank Statoil ASA for per-
1
mission to publish and providing the field data set. Ørjan Pedersen
would like to thank Joachim Mispel for help preparing the field data.
Bjørn Ursin has received financial support from Statoil ASA through
the VISTA project. Finally, we thank the associate editor Yike Liu,
Bertram Nolte, and three anonymous reviewers at GEOPHYSICS
2 for their help in improving this paper.
Depth (km)

APPENDIX A

FINITE-DIFFERENCE CORRECTIONS FOR

3
ONE-WAY THIN-SLAB PROPAGATORS
Using equation 12, we can represent the phase-correcting term
⌬q␣ as

1
4 ⌬q␣ ⳱ q␣ ⳮ q␣0 ⳱ 共1 ⳮ r兲 ⳮ 兺 共ã j ⳮ ã0j r2jⳭ2兲共␣ 0 p兲2jⳭ2
␣0 jⱖ0

共A-1兲
Figure 19. Migrated qPⳮ qSV section from crossline 232 of the real where r ⳱ ␣ 00 / ␣ 0 and ã0j are the coefficients in equations 13 defined
data example from offshore Norway. by the background medium. By cascading ⌬q␣ as given in equation
18, we find that
CONCLUSIONS
1 ⌬␬ 1␣ p2
⌬q␣ ⬇ 共1 ⳮ r兲 Ⳮ ⌬␬ 2␣ p2 Ⳮ , 共A-2兲
We have developed a prestack migration scheme for compres- ␣0 1 ⳮ ⌬␬ 0␣ p2
sional and converted waves using one-way propagators. The method
builds on an FFD depth extrapolation scheme for which we derive where
approximations of the vertical qP and qSV slowness as a function of
⌬s21
horizontal slowness p. We also derive a qP approximation by using ⌬␬ 0␣ ⳱ ␣0
the quasi-acoustic approximation, which can be used in convention- ⌬s2
al processing where the shear-wave velocity is not known. We notice
⌬s2
that the quasi-acoustic approximations provides similar accuracy as ⌬␬ 1␣ ⳱ ␣0
the qP approximations. This is expected, since the impact of ␤ 0 on a ⌬s1
propagating qP wave is small.
By analyzing the group-velocity as a function of group-angle cor-
responding to the derived approximations, we show that the approxi-
冉
⌬␬ 2␣ ⳱ ⌬s0 ⳮ
⌬s2
冊
⌬s21
␣0 共A-3兲

mations used in the compressional and C-wave depth-extrapolation and

scheme are accurate for high angles using small anisotropy parame-
ters and weak nonhyperbolicity. ⌬s0 ⳱ ã0 ⳮ ã00r
Impulse responses demonstrate the accuracy of the derived propa-
⌬s1 ⳱ ã1 ⳮ ã01r3
gators through good dip response and accurate kinematic behavior
for weak anisotropy. We show that for higher values of the nonhyper- ⌬s2 ⳱ ã2 ⳮ ã02r5 . 共A-4兲
bolicity parameter 共␧ ⳮ ␦ 兲 / 共1 Ⳮ 2␦ 兲, the accuracy of the compres-
sional propagators are still valid for high angles of propagation, Similarly, from equation 16 we find
while the C-wave propagators become more inaccurate for higher
1
angles. ⌬q␤ ⳱ q␤ ⳮ q␤0 ⳱ 共1 ⳮ t兲 ⳮ 兺 共c̃ j ⳮ c̃0j t2jⳭ2兲共␤ 0 p兲2jⳭ2
The 2D synthetic data example illustrates the methods ability to ␤0 jⱖ0
handle models with a realistic degree of laterally varying medium
共A-5兲
parameters. The data example also shows consistency between the
compressional and C-wave images. The 3D field data example where t ⳱ ␤ / ␤ 0 and c̃ are the coefficients in equations 17 defined
0
0
0
j
shows that the method is applicable to real data. Even though the by the background medium. By cascading ⌬q␤ as given in equation
compressional and C-wave images show differences in focusing and 20, we find that
 One-way VTI propagator S247

