The purpose of Dip Move Out is to map a nonzero offset dataset into a zero offset dataset. After migration both data sets should result in the same subsurface picture.
The introduction of the artificial offset k splits the `stacking NMO' step into a velocity independent DMO step and a velocity depending standard NMO. The NMO step is unbiased with respect to dipping events. All events regardless of reflector dip line-up and contribute to the stacked image constructively. This improves NMO velocity analysis and the poststack image.
The two-dimensional convolution is the only cross-trace calculation in the proposed algorithm. Due to its strongly parallel features the DMO procedure is inexpensive, when performed on a parallel machine.
Forel 1988 demonstrated that his DMO process extends to three-dimensional data sets. Following Forel's approach, the proposed two-dimensional process can be extended to three dimensions.