Surface related multiple elimination (SRME) is a powerful data driven tool to remove surface related multiples. However, it has a very strict requirement on the acquisition geometry to obtain a satisfactory result. The shortcomings of the method due to inadequate acquisition are enhanced if there are complex multiple generators in the subsurface. In this paper, I propose a simple methodology to perform antialiasing in the multiple contribution gather (MCG) domain to reduce the artifacts generated in the SRME predicted multiples.
SRME is a prediction and subtraction method (see, e.g., Verschuur and Berkhout, 1997, and Weglein et al., 1997). Surface related multiples are predicted using the appropriately preprocessed input data and then subtracted from the dataset without multiple attenuation. Written explicitly, the 3D SRME prediction is the sum of autoconvolutions of the data given by the equation.
Here M is the predicted multiple for a trace with source location (xs,ys) and receiver location (xg,yg). D is the input data. The summation is performed over the defined (x,y) aperture. The MCG contains traces generated by these individual autoconvolutions of the data D, prior to summation.
SRME is a very popular and effective algorithm for removing surface related multiples. Ideally, this algorithm requires seismic sources at every receiver location. This prerequisite is not satisfied for field datasets. Figures 1 and 2 (modified from Verschuur, (2006)) clearly demonstrate the artifacts generated in the predicted multiples due to inadequate acquisition. Figure1a is the MCG generated for a single trace using an adequate source spacing. Figure 1b is the MCG generated for the same trace using double source spacing. The spatial aliasing is very noticeable in figure 1b. Comparison of the predicted multiples in figure 2, created by the summation of the MCG, clearly shows the artifacts generated due to inadequate sampling of the source.