Surface-related multiple attenuation with least-squares multidimensional deconvolution for P and S reflections in OBC/OBN data

Due to the presence of the interfaces at the ocean surface and bottom, the subsurface reflected/scattered waves are strongly influenced by surface-related multiples, posing a great challenge to the processing of multicomponent ocean-bottom seismic data, especially in velocity model building and seismic migration. Conventional demultiple methods for streamer data are not only challenged in processing ocean-bottom data, buthard to handle converted waves. Based on data-driven redatuming, we propose an regularized least-squaresinversion to solve the problem of up/down multidimensional deconvolution (MDD), according to the multidimensional convolution relationship among the up/down-going P or S waves and the impulsive reflections, so that the effectsof water layer and free surface on PP and PS reflections are eliminated. Synthetic data examples have illustratedthe mechanism of least-squares up/down MDD method, and demonstrated its precision advantage overconventional up/down deconvolution (UDD). These experiments have demonstrated the potential of the proposedapproach to improve imaging of the PP and PS reflections through attenuating the surface-related multiples,especially in the presence of complex geologic structures in shallow water.