Adaptive curvelet-domain primary-multiple separation

Theory

The proposed separation method consists of two stages. During the first adaptive stage, the predicted multiples, ${\breve{\mathbf{s}}}_2$ , are fitted through a correction operator to the multiples present in the total data, $\mathbf{p}=\mathbf{s}_1+\mathbf{s}_2$ , which consists of the sum of primaries, $\mathbf{s}_1$ , and multiples, $\mathbf{s}_2$ . During the second stage, the primaries and multiples are separated by a thresholding procedure, defined in terms of the scaled magnitudes of the curvelet coefficients of the predicted multiples. Since SRME-predicted multiples are used as input, the wavelet and source directivity will not be properly compensated (Verschuur et al., 1992).