The use of sensible velocities for migration

Abstract

Conventional seismic processing is unable to properly image complex geological structures such as those found on the North West Shelf of Australia. Lateral variations resulting from faulting hinder complete imaging of the underlying structures. Prestack depth migration is the correct way to image such structures but requires very accurate velocity field information in order to do so. Prestack time migration and post-stack migration are quicker methods but they make assumptions that are quite often invalid. Rules are needed to identify the cases when simpler methods will work, and to decide the velocity models needed for each type of migration. A synthetic seismic dataset was developed, based on the Oliver line from the North West Shelf of Australia, containing lateral velocity contrasts of 60%. Both time and depth algorithms were used to assess the amount of smoothing the velocity model was able to tolerate while still remaining suitable for interpretation. Migrating with a smoothed velocity can cause two undesired effects. Events can be mispositioned vertically, and their amplitudes can be degraded. For different amounts of smoothing these two effects can be measured as a function of lateral position to evaluate the performance of the migration on the Oliver model. Depth errors of 70 m and a 60% reduction in amplitude for a reflector at 2800 m resulted when the velocity was smoothed laterally over 200 m in prestack depth migration. The same smoothing resulted in a 40 ms time error with 50% amplitude reduction for prestack time migration. Rotating the smoothing to the vertical direction produced a 45 m depth error and an amplitude reduction of 40% for prestack depth migration. Prestack time migration produced results similar to those in the vertical direction.