Application of true-azimuth 3D SRME in the Northwest Shelf, Australia

3D general surface multiple prediction (GSMP) is an accurate and flexible implementation of 3D surface-related multiple elimination (SRME). We use a case history from the Northwest Shelf to demonstrate the effectiveness of the algorithm. A feasibility study explores the utility of reprocessing the existing data with 3D SRME, and improvement in velocity model delineation is predicted. Assuming that all traces have source-to-detector azimuths aligned with the nominal grid is an approximation exploitable when designing 3D SRME algorithms in order to simplify implementation and reduce cost. One of the major advantages of 3D GSMP over such methods is its ability to predict multiples corresponding to the true acquisition geometry. A comparison of results corresponding to multiples predicted with nominal- and true-azimuth geometries clearly shows the improvement in demultiple performance for the latter case. Compute cost is a concern for all 3D SRME algorithms, and is related to the sampling and aperture used for the multiple contribution gathers. We show how these parameters may be optimized to control the compute cost without degrading results. A substantial improvement in image quality is obtained following reprocessing with 3D GSMP, even where the multiples are not strong compared to the signal.