Anisotropy Estimation for Prestack Depth Imaging A Tomographic Approach

Tony Huang, Sheng Xu, Yu Zhang, CGGVeritas, 10300 Town Park Drive, Houston, TX 77072, USA
Summary

The work flow is as follows:

Start with isotropic velocity

Building an anisotropic velocity model from surface

seismic data generally requires complete elastic wave
records. P wave anisotropy parameter determination
normally occurs in the frame work of high order moveout
analysis of common offset gathers in the time domain.
This is the conventional approach for vertical velocity
analysis and effective estimation.
In areas where well data (sonic logs, check shots et al) are
available, we are able to get accurate vertical velocity by
calibrating seismic migration velocity with check shot
velocity. We present a method to accurately determine
anisotropic parameters in transversely isotropic media
without the weak anisotropy assumption. We test the
algorithm on synthetic data and present a work flow to
determine anisotropic parameters for prestack depth
imaging in the Gulf of Mexico.
Introduction
In isotropic media, 3D tomography works well to give an
accurate interval velocity for prestack depth imaging.
Correct velocity field flattens gathers, and positions events
correctly in depth. However, presence of seismic
anisotropy adds difficulties for prestack depth imaging.
Estimation of reliable anisotropy parameters remains a big
challenge. There are a number of approaches developed to
estimate anisotropy parameters. Anisotropic travel time
inversion has been done for transversely isotropic media
(Alkhalihah and Tsvankin, 1995; Grechka and Tsvankin,
1998). In this approach, non-hyperbolic moveout in the
NMO equation is used to determine effective for time
migration. To use this approach in prestack depth imaging
we need to convert effective to interval , and convert
rms velocity to interval velocity using the Dix formula.
Krebs, et al (2003) present an integrated velocity estimation
technique by fitting surface seismic data and well data in an
appropriate data domain. This approach uses 1D update and
global constraints to obtain anisotropic parameters at well
locations and extrapolates parameters along geological
horizons to areas outside the wells.
We present an approach for anisotropy estimation in the
areas where vertical velocity ( v 0 ) can be accurately
obtained at well locations with vertical check shot surveys.
Our procedure is to invert for and parameters with a
joint tomographic inversion.

Anisotropic Estimation for Prestack Depth Imaging

Obtain

v0

Generate

from calibration with check shot velocity

3D

migration

mini

volume

1000mx1000m) around wells using

v0

(e.g.,

(==0)

Jointly invert for one , function per well

Generate a & volume by hanging 1d function
from WB
Run 3D anisotropic migration to generate CIGs for
3D tomographic velocity update ( v 0 only, fix
& )

Joint inversion of

and

For P wave imaging, the required anisotropic parameters

are v 0 , , and

. When the input is restricted to P wave

surface seismic, even in 1D case, an ambiguity exists in

inverting all three parameters. For example, if the seismic
event in CIG gather exhibits a hyperbolic moveout, we can
not distinguish whether it is caused by velocity perturbation
or by a perturbation of elliptical anisotropy ( = )
parameters. Either perturbation can flatten the gathers in
the depth domain. When non surface seismic data (e.g.,
check shots) are used, we can accurately determine v 0 and
ambiguity is no longer a problem.

Knowledge of v 0

allows us to invert for the anisotropic parameters

and

simultaneously.

After fixing v 0 , we migrate gathers around the well area,

and pick actual curvature in common image gathers (CIG).
An algorithm is developed to invert and
simultaneously. We start from the Hamiltonian:
v 2 ( x)
(1)
(G 1)
H ( x, p) = 0
4

Under acoustic approximation for VTI media (Alkhalihah

and Tsvankin, 1995), the eikonal G is defined as
2
1
G = (1+ 2 ) px2 + pz2 + (1+ 2 ) px2 + pz2 8( ) px2 pz2 1,
2

Where

px , p z are

the normalized slowness components

(Zhou et al, 2003)

p x = p x / v0 ,

p z = p z / v0 ,

Anisotropy Estimation for Prestack Depth Imaging

From equation (1), we derive the group momentum for the

ray tracing,
dx v02 p x (1 + 2 ) 2( ) p z2
=
(2)
d
1 2( ) p x2 p z2

dz v02 p z 1 2( ) p x2
=
d
1 2( ) p x2 p z2

We derive equation (2) without weak anisotropy

assumption (Thomsen 1986). Notice that if we introduce
weak anisotropy assumption, the denominators of equation
(2) turn to unit. As the ray momentum on both horizontal
and vertical share the same denominator, the group
direction of the ray would be the same with or without
weakness anisotropy assumption:

dx p x (1 + 2 ) 2( ) p z2
tan =
=
dz
p z 1 2( ) p x2

(3)

