Kirchhoff prestack depth migration in simple models

of various anisotropy
Václav Bucha
Department of Geophysics, Faculty of Mathematics and Physics, Charles University,
Ke Karlovu 3, 121 16 Praha 2, Czech Republic, E-mail: bucha@seis.karlov.mff.cuni.cz

Summary
We apply the Kirchhoff prestack depth migration to calculation of migrated sections
in 2-D and 3-D simple anisotropic velocity models. Each velocity model is composed
of two homogeneous layers separated by one curved interface. We use different types
of anisotropy in the upper layer: isotropy, transversely isotropic media with a horizon-
tal symmetry axis, triclinic anisotropy and monoclinic anisotropy. We test Kirchhoff
prestack depth migration in two ways: a) the anisotropy used for computation of the
recorded wave field is the same as the anisotropy used for migration, b) the anisotropy
used for computation of the recorded wave field differs from the anisotropy used for
migration.

Keywords
Kirchhoff prestack depth migration, anisotropic velocity model

1. Introduction
We approximate the ”recorded wave field” by ray-theory synthetic seismograms, and
then apply ray-based Kirchhoff prestack depth migration to calculation of migrated
sections. For the calculation of the recorded wave field, we use simple anisotropic
velocity models composed of two homogeneous layers separated by one curved interface.
The upper layer is anisotropic and velocity models differ by the anisotropy in the upper
layer. The bottom layer is isotropic. The curved interface is the same in all models.
Computation of the recorded wave field in the models is performed using the AN-
RAY software package (Gajewski & Pšenčı́k, 1990). Two-point rays are calculated for
reflected P-wave in models with isotropy (ISO), transversely isotropic media with a hor-
izontal symmetry axis (HTI), triclinic anisotropy (TA) and special case of monoclinic
anisotropy (MA) in the upper layer.
We use MODEL, CRT, FORMS and DATA packages for the Kirchhoff prestack
depth migration (Červený, Klimeš & Pšenčı́k, 1988; Bulant, 1996). The velocity models
for migration are homogeneous.
In the first test, the anisotropy in the upper layer of the velocity model used for
computation of the recorded wave field is the same as the anisotropy in the velocity
model used for each migration. The bottom layer is isotropic in all models.
We also test behaviour of 2-D Kirchhoff prestack depth migration in 2-D models in
which the anisotropy in the upper layer of the velocity model used for computation of
the recorded wave field is different from the anisotropy in the velocity model used for
migration. We compare 24% HTI anisotropy used for the recorded wave field with 15%
HTI and 3.5% HTI anisotropy used for the migration. Another test is a comparison

In: Seismic Waves in Complex 3-D Structures, Report 20 (Department of Geophysics, Charles University, Prague, 2010), pp. 35–52

35
between 24% HTI anisotropy used for the recorded wave field and two values of isotropic
velocity 2.5 km/s and 4.5 km/s used for the migration. Monoclinic anisotropy used for
the recorded wave field is compared with isotropic velocity 2.45 km/s used for the
migration.
We use 3-D Kirchhoff prestack depth migration in a simple 3-D models with triclinic
anisotropy (TA) and monoclinic anisotropy (MA). The anisotropy in the upper layer
of the velocity model used for computation of the recorded wave field is the same as
the anisotropy in the velocity model used for migration. Shot-receiver configuration
consists of 81 parallel profile lines. We also test 3-D migration with profile line step
twice enlarged (41 parallel profile lines).
Programs used for the Kirchhoff prestack depth migration were introduced on com-
pact disk SW3D-CD-9 (Bucha & Bulant, 2005). The GOCAD program has been used
for visualization of the 3-D model with curved interface, shot-receiver configurations
and computed two-point rays. The corresponding 3-D figures are screen snapshots of
limited resolution.

2. Anisotropic velocity models

The dimensions of the velocity models and measurement configurations are derived from
the Marmousi model and dataset (Versteeg & Grau, 1991). The horizontal dimension
of the model is 9.2 km and the depth is 3 km. Each velocity model is composed of
two homogeneous layers separated by one curved interface (see Figures 1, 2). Velocity
models differ by the anisotropy of the upper layer. The bottom layer is isotropic in all
models and has P-wave velocity Vp = 3.6 km/s.

