Surface Related Multiple Elimination for WATS Data

A version of the 3D SRME algorithm for the WATS data is considered. In this version, missing traces required for 3D multiple prediction are reconstructed on-the-fly based on the azimuth dependent NMO for a few selected horizons. The subtraction scheme with multichannel multidimensional filters averaged over a streamer, which avoids re-sorting of traces, proved effective for such data. The use of the synthetic data, generated from a realistic model and with the acquisition pattern corresponding to the real case, helps to analyze strong and weak points of the studied demultiple scheme. The results show that the suggested algorithm is effective for the WATS data, especially at near-to-middle offsets. As for far offsets combined with complex geological settings, there is still a room for improvements. An effective data handling scheme for scalable implementation of the algorithm for compute clusters is suggested, where receiver-side trace are accessed in the sequential order and processed according to the pre-computed index tables of trace contributions.