Conclusions and Future Work

Overall, estimating a non-stationary prediction-error filter with multiple scales of data appears to be successful. The method interpolates a very heavily decimated 2D test case successfully. Results for a 3D case are even more successful, even though the amount of data removed from the case was greater than in the 2D case.

In the future, this method can be used on real seismic data in two, three, or even five dimensions, so that prestack 3D data can be interpolated in cases where surface topography, structures, or other obstacles cause irregularly sampled data.