0 2 4 6 distance (km) 8
0

1
depth (km)

3
Figure 1. 2-D model with a curved interface.

We also perform 3-D calculations in the model simply derived from the 2-D model by
extension in the perpendicular direction (see Figure 2).
We compute recorded wave field in models with isotropy and various types of
anisotropy in the upper layer:
a) Isotropic (ISO) upper layer with P-wave velocity Vp = 2.5 km/s.
b) Transversely isotropic media with a horizontal symmetry axis (HTI) representing
a medium with aligned thin dry cracks (Shearer & Chapman, 1989). The axis of
symmetry is parallel with the 2-D velocity model. Anisotropy of P waves is 24%.

36
Figure 2. 3-D model with a curved interface, top and bottom model planes. The horizontal dimensions
of the model are 9.2 km x 3 km, the depth is 3 km.

Reflected two-point rays are situated in the 2-D plane. Matrix of elastic parameters
is defined by a table

11.91 4.4 4.4 0. 0. 0.

 
 19.11 6.35 0. 0. 0. 
19.11 0. 0. 0. 
 
.

6.38 0. 0. 


5.1 0.
 
5.1

c) Triclinic anisotropy (TA) representing dry Vosges sandstone (Mensch & Rasolofos-
aon, 1997). Reflected two-point rays propagate in 3-D volume. Matrix of elastic
parameters is defined by a table

10.3 0.9 1.3 1.4 1.1 0.8

 
 10.6 2.1 0.2 −0.2 −0.6 
14.1 0.0 −0.5 −1.0 
 
.

5.1 0.0 0.2 


6.0 0.0
 
4.9

d) Monoclinic anisotropy (MA). Reflected two-point rays are situated in 2-D plane
x1 x3 . Matrix of elastic parameters is defined by a table

4 −2 4 0 1 0
 
 6 −2 0 0 0
6 0 1 0
 
.

2 0 0


2 0
 
2

37
We perform the migration in homogeneous models (without curved interface) with
anisotropy specified in items a) - d) and also in the following isotropic and anisotropic
models:
e) Isotropic (ISO) models with P-wave velocities Vp = 2.45 km/s and Vp = 4.5 km/s.
f) Transversely isotropic media with a horizontal symmetry axis (HTI) representing
a medium with aligned thin water-filled cracks (Shearer & Chapman, 1989). The
axis of symmetry is parallel with the 2-D velocity model. Anisotropy of P waves
is 15%. Reflected two-point rays are situated in the 2-D plane. Matrix of elastic
parameters is defined by a table

19.63 7.26 7.26 0. 0. 0.

 
 20.16 7.4 0. 0. 0. 
20.16 0. 0. 0. 
 
.

6.38 0. 0. 


3.48 0.
 
3.48

g) Transversely isotropic media with a horizontal symmetry axis (HTI) representing

again a medium with aligned thin water-filled cracks (Shearer & Chapman, 1989).
The axis of symmetry is parallel with the 2-D velocity model. Anisotropy of P
waves is 3.5%. Reflected two-point rays are situated in the 2-D plane. Matrix of
elastic parameters is defined by a table

20.04 7.41 7.41 0. 0. 0.

 
 20.22 7.46 0. 0. 0. 
20.22 0. 0. 0. 
 
.

6.38 0. 0. 


5.1 0.
 
5.1

3. Shots and receivers

The 2-D measurement configuration is derived from the Marmousi model and dataset
(Versteeg & Grau, 1991). One profile line is used for 2-D calculations. The first shot is
at 3 km, the last shot is at 8.975 km, distance between shots is 0.025 km, the depth of
shots is 0.008 km. The total number of shots is 240. The number of receivers per shot
is 96, the first receiver is at offset 2.575 km left of shot location, the last receiver is at
offset 0.2 km left of shot location, the distance between receivers is 0.025 km, the depth
of receivers is 0 km.
The 3-D measurement configuration consists of 81 parallel profile lines, see Figures 3
and 4. The measurement configuration along each profile line is the same as for the
2-D measurement. The distance between parallel profile lines is 0.025 km. We also test
measurement configuration with twice greater line step 0.05 km (41 profile lines).

