Professional Documents
Culture Documents
Velocity Model Representation and Updating: Layer-Based Versus Gridded Models
Velocity Model Representation and Updating: Layer-Based Versus Gridded Models
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
Model-building techniques generally are divided recover in a complex model often will manifest
into two sequential phases: picking and inverting. itself in the nonhyperbolicity of the moveout in the
The errors and assumptions in these two phases are prestack data. In normal moveout (NMO) stacking,
quite distinct (Jones, 2003). In this chapter, we deal we assume hyperbolicity, so stacking destroys use-
primarily with the model-representation and pick- ful information in that case.
ing aspects of model building. In Chapter 6, we Prior to the 1980s, most model-building tech-
expand on picking, and in Chapter 7, we consider niques implicitly assumed the earth to be a 1D (lay-
the inversion aspects of model updating. ered) medium. The papers in this reprint volume
Picking can be performed (either manually or concern our attempts to move beyond that restrictive
using automation) on a single offset (or the stack) assumption so that geoscientists could address
or for a series of offsets in the gathers. Picking lateral velocity variation, both in the model and in
also can be performed either before or after an ini- the associated update schemes (e.g., Reshef, 1994;
tial migration. Picking can be done along horizons Tieman, 1995).
or for a scatter of discontinuous events. For beam
migration, for example, the picking of local time
dips is performed concurrently in two domains Layer-based versus gridded models
(e.g., shot and receiver). If picking is done before
migration, the picking error might be large because Models themselves also fall into two major cate-
tracking or distinguishing the various arrival gories, thereby reecting the underlying geologic
branches of diffraction hyperboloids is difcult environments: layer-based and nonlayer-based
for complex environments. (gridded) models. Examples of these representa-
Given the fact that manual picking can be tedi- tions and their associated problems can be found in
ous, various schemes have been developed for auto- van Trier (1988, 1989), Wyatt et al. (1992), Wiggins
mating this part of the process. In addition, various et al. (1993), and Yilmaz et al. (1994).
data-reduction schemes can be employed, such as In layer-based models, velocity gradients and
stacking, so that the picking need be done on only vertical-compaction gradients are bounded by sedi-
one data volume (on the zero-offset cube, for exam- mentary interfaces. There, it is sufcient to pick
ple) instead of on many nite-offset volumes. For seismic-reection events as the partitions to the
example, the trade-off associated with picking on velocity regions in the model.
stacked data constitutes a speedup in the overall Some nonlayer-based models have velocity
inversion process, but by stacking, we lose infor- regimes dominated by compaction gradients that
mation. That is, the complexity we are trying to are subparallel to the seabed in the marine case. In
397
398 Prestack Depth Migration and Velocity Model Building
the case of salt or shale tectonism, the scenario is we make a scatter of picks and input the resulting
complicated by the presence of these irregular cloud of values into the inversion scheme.
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
Density of picks and automation for complex areas, where diffraction branches from
various arrivals will be difcult to distinguish and
Regardless of the technique employed, another
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
In Thomsens notation and for the simple case of interval values are obtained, then can be deter-
a at-lying (1D) medium, the vertical and horizon- mined for use in depth migration.
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
tal velocities are related to the surface seismic For a 1D medium, Alkhalifah and Tsvankin
near-offset moveout velocity (Vnmo) by (1995) and Alkhalifah (1997) described a cumula-
tive effective anisotropy that incorporates various
Vnmo = Vv (1 + 2d ) Vv (1 + d ) (1) nonhyperbolic moveout effects:
Vh = Vv (1 + 2e ) Vv (1 + e ) (2) t j v 4j (1 + 8h j )
1 j
heff (To ) = 4
- 1 , (4)
8 T0 Vnmo
and
Vh = Vnmo (1 + 2h) Vnmo (1 + h) (3) where vj is the vertical interval velocity derived
from short-offset NMO velocities Vnmo using a Dix
where Vnmo is the near-offset velocity estimated inversion, tj is the two-way vertical time in the jth
from stacking velocity analysis, Vv is the vertical layer, and T0 is the total vertical time to the layer.
velocity seen in well logs (after calibration to the We must invert this equation to recover the interval
seismic band width), and Vh is the horizontal com- values of .
