|MULTI- AND WIDE-AZIMUTH HIGH-RESOLUTION TOMOGRAPHY - APPLICATIONS FROM AROUND THE WORLD|
An accurate velocity model is key to improved seismic imaging. Two case studies are presented that illustrate how multi-azimuth and wide-azimuth reflection tomographic inversion leads to superior velocity model building due to the azimuthal diversity. In a case study in the Deep Water Nile Delta it is demonstrated that multi-azimuth tomography is able to resolve smaller scale velocity variations than narrow-azimuth tomography, and hence enables more accurate velocity model building in depth. In another case study in the Gulf of Mexico it is shown that with wide-azimuth tomography it is possible to successfully build a TTI-anisotropic model, which leads to more accurate reflector positioning and better well tie in the migration image.
The inversions were all carried out using an azimuth-preserving reflection tomographic solution for building velocity models (Jiao et al., 2009). Automatic full-volume picking of reflectors and associated residual moveout is utilized to minimize user interference. Extensive QC is available at several stages of the tomography. Rapid convergence to variable wavelength is achieved due to integration of fast beam migration (Sherwood et al., 2009) and a conjugate gradient solver regularized by 3D Gaussian filters (Zhou et al., 2009). The data is binned in common vector offsets, providing the flexibility to handle narrow-, wide- and multi-azimuth surveys all within the same framework.
|MULTI-AZIMUTH INVERSION IN THE NILE DELTA|
The multi-azimuth investigation (van der Burg et al., 2010) was performed on a dataset acquired in the Deep Water Nile Delta, consisting of six overlapping narrow-azimuth towed streamer (NATS) surveys at 30° sail-line increments (Figure 1, left). The study area contains anhydrite pockets below the top-Messinian and shallow channels. Both are causing substantial lateral velocity heterogeneity to be resolved by the multi-azimuth (MAZ) inversion. Separate narrow-azimuth fast beam migration with intrinsic multiple reject and residual moveout (RMO) analysis are applied for each survey. This is followed by joint, multi-azimuth, inversion of all surveys.
The RMO analysis was automatically conducted on the common image point gathers, using reflection points and local reflector dips automatically picked on the full-volume stack. The auto-picking was controlled by adjusting parameters such as the minimum spacing between reflectors, semblance from the residual moveout analysis and reflector coherency. QC was conducted by viewing 3D RMO volumes, reflector picks and dips, and by overlaying the picked moveout on common image point gathers. The inversion step was quality checked by looking at ray-density plots as well as at the slowness update. The editing of bad picks or the application of lateral smoothing to the picked moveout was not needed; it was left to the inversion to dampen out the effect of occasional bad picks.
To control the wavelength of the updated velocity field and to regularize the inversion, a set of 3D Gaussian smoothers are applied. In four iterations, the lateral smoothing standard deviation was decreased from 2000m to 250m to build up the details in the velocity field in a progressive fashion, from long wavelength to short wavelength components. Starting with a smooth initial velocity model derived from the final RMS velocities used for pre-stack time migration of all datasets, with this approach it was possible to resolve small-scale velocity anomalies caused by the shallow channel and the anhydrite pockets (Figure 1 and 2). A final comparison was made with results from narrow-azimuth tomography. The depth slices through the final velocity models and migrated stacks of the 0° survey in Figure 3 show much improved resolution below the top-Messinian in the velocity model obtained from multi-azimuth inversion. The common image point gathers obtained by migrating with the velocity depth model from multi-azimuth inversion are flatter than those obtained from narrow-azimuth inversion. The stack from multi-azimuth is improved compared to the stack from narrow-azimuth inversion. The resolved small scale velocity variations correlate very well with geology.
The wide-azimuth anisotropic investigation (Schleicher et al., 2010) was performed on the Crystal survey in Keathley Canyon, Gulf of Mexico. Wide azimuth tomography projects back depth errors in both offset-x and offset-y into velocity updates. In the survey area, the subsurface exhibits tilted transverse isotropy (TTI). This requires the estimation of anisotropy parameters epsilon, delta, and axis of symmetry in addition to the velocity. The left panel of Figure 4 shows a section through the isotropic beam migrated stack volume and gathers. In presence of anisotropy, isotropic depth migration overestimates depth. Depth differences between a calibration well and the isotropic migrated image were used to make an initial estimate of delta. The delta field was assumed to be spatially smooth and was interpolated conformable to horizon data. For epsilon elliptical anisotropy was assumed in this stage. The isotropic velocity was scaled by delta in the vertical travel time domain to obtain an initial estimate of the velocity normal to the bedding planes. This resulted in an initial model for vertical transverse isotropic (VTI) migration. As expected, after VTI migration the position of reflectors appear shallower (indicated with an arrow on the middle panel of Figure 4) and has a good well tie.
In the Gulf of Mexico, the geometry of the high velocity salt bodies is an important step in velocity model building. The geometry is estimated using the sediment flood, salt flood, full salt processing sequence. After estimating sediment velocity above salt, a sediment flood migration using only the sediment velocity is interpreted for the top salt. A salt flood migration using salt velocity everywhere beneath salt is used to locate the base salt. The full salt migration that incorporates both the top and base salt is the final migration. Additional migrations were required to image top salt and base salt for overhung and stacked salt bodies. Cycle time is much reduced by using a fast 3D, wide azimuth, beam, prestack depth migration.
It has been demonstrated that due to increased ray coverage associated with azimuthal diversity, multi-azimuth tomography is able to resolve smaller scale velocity variations than narrow-azimuth tomography, and hence enables more accurate velocity model building in depth. Also it has been shown that iterating TTI migration and wide-azimuth tomography creates flatter gathers and improved reflector positioning in the presence of anisotropy.
BP, RWE Dea and EGAS are thanked for permission to publish the Nile Delta data. PGS is thanked for permission to show the Crystal Gulf of Mexico data. John Cramer, Clive Gerrard and Armando Sosa performed the processing and inversion of the Crystal data.
Jiao, J., Lin, S., Zhou, C., Brandsberg-Dahl, S., Schleicher, K. and Tieman, H.  Multi-parameter controlled automatically picking and variable smoothing for tomography with fast 3D beam prestack depth migration, 79th SEG Annual Meeting, 28, Expanded Abstracts, pp. 3989-3993.