38
Figure 3. 3-D model with 81 parallel profile lines, curved interface and bottom model plane. The
horizontal dimensions of the model are 9.2 km x 3 km, the depth is 3 km. We compute and stack
migrated sections in 2-D plane (blue) located at the middle of the shot-receiver configuration (at
horizontal coordinate of 1.5 km).

Figure 4. Top view of 81 parallel profile lines.

4. Recorded wave field

Computation of the recorded wave field in the models has been performed using the
ANRAY software package (Gajewski & Pšenčı́k, 1990).
3-D ray tracing (program anray.for) is used for calculation of two-point rays of
reflected P-wave. Program fresan.for computes the frequency response at a system
of receivers (vertical component, explosive source). Program synfan.for calculates
ray-theory seismograms. The sampling interval is 4 ms and the number of samples per
trace is 1024. Seismograms are converted into GSE format and gridded. History files
summarize parameters and execute programs for all 240 shot-receiver configurations in
one profile line. In view of the fact that the layers are homogeneous and the curved
interface is independent of the distance perpendicular to the profile lines, the recorded

39
wave field is the same for all parallel profile lines.
Figures 5-9 show examples of two-point rays and seismograms for three selected
shot-receiver configurations. Shots 1, 120 and 240 are at horizontal coordinates of 3
km, 5.975 km and 8.975 km. Rays and seismograms are computed in models with
a) isotropy (ISO), b) transversely isotropic media with a horizontal symmetry axis
(HTI), c) triclinic anisotropy (TA) and d) monoclinic anisotropy (MA) in the upper
layer.
Note the different illumination of the curved interface by two-point rays for the
selected shots and receivers.

shot 1 shot 120 shot 240

0 30 traces 60 90 0 30 traces 60 90 0 30 traces 60 90

0 0 0

1 1 1
time (s)

time (s)

time (s)

2 2 2

3
shot 1 3
shot 120 3
shot 240
Figure 5: Isotropy (ISO) in the upper layer. Two-point rays and seismograms computed in
isotropic model with P-wave velocity in the upper layer 2.5 km/s. Two-point rays of reflected P-wave
propagate in 2-D plane. Shots 1, 120 and 240 are located at horizontal coordinates of 3 km, 5.975 km
and 8.975 km.

40
 shot 1 shot 120 shot 240

0 30 traces 60 90 0 30 traces 60 90 0 30 traces 60 90

0 0 0

1 1 1
time (s)

time (s)

time (s)

2 2 2

3
shot 1 3
shot 120 3
shot 240
Figure 6: Transversely isotropic media with a horizontal symmetry axis (HTI) 24% in
the upper layer. Two-point rays and seismograms computed in model with HTI 24% for P-waves in
the upper layer. Two-point rays of reflected P-wave propagate in 2-D plane. Shots 1, 120 and 240 are
located at horizontal coordinates of 3 km, 5.975 km and 8.975 km.

41
 shot 1 shot 120 shot 240

0 30 traces 60 90 0 30 traces 60 90 0 30 traces 60 90

0 0 0

1 1 1
time (s)

time (s)

time (s)

2 2 2

3
shot 1 3
shot 120 3
shot 240
Figure 7: Triclinic anisotropy (TA) in the upper layer. Two-point rays and seismograms
computed in model with triclinic anisotropy (TA) in the upper layer. Two-point rays of reflected P-
wave propagate in 3-D volume. Shots 1, 120 and 240 are located at horizontal coordinates of 3 km,
5.975 km and 8.975 km.

42
 shot 240 shot 120

shot 1

Figure 8: Triclinic anisotropy (TA) in the upper layer. Detailed views of two-point rays
computed for shots 1, 120 and 240 in model with triclinic anisotropy (TA) in the upper layer and
displayed in Figure 7. Note the curved path of reflections at the interface.

