We have applied 2-D wave equation prestack depth migration to a long-offset dataset acquired over the flood basalts of the Faeroes-Shetland Basin. We attempted to migrate PSSP, PSPP, and PPSP converted modes by making a suitable replacement of compressional wave velocity with shear wave velocity. In spite of very low subsalt signal-to-noise ratio, we demonstrate that our use of angle domain common image gathers (ADCIGs) facilitated the successful discrimination between a probable PSSP converted wave and multiples in the migrated domain. Still, more investigation and careful processing is required to determine the validity of that interpretation.
Looking to the future, we see many avenues toward improvement. As noted by Kostov et al. (2000) the strong cable feathering in the acquisition of this data contributed to incomplete suppression of multiples at far offsets. Since we used the theoretical amplitude-versus-angle behavior to discriminate between converted waves and multiples, it is crucial either to suppress multiples at all offsets or to image with the raw data.
To improve the performance of the SRME algorithm, we plan to apply azimuth moveout (AMO) Biondi et al. (1998) to correct for the cable feathering. Although AMO is designed for use with 3-D acquisition geometries, we can make a first order correction for the feathering by assuming a 2-D earth.
As normally applied in the industry, SRME attenuates multiples from the free surface only. As mentioned in the body of the paper, we noted that the timing of an energetic internal multiple may have been misinterpreted as the base of basalt. At the very least, we need to time-migrate the raw data, as this domain is superior for interpretation of multiples. Next, we believe that careful wave equation or ray-based modeling is crucial to the identification of these more complex multiples. Alternatively (or additionally), if we construct an excellent velocity model of the region above the top basalt, we should be able to migrate the internal multiples.
The strong velocity contrast over the top basalt makes prestack depth migration a must. Unfortunately, the same velocity contrast also seriously inhibits construction of a velocity model, on which depth migration is notoriously dependent. In order to obtain rough bounds on sub-basalt velocity and structure, we may make use of refracted modes. Fliedner and White (2001) demonstrate the use of diving waves (overturning rays), in conjunction with reflection data, to determine basalt thickness and velocity gradient. We also need to investigate the validity of the simplistic substitution we made for shear wave velocity (constant Vp/Vs ratio). While the simplicity of a single-parameter model is appealing, it is likely unrealistic, given the complexity of most basalt flows. Lastly, migration velocity analysis for the asymmetric modes is poorly defined. The recent work of Rosales and Biondi (2001) on Ocean Bottom Cable PS data shows promise to this end.
Although we record a single wavefield, the individual modes within that wavefield may be viewed as multiple, possibly independent, sources of data. Lu et al. (2000) demonstrate excellent results from the prestack separation, then depth imaging, of converted waves under salt. Taking this philsophy beyond validation of interpretation of base basalt, we can think of having multiple data sources (prestack separation of converted waves and multiples) which produce a single, self-consistent image (obtained by least squares optimization).