,
,

for

Layer

z (m)

V0 (m/s) Delta

Epsilon Epsilon
weak

strong

1500

1500

3000

0.03

0.06

0.06

3000

3000

0.03

0.06

0.18

4500

3000

0.03

0.06

0.18

6000

For tomography, the shot and receiver locations are

invariants. We need to derive the travel time perturbation
with respect to anisotropic parameter and in a
condition of keeping group direction constant. Using
eikonal equation and equation (3), we then derive the
derivative

Table 1: 1D model parameters

and

joint

travel time inversion.

Synthetic Studies

We perform depth migration using two modeled input

gathers, and pick curvatures in the common offset imaging
gathers (CIG) to simulate the real case application. Figure 1
shows CIG gathers and curvature picks that went into the
tomography inversion.
Figure 2 shows inversion results for two models. The
derived anisotropic parameters are within one percent off
from the parameters of true models. We are able to
distinguish weak and strong .

To check the inversion process, we perform analysis with

modeling data. The velocity model consists of two layers:
one water layer with velocity of 1500m/s, and another earth
layer with velocity of 3000 m/s. There are two anisotropy
models: one with weak parameter; and the other with
strong parameter. There are 4 reflectors at depth of
1500m, 3000m, 45000, and 6000m, respectively. Table 1
describes the modeling parameters.

(a) Weak anisotropy)

(b) Strong anisotropy

Figure 1: CIG gathers from migration using V0 for weak anisotropy

(a) and strong anisotropy (b). Blue lines denote curvature picks.

Anisotropy Estimation for Prestack Depth Imaging

Figure 4 shows the improvement of steeply dipping TOS

imaging near a well. There are improvements in structural
focusing and seismic/well mistie. Anisotropic imaging
shows more details of small salt wings which is confirmed
by well data.

2 Layers weak epsilon

0.07
0.06
0.05
2ly_eps_true

0.04

2ly_del_true

0.03

2ly_eps_tom
2ly_del_tom

0.02
0.01
0
0

1000

2000

3000

4000

5000

6000

7000

Conclusions
We presented a tomographic approach to determine
anisotropic parameters for transversely isotropic media. We
tested the methodology in model data as well as real data in
a marine sedimentary basin.

Depth

(a) Weak anisotropy

Acknowledgments

2 Layer strong epsilon

0.2
0.18
0.16
0.14
2ly_eps_true
2ly_del_true

0.12
0.1

2ly_eps_tomo
2ly_del_tomo

0.08
0.06
0.04
0.02
0
0

1000

2000

3000

4000

5000

6000

7000

Depth

(b) Strong anisotropy

Figure 2: Inversion results of weak anisotropy (a) and strong
anisotropy (b) model data. Delta and epsilon are very close to
true model.

Gulf of Mexico Example

The real data set is over marine sediment basins in the Gulf
of Mexico. The objective of anisotropic depth imaging is to
improve salt positioning, which is expected to give better
subsalt imaging. There are sonic logs and check shots
available. The vertical velocity ( v 0 , and) is obtained by

calibrating isotropic velocity with vertical check shot

velocity. Anisotropic parameters are not expected to change
greatly in the area. The objective is to derive a v(z)
anisotropy function ( and ) using seismic and well

data. We average and functions at well location, and

extrapolate the functions along water bottom, which is
consistent with compaction driven sediment (Figure 3).

We would like to thank CGGVeritas for giving us

permission to present the paper. We thank Mike Howard of
BHPB for discussion of anisotropic work flow, and
encouragement in the work. We also thank Bruce Ver West
for generating the synthetics and discussion in the topics.
Finally, we would like to thank David Sixtra and Arnold
Rodriguez of Anadarko Petroleum Company, Judy Mooney
and Clive Hurst of ENI Petroleum for providing check
shots in the study area.

Anisotropy Estimation for Prestack Depth Imaging

(a) Isotropic sedimentary flood migration

(b) Anisotropic sedimentary flood migration

Figure 4: Anisotropy prestack depth imaging improves definition of TOS, minimizes seismic/well mistie. (a) Isotropic sedimentary flood
migration. (b) Anisotropic sedimentary flood migration.

(a) Isotropic CIG Gathers

(b) Anisotropic CIG gathers

Figure 5: CIG depth gathers. Anisotropic imaging yields flatter gathers. (a) CIG of isotropic migration. (b) CIG of anisotropic migration.