43
 shot 1 shot 120 shot 240

0 30 traces 60 90 0 30 traces 60 90 0 30 traces 60 90

0 0 0

1 1 1
time (s)

time (s)

time (s)

2 2 2

3
shot 1 3
shot 120 3
shot 240
Figure 9: Monoclinic anisotropy (MA) in the upper layer. Two-point rays and seismograms
computed in model with monoclinic anisotropy (MA) in the upper layer. Two-point rays of reflected
P-wave propagate in 2-D plane. Shots 1, 120 and 240 are located at horizontal coordinates of 3 km,
5.975 km and 8.975 km.

44
5. Kirchhoff prestack depth migration
We use MODEL, CRT, FORMS and DATA packages for the Kirchhoff prestack depth
migration (Červený, Klimeš & Pšenčı́k, 1988; Bulant, 1996). The migration consists of
one or two-parametric ray tracing from individual surface points (program crt.for),
calculating grid values of travel time and amplitude (program mtt.for), common-shot
migration (program grdmigr.for) and stacking of migrated images. History files sum-
marize parameters and execute programs for all 240 shot-receiver configurations in pro-
file lines.

5.1 2-D migration

Figure 10 shows stacked migrated sections calculated in models with isotropy (ISO),
transversely isotropic media with a horizontal symmetry axis (HTI) 24%, triclinic
anisotropy (TA) and monoclinic anisotropy (MA). The anisotropy in the upper layer of
the velocity model used for computation of the recorded wave field is the same as the
anisotropy in the velocity model used for migration. The crosses denote the interface
in the velocity model used for computation of the recorded wave field.
Figures 11-13 show comparison of stacked migrated sections when the anisotropy
in the upper layer of the velocity model used for computation of the recorded wave field
is different from the anisotropy in the velocity model used for migration. We com-
pare 24% HTI anisotropy used for the recorded wave field with 15% HTI and 3.5% HTI
anisotropy used for the migration (see Figure 11). Another test is a comparison between
24% HTI anisotropy used for the recorded wave field and two values of isotropic velocity
2.5 km/s and 4.5 km/s used for the migration (see Figure 12). Figure 13 shows com-
parison of monoclinic anisotropy (MA) used for the recorded wave field with isotropic
velocity 2.45 km/s used for the migration. These tests should simulate situations when
we made a bad guess of the velocity model for migration.
The computing time of the Kirchhoff migration, for one 2-D stacked section com-
puted on a grid of cells 16×16 metres and interpolated to grid of cells 4×4 metres, is
approximately 2 hours. The time of the computation corresponds to an Athlon one core
2.3 GHz processor.

5.2 3-D migration

Figures 14, 15 show stacked migrated sections calculated in models with triclinic aniso-
tropy (TA) and monoclinic anisotropy (MA). The anisotropy in the upper layer of
the velocity model used for computation of the recorded wave field is the same as
the anisotropy in the velocity model used for migration. Shot-receiver configuration
consists of 81 parallel profile lines with step 0.025 km (see Figures 3, 4). We also tested
3-D migration with profile line step twice enlarged (from 0.025 km to 0.050 km, 41
parallel profile lines). First profile line starts at horizontal coordinate of 0.5 km and
the last profile line ends at horizontal coordinate of 2.5 km. We compute and stack
81 or 41 migrated sections in the 2-D plane located in the middle of the shot-receiver
configuration (at horizontal coordinate of 1.5 km, see Figure 3).
Note that the migrated sections computed with profile line step 0.050 km (41 lines)
for monoclinic anisotropy (MA, Figure 15) contain greater residua of individual migrated
sections used for stacking than for triclinic anisotropy (TA, Figure 14).

