3D velocity-independent elliptically-anisotropic moveout correction |
Plane-wave destruction filters provide a truly automated approach to velocity analysis, as they can be used without any user-selected input parameters. Here, we have used the finite-difference plane-wave destructors, which, as described by Fomel (2002), can be given a user-supplied initial estimate of the slope field. Providing an initial slope estimate helps improve the efficiency of the slope-detection and can help estimate conflicting slopes (Fomel, 2002). In all of the examples above, no initial slope field was provided. The output slope fields are computed using smoothing regularization, which helps make the moveout correction more robust, and provides a way for the user to interact with the slope detection performance. If the seismic data is particularly noisy, a more aggressive smoothing can help make a more consistent automatic NMO correction, while for clean data, less smoothing yields a better resolved localized slope field.
We would like to comment here on the performance of the method for realistic cases containing a stack of layers, each with a different orientation of azimuthal anisotropy. The azimuthally-dependent traveltime variations caused by wave propagation in the upper layers will be superimposed on the reflection events corresponding to underlying layers. While inverting for NMO parameters is shown to be straightforward through the velocity-independent approach, solving for interval parameters would require these effects to be unraveled through the use of layer-stripping (Hake, 1986) or a Dix-type inversion (Grechka et al., 1999; Grechka and Tsvankin, 2002). If the effects from overlying layers distort later traveltime surfaces enough such that they are no longer elliptically hyperbolic as suggested by equation 1, then the moveout correction will not be complete for the entire section. However, as seen in the second example in the previous section, the velocity-independent moveout method can be used as a residual correction, with no changes to the procedure. The later events with incomplete moveout correction can therefore be corrected with iterated applications of the method. Another complication arises in the residual moveout case though, if one wants to extract parameters such as the azimuth angle or moveout slownesses. The equations presented here for parameter extraction were derived for a single pass NMO correction; it remains for further work to extend the parameter estimation methods to cases where residual moveout correction is necessary.
3D velocity-independent elliptically-anisotropic moveout correction |