1 ⌬␬ 1␤ p2 REFERENCES
⌬q␤ ⬇ 共1 ⳮ t兲 Ⳮ ⌬␬ 2␤ p2 Ⳮ , 共A-6兲
␤0 1 ⳮ ⌬␬ 0␤ p2 Alkhalifah, T., 1998, Acoustic approximations for processing in transversely
isotropic media: Geophysics, 63, 623–631, doi: 10.1190/1.1444361.
——–, 2000a, An acoustic wave equation for anisotropic media: Geophysics,
where 65, 1239–1250, doi: 10.1190/1.1444815.
Downloaded 10/14/14 to 129.241.69.108. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

——–, 2000b, Prestack phase-shift migration of separate offsets: Geophys-

⌬u21 ics, 65, 1179–1194, doi: 10.1190/1.1444811.
⌬␬ 0␤ ⳱ ␤0 Alkhalifah, T., and K. Larner, 1994, Migration error in transversely isotropic
⌬u2 media: Geophysics, 59, 1405–1418, doi: 10.1190/1.1443698.
Amundsen, L., and A. Reitan, 1995, Decomposition of multicomponent sea-
⌬u2 2 floor data into upgoing and downgoing P- and S-waves: Geophysics, 60,
⌬␬ 1␤ ⳱ ␤ 563–572, doi: 10.1190/1.1443794.
⌬u1 0 Barkved, O., B. Bartman, B. Compani, J. Gaiser, R. VanDok, T. Johns, P.

冉 冊
Kristiansen, T. Probert, and M. Thompson, 2004, The many facets of mul-
⌬u21 ticomponent seismic data: Oilfield Review, Summer, 42–56.
⌬␬ 2␤ ⳱ ⌬u0 ⳮ ␤0 共A-7兲 Berryman, J. G., 1979, Long-wave elastic anisotropy in transversely isotro-
⌬u2 pic media: Geophysics, 44, 896–917, doi: 10.1190/1.1440984.
Biondi, B., 2006, 3D Seismic imaging: Investigations in Geophysics Series
No. 14, SEG.
and Bleistein, N., 1987, On the imaging of reflectors in the earth: Geophysics, 52,
931–942, doi: 10.1190/1.1442363.
⌬u0 ⳱ ũ0 ⳮ ũ00t Červený, V., 2001, Seismic ray theory: Cambridge University Press.
Claerbout, J. F., 1971, Toward a unified theory of reflector mapping: Geo-
⌬u1 ⳱ ũ1 ⳮ ũ01t3 physics, 36, 467–481.
——–, 1985, Imaging the earth’s interior: Blackwell Scientific Publications.
⌬u2 ⳱ ũ2 ⳮ ũ02t5 . 共A-8兲 Gazdag, J., 1978, Wave equation migration with the phase-shift method:
Geophysics, 43, 1342–1351.
Gazdag, J., and P. Sguazzero, 1984, Migration of seismic data by phase shift
plus interpolation: Geophysics, 49, 124–131.
Finally, we notice that in the isotropic case, ␨ ⳱ ␴ ⳱ 0 and ␹ ⳱ 1, Granli, J. R., B. Arntsen, A. Sollid, and E. Hilde, 1999, Imaging through gas-
thus a0 ⳱ 1 and a1 ⳱ a2 ⳱ 0; hence, filled sediments using marine shear-wave data: Geophysics, 64, 668–677.
Han, Q., and R.-S. Wu, 2005, A one-way dual-domain propagator for scalar
qP-waves in VTI media: Geophysics, 70, no. 2, D9–D17.
Hokstad, K., 2000, Multicomponent Kirchhoff migration: Geophysics, 65,
1 861–873.
ã0 ⳱ , Holberg, O., 1988, Towards optimum one-way wave propagation: Geophysi-
2 cal Prospecting, 36, 99–114.
Larner, K. L., and J. K. Cohen, 1993, Migration error in transversely isotro-
1 pic media with linear velocity variation in depth: Geophysics, 58,
ã1 ⳱ , 1454–1467.
8 MacLeod, M., R. Hanson, M. Hadley, K. Reynolds, D. Lumley, S. McHugo,
and T. Probert, 1999, The Alba field OBC seismic survey: 69th Annual In-
1
ã2 ⳱ . 共A-9兲 ternational Meeting, SEG, Expanded Abstracts, 725–727.
Musgrave, M. J. P., 1970, Crystal acoustics: Holden-Day.
16
Nolte, B., 2005, Converted-wave migration for VTI media using Fourier fi-
nite-difference depth extrapolation: 67th Conference and Exhibition,
EAGE, Extended Abstracts, P001.
This gives Osen, A., L. Storelvmo, L. Amundsen, and A. Reitan, 1996, Decomposition
of particle velocity fields into up- and downgoing P- and S-waves: 66th
1 Annual International Meeting, SEG, Expanded Abstracts, 743–746.
⌬s0 ⳱ 共1 ⳮ r兲, Pedersen, Ø., B. Ursin, and A. Stovas, 2007, Wide-angle phase-slowness ap-
2 proximations in VTI media: Geophysics, 72, no. 4, S177–S185.
Ristow, D., and T. Rühl, 1994, Fourier finite-difference migration: Geophys-
1
⌬s1 ⳱ 共1 ⳮ r3兲, ics, 59, 1882–1893.
Rousseau, J. H. L., and M. V. de Hoop, 2001a, Modeling and imaging with
8
the scalar generalized-screen algorithms in isotropic media: Geophysics,
66, 1551–1568.
1
⌬s2 ⳱ 共1 ⳮ r5兲. 共A-10兲 ——–, 2001b, Scalar generalized-screen algorithms in transversely isotropic
media with a vertical symmetry axis: Geophysics, 66, 1538–1550.
16
Schneider, W. A., 1978, Integral formulation for migration in two and three
dimensions: Geophysics, 43, 49–76.
Further, Sollid, A., and B. Ursin, 2003, Scattering-angle migration of ocean-bottom
seismic data in weakly anisotropic media: Geophysics, 68, 641–655.
Stoffa, P. L., J. T. Fokkema, R. M. de Luna Freire, and W. P. Kessinger, 1990,
1 共1 ⳮ r3兲2
⌬␬ 0␣ ⳱ ␣ 0, Split-step Fourier migration: Geophysics, 55, 410–421.
4 1 ⳮ r5 Stolt, R. H., 1978, Migration by Fourier transform: Geophysics, 43, 23–48.
Stovas, A., and B. Ursin, 2003, Reflection and transmission responses of lay-
ered transversely isotropic viscoelastic media: Geophysical Prospecting,
1 1 ⳮ r5 2
⌬␬ 1␣ ⳱ ␣ , 51, 447–477.
2 1 ⳮ r3 0 Sun, Y., F. Qin, S. Checkles, and J. P. Leveille, 2000, 3-D prestack Kirchhoff