45
 0 2 4 6 distance (km) 8
0
ISO

1
depth (km)

3
0 2 4 6 distance (km) 8
0
24% HTI

1
depth (km)

3
0 2 4 6 distance (km) 8
0
TA

1
depth (km)

3
0 2 4 6 distance (km) 8
0
MA

1
depth (km)

3
Figure 10. Stacked migrated sections calculated in models with isotropy (ISO), 24% HTI anisotropy,
triclinic anisotropy (TA) and monoclinic anisotropy (MA). The anisotropy in the upper layer of the
velocity model used for computation of the recorded wave field is the same as the anisotropy in the
velocity model used for migration. The crosses denote the interface in the velocity model used for
computation of the recorded wave field.

46
 0 2 4 6 distance (km) 8
0
24% HTI

1
depth (km)

3
Recorded wave field - 24% HTI anisotropy
Migration - 24% HTI anisotropy

0 2 4 6 distance (km) 8
0
24% HTI x 15% HTI

1
depth (km)

3
Recorded wave field - 24% HTI anisotropy
Migration - 15% HTI anisotropy

0 2 4 6 distance (km) 8
0
24% HTI x 3.5% HTI

1
depth (km)

3
Recorded wave field - 24% HTI anisotropy
Migration - 3.5% HTI anisotropy
Figure 11. Comparison of stacked migrated sections when the percentage of HTI in the upper layer
of the velocity model used for computation of the recorded wave field is different from the percentage
of HTI in the velocity model used for migration. The crosses denote the interface in the velocity model
used for computation of the recorded wave field.

47
 0 2 4 6 distance (km) 8
0
24% HTI

1
depth (km)

3
Recorded wave field - 24% HTI anisotropy
Migration - 24% HTI anisotropy

0 2 4 6 distance (km) 8
0
24% HTI x 2.5km/s ISO

1
depth (km)

3
Recorded wave field - 24% HTI anisotropy
Migration - isotropy (ISO), P-wave velocity is 2.5 km/s

0 2 4 6 distance (km) 8
0
24% HTI x 4.5km/s ISO

1
depth (km)

3
Recorded wave field - 24% HTI anisotropy
Migration - isotropy (ISO), P-wave velocity is 4.5 km/s
Figure 12. Comparison of stacked migrated sections when the upper layer of the velocity model used
for computation of the recorded wave field is anisotropic (24% HTI) and the velocity model used for
migration is isotropic (ISO). The crosses denote the interface in the velocity model used for computation
of the recorded wave field.

48
 0 2 4 6 distance (km) 8
0
MA

1
depth (km)

3
Recorded wave field - monoclinic anisotropy (MA)
Migration - monoclinic anisotropy (MA)

0 2 4 6 distance (km) 8
0
MA x 2.45km/s ISO

1
depth (km)

3
Recorded wave field - monoclinic anisotropy (MA)
Migration - isotropy (ISO), P-wave velocity is 2.45 km/s
Figure 13. Comparison of stacked migrated sections when the upper layer of the velocity model used
for computation of the recorded wave field is anisotropic (MA) and the velocity model used for migration
is isotropic (ISO). The crosses denote the interface in the velocity model used for computation of the
recorded wave field.

49
 0 2 4 6 distance (km) 8
0
TA 3-D, 81 profile lines

1
depth (km)

3
Stacked migrated section for 81 parallel measurement lines

0 2 4 6 distance (km) 8
0
TA 3-D, profile line at 0.5 km

1
depth (km)

3
Migrated section for 1 measurement line (coordinate of 0.5 km)

0 2 4 6 distance (km) 8
0
TA 3-D, 41 profile lines

1
depth (km)

3
Stacked migrated section for 41 parallel measurement lines
Figure 14. Stacked migrated sections calculated in 3-D model with triclinic anisotropy (TA). The
sections are stacked at the 2-D plane section located in middle of measurement configuration (coordinate
of 1.5 km). The anisotropy in the upper layer of the velocity model used for computation of the recorded
wave field is the same as the anisotropy in the velocity model used for migration. The crosses denote
the interface in the velocity model used for computation of the recorded wave field.

