Perturbed prediction-driven multiple elimination: A PEF scheme
Simon Spitz · 2007
Primary-multiple separation is a two steps process. At the first step the multiples are predicted. The prediction can be made in various ways, some better suited than others for the problem at hand, and a lot of research has been done on the subject. At the second step the multiples are extracted from the input dataset. Multiples are usually eliminated following a least squares matched filtering approach. Because the prediction and elimination stages are carried out independently, any imperfection generated at the first step is felt at the second. When the predicted multiple wavefield is perturbed with respect to the one that is actually present in the data the result of the matching is sub-optimal. This paper presents a subtraction scheme that is able to cope with such perturbed prediction.