Separating multiples from primaries has been a long-standing problem in exploration geophysics. Multiple reflections often destructively interfere with the primary reflections of interest Foster and Mosher (1992). Marine data often exhibits strong multiple-reflection problems. However, Ocean Bottom Seismic (OBS) data presents an ideal acquisition setting for removing the receiver ghost by PZ summation Barr and Sanders (1989); Soubaras (1996). However, the PZ summation does not remove all the multiples, and further processing is required.
This paper combines two post-imaging processing techniques: 1) Stolt Residual Migration (SRM) Sava (1999, 2000), and 2) Multiple Attenuation in the Image Space (MAIS) Sava and Guitton (2003). Integrating these two processing techniques in one process leads to a way of eliminating multiple reflections in areas with unresolved velocity problems. Stolt Residual Migration is used in the prestack domain to obtain an image focused with a different velocity model.
Sava and Guitton (2003) discuss multiple suppression in the image space (i.e., after migration) as an alternative technique to multiple suppression in the data space (i.e., before any imaging operation). Their main assumption is that after migration with the correct velocity model, primaries are flat in angle-domain common-image gathers, while multiples present a residual curvature; therefore, primaries and multiples can be separated using the Radon transform Foster and Mosher (1992). However, if we yet do not have the correct velocity model, the primaries and multiples may present a similar residual moveout in the angle domain, and they will not be distinguishable by their curvatures using the Radon transform.
The problem we address in this paper is how to discern between primaries and multiples in the image space if the events have similar curvatures. We present a real data example, the Mahogany field, a 2-D OBS data set in the Gulf of Mexico.