A new method for the migration velocity analysis of noisy seismic data

Abstract

We present a method for the determination of optimum migration velocities that is especially suited to datasets, such as deep crustal seismic data, in which both incoherent and coherent noise are present. Slowness weights are evaluated for each point in a time section in order to separate coherent and incoherent noise from signal, which can be identified by its characteristic slowness. These weights are calculated from slant stacks in a single pass through the data. Thereafter the weights may be summed along suites of diffraction curves corresponding to different velocity models to produce coherency estimates for each subsurface point as a function of velocity. After contouring these functions optimum migration velocities may be picked. Synthetic examples are presented illustrating the improved efficiency and velocity sensitivity of this method over previous ones, and showing the applicability of the technique to pre-stack, post-stack, and converted phase migration. A real example taken from a marine survey is used to demonstrate the utility of applying the method to deep crustal data.