f Multiple attenuation by predictive deconvolution in the p-t domain

Many multiple attenuation or suppression methods depend on discrimination of the moveouts of reflections in unstacked data. Most of these methods, such as NMO stacks and f-k dip filters, are not very effective when primary and multiple reflections have similar moveouts and stacking velocities. Predictive deconvolution in the p-tau domain does not require such velocity discrimination. Land reflection seismic data from North Africa contains many strong multiples with essentially the same moveout as weak underlying primaries. Predictive deconvolution successfully attenuated these multiples and improved the interpretation of the subsurface.