ponent of velocity (to which we do not usually It is also possible to incorporate inversion of
have access). these parameters into a tomography (e.g., Sexton
Alkhalifahs parameter can be related to and Williamson, 1998). By using a tomographic
Thomsens and formulation via = ( )/ solution for anisotropic parameter estimation, we
(1 + 2 ). can deal with a complex model rather than being
Unless we rst have obtained a reliable estimate limited to 1D anisotropic assumptions.
of the vertical velocity-compaction gradients, the
anisotropy parameter estimates will be in error
(Jones et al., 2002). However, assuming this has Addressing the full wavefield
been done, we usually estimate from well ties
(Reilly, 1993) and determine from higher-order Migration and model building for shear-wave
moveout estimates or during tomographic inver- and converted-wave data are still immature. Al-
sion incorporating long-offset data. No reliable though migrating a shear-wave recording inde-
inversion scheme based on surface seismic meas- pendently is relatively straightforward, we still
urements has yet been demonstrated for obtaining have the difcult problem of obtaining reliable
, although some methods have been discussed parameter estimates. Scanning techniques have
(e.g., Isaac and Lawton, 2002). A serious restric- been presented, for example, that correlate images
tion on anisotropic parameter estimation from obtained from different shooting directions to
surface seismic data in the past was the absence estimate the depth-varying ratio VP:VS. However,
of long-offset information (i.e., the maximum ac- those techniques are restricted by 2D assumptions
quired offset was usually less than the depth of (Audebert et al., 1999).
the target horizons; consequently, anisotropic A major challenge in dealing with converted-
higher-order moveout effects were not manifest wave data is identifying the same events on both
in the data). the P and S (or PS) seismic sections. Because of
In analyzing long-offset data, we actually meas- changes in their respective reectivities, horizons
ure a cumulative e, which must be inverted to of interest often are absent in one section. Thus,
yield the interval values of (analogously to rms relying on codepthing of events to constrain the
inversion to obtain interval velocity). Once the model building can be problematic. We really need
Chapter 5: Velocity Model Representation and Updating 401
to be able to identify such pairs of events to compute exible hybrid-model builder should permit use of
VP /VS over a known geologic interval. both types of representations. Perhaps the most
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
Progress also has been made in the joint inver- important elements to keep in mind are that the
sion of P-wave and converted-wave velocity picks processes of picking and then of inverting those
(Herrenschmidt et al., 2001; Brown et al., 2002; picks are quite distinct.
Foss et al., 2004). However, we are still some way In the selection of papers reprinted with this
from a full vector eld migration (i.e., a single run chapter, we have not distinguished between tech-
of the migration inputting the P, Sv, and Sh measure- niques that result in layered models and those that
ments; e.g., Wapenaar et al., 1987). result in gridded models, nor have we dealt with
Moving beyond inversion of just traveltime infor- azimuthal anisotropy. However, we have separated
mation, full-waveform inversion (e.g., Pratt et al., the reprints into four categories:
2002, 2008; Sirgue et al., 2008; Warner et al., 2008)
offers promising results by addressing both the kine- 1) general considerations about velocity error and
matic and dynamic aspects of the recorded data. representation of the velocity eld
2) papers that deal primarily with getting reliable
data to pick (as opposed to concentrating on
Discussion how to invert those picks to update the velocity
model)
Most techniques can yield an adequate starting 3) works that deal with the updating procedure
model for depth imaging. However, for an area (mostly tomographic updating)
with moderate to complex structural geology or 4) papers that deal with anisotropy in the model
an area with rapidly changing velocities, tomo- update.
graphic techniques are benecial. Tomography has
the advantage of attempting to reconcile errors
(arising both from measurements and from under- References
lying assumptions) throughout a 3D volume. A
tomographic solution also permits us to deal more Al-Chalabi, M., 1974, An analysis of stacking rms
average and interval velocities over a horizontally
readily with anisotropy parameter estimation for
layered ground: Geophysical Prospecting, 22, 458
geology that is not at. In contrast, 1D updating 475.
schemes, which are discussed in more detail in Al-Chalabi, M., 1994, Seismic velocity A critique:
Chapter 6, deal with each measurement in isolation First Break, 12, no. 12, 589596.
from all others. Alkhalifah, T., 1997, Velocity analysis using nonhyper-
Spatial resolution of the velocity is a function bolic moveout in transversely isotropic media: Geo-
of the number of rays that the inversion uses to physics, 62, 18391854.
