Practical parallel implementation of the nonstationary integral wavefield extrapolators is presented. Two approaches, the windowed split-step algorithm and full integral The integral extrapolators are suitable for large-step algorithm, were implemented. The integral extrapolators are suitable for large-step wavefield extrapolation, even when lateral velocity variations are strong, provided that the static phase-shift and the focusing phase-shift velocities are replaced by time-average and depth-average velocities, respectively. Linear vertical wavefield interpolation between reference wavefields produced by large-step extrapolation algorithms can be used to compute the intermediate depth image. This dual algorithm significantly reduces the run time of the integral wavefield extrapolators.
Nonstationary extrapolators are found to be very capable of imaging shallow events at large dipping angles as well as at deep events, even when severe topography variations and high near-surface velocity is present. The Marmousi synthetic data set and the Alberta Foothills synthetic data set were migrated and superior depth images were achieved.
P-P prestack depth imaging techniques can be conveniently applied to converted-wave data based on the primary-only P-S wave propagation model. Prestack P-S shot gather migration with the dual algorithm produced a very good depth image for the 1997 Blackfoot 3C-2D survey.
The concept of downward-continuation migration velocity analysis (DMVA) is proposed. It can be used to partially eliminate the assumption of laterally invariant velocity in the established migration velocity analysis techniques and hence provide a better velocity estimation.
View full article as PDF (20.00 Mb)