50
 0 2 4 6 distance (km) 8
0
MA 3-D, 81 profile lines

1
depth (km)

3
Stacked migrated section for 81 parallel measurement lines

0 2 4 6 distance (km) 8
0
MA 3-D, profile line at 0.5 km

1
depth (km)

3
Migrated section for 1 measurement line (coordinate of 0.5 km)

0 2 4 6 distance (km) 8
0
MA 3-D, 41 profile lines

1
depth (km)

3
Stacked migrated section for 41 parallel measurement lines
Figure 15. Stacked migrated sections calculated in 3-D model with monoclinic anisotropy (MA).
The sections are stacked at the 2-D plane section located in middle of measurement configuration
(coordinate of 1.5 km). The anisotropy in the upper layer of the velocity model used for computation
of the recorded wave field is the same as the anisotropy in the velocity model used for migration. The
crosses denote the interface in the velocity model used for computation of the recorded wave field.

51
6. Conclusions
We used Kirchhoff prestack depth migration for calculation of migrated sections in an
anisotropic velocity model with one curved interface. The upper layer of the velocity
model has several types of anisotropy. The bottom layer is isotropic.
We computed 2-D stacked migrated sections when the anisotropy in the upper
layer of the velocity model used for computation of the recorded wave field is the same
as the anisotropy in the velocity model used for migration. In this case, migrated
interface coincides nearly perfectly with the interface in the model used for computation
of the recorded wave field. The length of the migrated interface is dependent on the
illumination by rays that differs for different types of anisotropy.
We compared 2-D stacked migrated sections when the anisotropy in the upper layer
of the velocity model used for computation of the recorded wave field is different from
the anisotropy in the velocity model used for migration. In this case migrated interface
is shifted and deformed more by wrong guess of isotropic velocity than wrong guess of
HTI anisotropy in the model used for migration.
We tested 3-D Kirchhoff prestack depth migration in a simple 3-D model with suc-
cess. We also tested 3-D migration with profile line step twice enlarged (from 0.025
km to 0.050 km). The migrated section computed with 0.050 km step for monoclinic
anisotropy contains greater residua of individual sections used for stacking than mi-
grated section computed for triclinic anisotropy.

Acknowledgments
The author thanks Ivan Pšenčı́k and Luděk Klimeš for great help throughout the work
on this paper.
The research has been supported by the Grant Agency of the Czech Republic
under contracts 205/07/0032 and P210/10/0736, by the Ministry of Education of the
Czech Republic within research project MSM0021620860, and by the members of the
consortium “Seismic Waves in Complex 3-D Structures” (see “http://sw3d.cz”).
References
Bulant, P. (1996): Two-point ray tracing in 3-D. Pure appl. Geophys., 148, pp. 421-447.
Bucha, V. & Bulant, P. (eds.) (2005): SW3D-CD-9 (DVD-ROM). In: Seismic Waves in
Complex 3-D Structures, Report 15, pp. 345–345, Dep. Geophys., Charles Univ.,
Prague, online at “http://sw3d.cz”.
Červený, V., Klimeš, L. & Pšenčı́k, I. (1988): Complete seismic-ray tracing in three-
dimensional structures. In: Doornbos, D.J.(ed.), Seismological Algorithms, Aca-
demic Press, New York, pp. 89–168.
Gajewski, D. & Pšenčı́k, I. (1990): Vertical seismic profile synthetics by dynamic ray
tracing in laterally varying layered anisotropic structures. J. geophys. Res., 95B,
pp. 11301–11315.
Mensch, T. & Rasolofosaon, P. (1997): Elastic-wave velocities in anisotropic media of
arbitrary symmetry-generalization of Thomsens parameters ǫ, δ and γ. Geophys. J.
Int., 128, pp. 43–64.
Shearer, P.M. & C.H. Chapman (1989): Ray tracing in azimuthally anisotropic media: I.
Results for models of aligned cracks in the upper crust, Geophys. J., 96, pp. 51–64.
Versteeg, R. J. & Grau, G. (eds.) (1991): The Marmousi experience. Proc. EAGE
workshop on Practical Aspects of Seismic Data Inversion (Copenhagen, 1990), Eur.
Assoc. Explor. Geophysicists, Zeist.
52