We present a new strategy for depth velocity model building from pre-stack time migrated gathers. It is based on a dense volumetric dip and residual move-out ...
From time to depth imaging: an accurate workflow Gilles Lambaré*, Philippe Herrmann, Jean-Paul Touré, Laure Capar, Patrice Guillaume, Nicolas Bousquié, Damien Grenié, Serge Zimine, CGGVeritas. In terms of tomographic inversion the strategy proposed by Guillaume et al. (2001) may significantly reduce the number of PreStack Depth Migrations (PreSDM) required by the migration velocity analysis loop (Deregowski, 1990) (ideally one!). It is based on a tomographic inversion of demigrated kinematics’ attributes, obtained from a local structural dip and Residual Move-Out (RMO) picking. A unique picking step may then be sufficient, should the quality of the initial prestack depth migration be appropriate for a dense volumetric automated picking.
Summary We present a new strategy for depth velocity model building from pre-stack time migrated gathers. It is based on a dense volumetric dip and residual move-out (RMO) picking in the prestack time migrated domain. The kinematic information is demigrated to compute multioffset un-migrated attributes – called seismic invariants – used as input data for a multi-offset depth tomography. Compared to the corresponding existing strategy based on picking in the depth migrated domain, our new strategy allows to fully take advantage of a previous time imaging project. It bypasses the initial depth model building and initial prestack depth migration required for picking RMO in depth. Furthermore it takes advantage of the quality of a picking phase carried out on already reasonably focused time migrated dataset, with no compromise on the accuracy. We demonstrate the relevance of the approach on a real dataset.
We propose to adapt this workflow so that we can take advantage of existing time imaging results. In this scheme, depth velocity model building uses a dense volumetric automated dip and RMO picking from Pre-Stack Time Migration (PreSTM) gathers rather than from PreSDM gathers. The benefit is two-fold: first, we bypass both the initial depth model building step and the initial PreSDM step. Second, the automated picking takes advantage of the quality of the focusing in PreSTM domain, without any compromise on available kinematic information. We present an application to a real dataset from TOTAL demonstrating the relevance and accuracy of the approach. Our results are compared to those obtained with a PreSDM picking but also to those of the PreSTM project after depth to time conversion.
Introduction Depth velocity model building based on a dense volumetric automated picking (Woodward et al., 1998; Billette et al., 2000; Guillaume et al., 2001; Etgen et al., 2002; Chauris et al., 2002a, b; Nguyen et al., 2003) allows for an accurate estimate of depth velocity models, while getting rid of interpretative picking.
Image dip
CIP gather X, Y RMO Z
X, Y Z
CIP Zx,y (h) Offset,
Figure 1 : Kinematic information: the Residual move-out curve and local dip.
SEG/San Antonio 2007 Annual Meeting
3014
Delpth velocity model building Estimating a depth velocity model certainly remains one of the great challenges in seismic imaging. Velocity model building usually requires picking in the PreSDM domain, with the benefits of a better signal to noise ratio and of an easier interpretation (Al-Yahya, 1989; Liu and Bleistein, 1995).The use of a dense volumetric picking has imposed itself as a very efficient strategy and semi automated approaches are now available. A typical depth velocity model building workflow involves: 1) PreSDM using some initial velocity model, 2) RMO picking on Common Image Point (CIP) gathers (Figure 1), 3) Structural dip picking on depth migrated image, 4) Update of the depth velocity model by ray-based tomographic inversion. This procedure is usually repeated several times to improve the quality of the migrated image. The choice of the initial velocity model is an important issue since it governs the quality of the automated picking but also the number of iterations in the migration velocity analysis loop. Concerning this last point, the strategy proposed by Guillaume et al. (2001) avoids the dependency of velocity model update on the (velocity) model used for the prestack depth migration and picking. It is based on the use of kinematic invariants computed through kinematic demigration of events picked on CIP gathers (Figure 2). A locally coherent event in a common offset depth migrated image (left) is characterized by its central position and dip, and is associated in the un-migrated time domain (right) to a set of source (S) and receiver (R) positions, two-way time, TSR and a local time slope, GradT; these attributes are
Kinematic invariants describe the local kinematic information in the un-migrated time domain. One can “remigrate” them with any new depth velocity model, providing an assessment of the quality of the depth velocity model. Several tomographic inversion loops can be performed from a single picking. Moreover if the automated picking is dense and accurate enough, an accurate final depth velocity model can be obtained directly from a single picking step in the PreSDM domain. Starting from PreSTM gathers We propose to replace the picking in the PreSDM domain by a picking in the PreSTM domain. When a completed time imaging project is available, our approach allow to fully take advantage of this information. The picking is performed in the PreSTM domain, on already well focused images. No compromise is made on the quality of the kinematic invariants that are obtained via a pre-stack kinematic time de-migration. Compared to the original depth imaging workflow described in Guillaume et al. (2004), the only new requirement is for a prestack kinematic time de-migration. We developed such a tool. For any locally coherent event in a common offset PreSTM image described by its position, (X, t0), and its dip, we compute the corresponding shot and receiver positions, S, R, given the offset and azimuth. This computation uses the PreSTM travel time functions (see for example Castle (1994)). At computed shot and receiver positions, we derive the slope of the event
kinematic de-migration
X
Z
called the kinematic invariants because they don’t depend on the velocity model (as long as we use the same velocity model for migration and de-migration).
Dip
Common offset section
Velocity model
S, R
GradT
TSR
Common offset section
Figure 2: Computation of kinematic invariants by kinematic demigration.
