The methodology introduced in this paper targets those multiples that are not easily removed with known techniques, for example when stacking normal-moveout corrected seismic gathers does not eliminate all multiples Foster and Mosher (1992). In a previous report, Sava and Guitton (2003) introduce a technique to remove multiples after imaging. Their technique consists of applying the Radon analysis in the depth-angle domain instead of the standard Radon analysis done in the offset-time domain. Their main assumption is that primaries and multiples migrated with the correct velocity model for the primaries and transformed into angle domain common image gathers, map to distinctive angles, where primaries are flat and multiples present a residual curvature. Finally, the angle gathers are mapped into the Radon domain, where the multiples can be suppressed.
A potential issue with the previous methodology is the coupled problem of velocity analysis and multiple suppression. It is difficult to obtain an accurate velocity model when multiples are present. Moreover, the previous methodology relies on an accurate velocity model in order to suppress the multiples.
This paper proposes a solution for solving this problem. Our methodology combines two post-migration processing techniques. The first one, SRM, produces an intermediate image with a new velocity information given, by the ratio () between the original migration velocity and the new migration velocity; therefore, SRM induces a curvature change between primaries and multiples if is significantly different than 1. The second technique, MAIS, separates the primaries and multiples into two different groups, based on their distinctive curvature induced by SRM. This allows us to model the multiples and suppress them from the final section.
Figure shows the basic flow chart to eliminate multiples in the image space. The input data are our best result after processing in the data space, and after prestack migration with our best velocity model at this stage. Such processing may vary and depends on the data set under study. The next step is to perform SRM with a velocity ratio high enough to produce a difference in curvature between the primaries and the multiples (e.g., ). The difference in curvature yields to the application of the Radon transform in the angle domain to eliminate multiples Sava and Guitton (2003). Finally, we apply SRM with a ratio that should be the inverse of the ratio applied before (e.g., ).
Figure 1 Flowchart for multiple suppression in the image space with no flat events in the angle domain.