sample each element of the subsurface. Thus, in Alkhalifah, T., and I. Tsvankin, 1995, Velocity analysis
order of increasing resolving power, we would have: for transversely isotropic media: Geophysics, 60,
the vertical Dix update, the zero-offset normal-ray 15501566.
update, two-point parametric tomographic inversion, Armstrong, T., S. Merlin, K. Hawkins, 2002, Seismic
and multioffset tomographic inversion. Naturally, anisotropy from VSP PreSDM in Pierce eld,
central North Sea: 64th Annual Conference and
increasing the spatial sampling of any of these
Exhibition, EAGE, paper B31.
techniques would improve its precision but not Audebert, F., P. Y. Granger, and A. Herrenschmidt, 1999,
its accuracy. CCP-scan technique: True common conversion point
Most available tomographic solutions can be sorting and converted-wave velocity analysis solved
formulated for either gridded or layered models. by PP-PS prestack depth migration: 69th Annual
Both layered and gridded model representations International Meeting, SEG, Expanded Abstracts,
have strengths and weaknesses, and ideally, a 11861189.
402 Prestack Depth Migration and Velocity Model Building
Berkhout, A. J., 1997, Pushing the limits of seismic Muller, T., 1998, Common reection surface stack
imaging, Part I: Prestack migration in terms of double versus NMO/stack and NMO/DMO/stack: 60th An-
dynamic focusing: Geophysics, 62, 937953. nual Conference and Exhibition, EAGE, Extended
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
Annual Conference and Exhibition, EAGE, paper Jones, I. F., H. Baud, K. Ibbotson, F. Audebert, 2000,
F023. Continuous 3D preSDM velocity analysis: The Lead-
Woodward, M. J., P. Farmer, D. Nichols, and S. Charles, ing Edge, 19, 263269.
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
1998, Automated 3-D tomographic velocity analysis of Lines, L., 1993, Ambiguity in analysis of velocity and
residual moveout in prestack depth migrated common depth (short note): Geophysics, 58, 596597.
image point gathers: 68th Annual International Meeting, Reshef, M., 1994, The use of 3D pre-stack depth imaging
SEG, Expanded Abstracts, 12181221. to estimate layer velocities and reector positions
Yilmaz, ., P. Farmer, A. Pieprzak, and B. Godfrey, 1994, (short note): Geophysics, 62, 206210.
Estimation of velocity-depth models for structural Tieman, H. J., 1995, Migration velocity analysis: Ac-
targets: A case history from the North Sea: 64th Annual counting for the effects of lateral velocity variations:
International Meeting, SEG, Expanded Abstracts, Geophysics, 60, 164175.
12501252. Wiggins, W., U. Albertin, and G. Stankovic, 1993,
Building 3-D depth migration velocity models with
topological objects: 63rd Annual International Meet-
Reprints ing, SEG, Expanded Abstracts, 170173.
Wyatt, K. D., S. K. Towe, J. E. Layton, S. B. Wyatt,
Hubral, P., G. Hcht, and R. Jger, 1998, An introduc- D. H. von Seggern, and C. A. Brockmeier, 1992,
tion to the common reection surface stack: 60th Ergonomics in 3D depth migration: 62nd Annual
Annual Conference and Exhibition, EAGE, Extended International Meeting, SEG, Expanded Abstracts,
Abstracts, Session 01-19. 944947.
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
404
Prestack Depth Migration and Velocity Model Building
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
406
Prestack Depth Migration and Velocity Model Building
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
408
Prestack Depth Migration and Velocity Model Building
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
410
Prestack Depth Migration and Velocity Model Building
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
412
Prestack Depth Migration and Velocity Model Building
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
414
Prestack Depth Migration and Velocity Model Building
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
416
Prestack Depth Migration and Velocity Model Building
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
418
Prestack Depth Migration and Velocity Model Building
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
420
Prestack Depth Migration and Velocity Model Building
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
422
Prestack Depth Migration and Velocity Model Building
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
424
Prestack Depth Migration and Velocity Model Building
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
426
Prestack Depth Migration and Velocity Model Building
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
428
Prestack Depth Migration and Velocity Model Building
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
430
Prestack Depth Migration and Velocity Model Building
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
432
Prestack Depth Migration and Velocity Model Building
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
434
Prestack Depth Migration and Velocity Model Building
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
436
Prestack Depth Migration and Velocity Model Building
Downloaded 07/02/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
438
Prestack Depth Migration and Velocity Model Building