CRS stack using global simultaneous multiparameter optimization - A spatial velocity analysis for near-surface data

CRS stacking employs a spatial three-parameter stacking operator that extents in offset and midpoint directions. Since this operator covers several CMP gathers, a large number of traces contribute to every single stacking process, thus generating a large improvement in signal-to-noise ratio. However, the standard implementations use the spatial operator only for stacking and not for the global stacking parameter search. Instead, they rely on a sequence of three one-dimensional searches which decreases the computational effort compared to the simultaneous three-parameter search by two orders of magnitude, but does not always deliver optimal results. Particularly for near-surface data, characterized by low signalto- noise ratio and modest CMP fold, the efficiency gain is not crucial. Considering the huge computing power available today, we propose a new pragmatic search strategy using a spatial two-parameter diffraction operator for the global search, followed by a local optimization using the full CRS operator with the diffraction parameters as initial guess. For shallow shear-wave data, we show that while the computational cost of this hybrid approach reduces by an order of magnitude the results are still very close to those obtained by the full global three-parameter search and far superior to those obtained by three cascaded one-parameter searches