3-D TIME-VARIANT DIP MOVEOUT BY THE F-K METHOD

The standard seismic imaging sequence consists of normal moveout (NMO), dip moveout (DMO), stack, and zero-offset migration. Conventional NMO and DMO processes remove much of the effect of offset from prestack data, but the constant velocity assumption in most DMO algorithms can compromise the ultimate results. Time-variant DMO avoids the constant velocity assumption to create better stacks, especially for steeply dipping events. Time-variant DMO can be implemented as a 3-D, f-k domain process using the dip decomposition method. Prestack data are moved out with a set of NMO velocities corresponding to discrete values of in-fine and crossline dips. The dip-dependent NMO velocity is computed to remove the trace offset and azimuth dependence of event times for an arbitrary velocity function of depth. After stacking the moved out CMP gathers, a three-dimensional (3-D) dip filter is applied to select the particular in line and crossline dip. The final zero-offset image is obtained by summing all the dip-filtered sections. This process generates a saddle-shaped 3-D impulse response for a constant velocity gradient. The impulse response is more complicated for a general depth-variable velocity function, where the response exhibits secondary branches, or triplications, at steeper dips. These complicated impulse responses, including amplitude and phase effects, are implicitly produced by the f-k process. The dip-decomposition method of 3-D time-variant DMO is an efficient and accurate process to correct for the effect of offset in the presence of an arbitrary velocity variation with depth. The impulse response of this process implicitly contains complex features like a 3-D saddle shape, triplications, amplitude, and phase. Field data from the Gulf of Mexico shows significant improvement on a steep salt flank event.