SEG/San Antonio 2007 Annual Meeting
3015
Conclusion in the un-migrated domain, thus forming a full set of kinematic invariants. Compared to the kinematic demigration in depth, the kinematic de-migration in time is a simpler process due to the simplicity of the one-way travel time curves used by the PreSTM. Application to a real dataset To illustrate the relevance of our approach, let us consider the case where a depth imaging project has followed a former time imaging project. Figure 3 shows the PreSTM stack (left) and a few associated CIP gathers (right) with superimposed RMO picks. Figure 4 shows CIP gathers in the time domain (PreSTM or PreSDM converted to time) for comparison. Figure 4 top shows the PreSTM gathers used for the picking. Figure 4 middle shows the CIP gathers (PreSDM converted to time) obtained with our final depth velocity model (tomographic inversion from a picking on PreSTM gathers). Finally Figure 4 bottom shows the CIP gathers (PreSDM converted to time) obtained within a “standard“ depth imaging project (tomographic inversion from a picking on PreSDM gathers). Note that for this last result only half of the full offset range was used, which explains the poor flattening of the gathers at large offsets. We see the reasonably good focusing of our result, which can be advantageously compared to the “standard” result (Figure 4 bottom). We see that we have improved flattening even when compared to PreSTM, which clearly demonstrate that our strategy allows fully taking advantage of an existing time imaging project, with no compromise on accuracy. Not only time to depth conversion has been performed but also time imaging has been improved (in fact depth imaging convert to time).
We propose a new workflow for depth velocity model building. It involves an automated dense volumetric picking on PreSTM gathers. Compared to the “standard” workflow, where picking is done on PreSDM gathers, our approach preserves the full information gained in the time imaging project. Indeed in the standard approach the initial PreSDM gathers used for the picking are obtained with a depth velocity model derived from the time velocity model assuming zero-dip and no-RMO. As a result they may exhibit a poor focusing, which penalize automated picking. On the contrary in our approach the automated picking may benefit from the good focusing of the PreSTM gathers (final result of a time imaging project). In fact our approach decouples focusing, done in the time domain, and depthing done within the tomographic inversion in depth. Focusing is important for automated picking, which can be done with no compromise on accuracy thanks to the concept of kinematic invariants. We have tested our approach with an application to a real dataset. The comparison with the “standard” workflow demonstrates the relevance of our approach. Moreover our final PreSDM gathers converted to time exhibit also a real improvement compared to the final PreSTM gathers. Acknowledgments We thank TOTAL & CHEVRON for the permission to show the results obtained on the real dataset. We thank Ariane Herrenschmidt (CGG Massy) for her help, as well as Risto Siliqi and Eric Suaudeau (CGG Massy) for very fruitful discussions.
Figure 3: Time imaging project. Left) the PreSTM stack, Right) some CIP gathers in time with superimposed RMO curves (location of CIP gathers is indicated on the stack by the dotted line).
SEG/San Antonio 2007 Annual Meeting
3016
PreSTM
Our result
PreSDM Figure 4: CIP gathers in time: Top) PreSTM ; Middle) PreSDM (converted to time!) using the depth velocity model obtained from a picking on PreSTM gathers; Bottom) PreSDM (converted to time!) using the depth velocity model obtained from a picking on PreSDM gathers (only half of the offset range was used for this project!).
SEG/San Antonio 2007 Annual Meeting
3017
EDITED REFERENCES Note: This reference list is a copy-edited version of the reference list submitted by the author. Reference lists for the 2007 SEG Technical Program Expanded Abstracts have been copy edited so that references provided with the online metadata for each paper will achieve a high degree of linking to cited sources that appear on the Web. REFERENCES Al-Yahya, K., 1989, Velocity analysis by iterative profile migration: Geophysics, 54, 718–729. Billette, F., J. Etgen, and W. Rietveld, 2000, Prestack depth migration using fast 3D tomography: 62nd Annual Conference and Exhibition, EAGE, Extended Abstracts, L37. Castle, R. J., 1994, A theory of normal moveout: Geophysics, 59, 983–999. Chauris, H., M. Noble, G. Lambaré, and P. Podvin, 2002, Migration velocity analysis from locally coherent events in 2-D laterally heterogeneous media, Part I : Theoretical aspects: Geophysics, 67, 1213–1224. ———, 2002, Migration velocity analysis from locally coherent events in 2-D laterally heterogeneous media, Part II : Practical aspects: Geophysics, 67, 1202–1212. Deregowski, S. M., 1990, Common-offset migrations and velocity analysis: First Break, 8, 224–234. Guillaume, P., F. Audebert, P. Berthet, B. David, A. Herrenschmidt, and X. Zhang, 2001, 3D finite-offset tomographic inversion of CRP-scan data, with or without anisotropy: 71st Annual International Meeting, SEG, Expanded Abstracts, 718–721. Guillaume, P., F. Audebert, N. Chazanoel, V. Dirks, and X. Zhang, 2004, Flexible 3D finite-offset tomography velocity model building: 66th Annual Conference and Exhibition, EAGE , Extended Abstracts, D045. Liu, Z., and N. Bleistein, 1995, Migration velocity analysis: Theory and an iterative algorithm: Geophysics, 60, 142–153. Nguyen, S., M. Noble, R. Baina, M. Alerini, and V. Devaux, 2003, Slope tomography assisted by migration of attributes: 65th Annual Conference and Exhibition, EAGE, Extended Abstracts, D28. Woodward, M., P. Farmer, D. Nichols, and S. Charles, 1998, Automated 3D tomographic velocity analysis of residual move-out in pre-stack migrated common image point gathers: 68th Annual International Meeting, SEG, Expanded Abstracts, 1218–1221.
SEG/San Antonio 2007 Annual Meeting
3018