冉 冊
beam migration for depth imaging: Geophysics, 65, 1592–1603.
Szydlik, T., P. Smith, S. Way, L. Aamodt, and C. Friedrich, 2007, 3D PP/PS
1 1 共1 ⳮ r3兲2
⌬␬ 2␣ ⳱ 共1 ⳮ r兲 ⳮ ␣ 0, 共A-11兲 prestack depth migration on the Volve field: First Break, 25, 43–47.
2 4 1 ⳮ r5 Thomsen, L., 1986, Weak elastic anisotropy: Geophysics, 51, 1954–1966.
Thomsen, L., and J. Dellinger, 2003, On shear-wave triplication in trans-
versely isotropic media: Journal of Applied Geophysics, 54, 289–296.
Ursin, B., and A. Stovas, 2006, Traveltime approximations for a layered
which is equivalent to the isotropic third-order FFD approximation transversely isotropic medium: Geophysics, 71, no. 2, D23–D33.
derived by Ristow and Rühl 共1994, their equation A-22兲. Ursin, B., and M. Tygel, 1997, Reciprocal volume and surface scattering in-
 S248 Pedersen et al.

tegrals for anisotropic elastic media: Wave Motion, 26, 31–42.
Vestrum, R. W., D. C. Lawton, and R. Schmid, 1999, Imaging structures be-
low dipping TI media: Geophysics, 64, 1239–1246.
Wu, R.-S., and L.-J. Huang, 1992, Scattered field calculation in heteroge-
Downloaded 10/14/14 to 129.241.69.108. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
neous media using a phase-screen propagator: 62nd Annual International
Meeting, SEG, Expanded Abstracts, 1289–1292.
Xie, X.-B., and R.-S. Wu, 2005, Multicomponent prestack depth migration
using the elastic screen method: Geophysics, 70, no. 1, S30–S37.

View publication stats