Seismic imaging using matched filters for operator weighting

Abstract

Given a particular statistical measure of signal to noise (S/N), the 'matched filter' is an ideal linear filter for maximizing the S/N ratio of a signal amongst random, white noise. A matched filter approach to event detection in prestack migration is proposed, where 'signal' is defined as a particular amplitude variation with offset (AVO) reflection coefficient surface, and all other AVO response surfaces are considered 'noise'. Matched filtering of the prestack data with the signal illuminates reflection events whose AVO response curve matches that of the signal; other reflection energy is suppressed. Matched filter imaging enhances the detection of Class 2 AVO events, and results in an overall S/N improvement over conventional P-SV wave prestack migration.