The recent pattern-based techniques in the literature are approximately equivalent to Wiener optimal estimation Castleman (1996) since they utilize the PEF to approximate the signal and noise power spectra. For instance, Spitz (2000) uses f-x domain PEF while Brown and Clapp (2000) and Clapp and Brown (2000) use t-x domain PEF.
In this report, Guitton (2001) presents a method that is not based on the Wiener reconstruction of the signal but that still uses t-x domain PEF. Following Guitton's idea, we use the PEFs as prediction operators as opposed to filtering operators in the Wiener approach. Nonetheless, this method belongs to the pattern-based type since PEFs are still estimated for the noise separation. Our goal is to show that this new methodology leads to a proper attenuation of ground-roll and multiples, and has the potential to rival the classical Wiener formulation.
In the first part of this paper, we briefly review the theoretical developments of both the Wiener-like scheme and the new proposed technique, discussing their differences and similarities. Then, we compare the two strategies applied to a 3-D shot gather infested with ground-roll, and multiple-infested CMP gathers in a region of complex geology.