Prestack Rank-Reduction-Based Noise Suppression

Prestack random-noise suppression is an important but inadequately-solved problem in land seismic processing, with the potential to significantly improve AVO analysis and the stacked section. To address this problem, we describe a family of filters that perform matrix-rank reduction on constant-frequency slices. These filters include two existing techniques – eigenimage and Cadzow filtering – as well as a novel hybrid method with properties from both. This general class of noise suppressors is powerful, versatile, and can be applied in any number of spatial dimensions. We explain the theory behind these filters, and demonstrate how to apply them in practice.