The zero‐velocity layer: Migration from irregular surfaces

Abstract

Seismic data acquired in areas with irregular topography are usually corrected to a flat datum before migration. A time‐honored technique for handling elevation changes is to time shift the data before application of migration. This simple time shift, or elevation‐static correction, cannot properly represent wide‐angle or dipping reflections as they would have been recorded at the datum. As a result, when elevation varies significantly, accuracy in event positioning may be compromised for migration and other wave‐equation processes, such as dip moveout processing (DMO). Traditionally, such over‐ and under‐migration artifacts have been dealt with by increasing or decreasing the migration velocity. However, simple adjustment of the migration velocity cannot undo the wave‐field distortions induced in seismic data acquired over varying elevations. More sophisticated and accurate solutions such as wave‐equation datuming are too computationally demanding for routine use. Here, we propose an efficient and accurate technique for doing migration from irregular surfaces using conventional migration algorithms. As in elevation‐static corrections, surface‐recorded data are time‐shifted to a horizontal datum; for our process, we choose to have that datum elevation lie at or above the highest elevation in the survey. After migration, the datum elevation can always be adjusted to any other level by means of a bulk time shift. In the migration step, the velocity is set to zero (or some very small value) in the layer between the surface and the datum; below the original surface, the interval velocity represents the best estimate of the subsurface geology. By adding a zero‐velocity layer, the migration algorithm is applied to the data from the flat datum and no lateral propagation is allowed until a nonzero velocity is encountered at the recording surface. Synthetic and field data examples demonstrate that use of the “zero‐velocity layer” significantly improves imaging accuracy relative to conventional migration from a flat datum. Moreover, the geologically derived migration‐velocity field need not be adjusted to compensate for shortcomings in the datum‐static procedure. The technique can be extended to prestack processes such as DMO, shot‐ and receiver‐gather downward extrapolation, and migration and thus suggests a unified approach to processing data from irregular surfaces.