Conclusions and Future Work

By estimating a pair of 1D PEFs in a warped data space, the sparse, curved tracks of the Madagascar dataset were successfully interpolated. Other methods that require a starting guess or rescaled proxy data were much less successful, as they were estimated in a domain where the data are not optimally distributed.

The tracks in the Madagascar dataset have several similarities with the tracks seen in 3D seismic data, such as sail lines, receiver cables, and cut lines for sources. When estimating PEFs on this data, the choice of doing it in the model space and the data space needs to be investigated.

To obtain a better result for this data, the next obvious step is to use non-stationary PEFs, as the character of the Madagascar data changes greatly with position.