Articles from the archive: Common-offset migrations and velocity analysis

Abstract

Prestack depth migration is very sensitive to velocity errors. This can make it an excellent tool for velocity analysis. However, to get maximum benefit from this Sensitivity to velocity we must choose our algorithm carefully. Most of the sensitivity to velocity arises from the normal moveout (NMO) term, i.e. from the alignment corrections to data recorded at different offsets. It therefore makes sense, in the context of velocity analysis, to design the prestack migration so that the overall procedure does not apply NMO according to a preconceived velocity model. Instead, the NMO should be calculated after migration in the standard way by fitting hyperbolas to common-midpoint gathers. Unfortunately, the strict application of such a design criterion eliminates most of the standard migration techniques, such as datuming, downward continuation and reverse time migration. It also eliminates shot record imaging by integral methods. I do not mean to imply that such imaging methods cannot be used for velocity estimation, but simply that the initial velocity model has to be significantly more accurate than that required if the migration technique does not mix data recorded at different offsets. For example, shot-record dip moveout (DMO) requires a significantly more accurate initial velocity estimate than common-offset DMO.