SEPARATION AND SHOT INTERPOLATION OF SIMULTANEOUS-SOURCE DATA WITH INTERPOLATED MSSA (I-MSSA)

We present an algorithm for simultaneous data deblending and source interpolation. The algorithm adopts the projected gradient descent method with a denoiser that acts as a projection to iteratively deblend and reconstruct shots onto a regular grid. We study two problems. Data are assumed to lie on a regular grid with missing shot positions in the first problem. These data were obtained by simple nearest neighbour interpolation (binning). The Multichannel Singular Spectrum Analysis (MSSA) filter is used as the denoiser. In the second case, we honour true spatial shot positions and adopt Interpolated MSSA (I-MSSA) as the denoiser. We show the superior performance of the deblending and reconstruction algorithm when the I-MSSA projection is adopted. In essence, we propose an algorithm that allows one to deblend and reconstruct shot positions. Hence, one can achieve extra acquisition savings by blending sources and source decimation.