Compressive sensing seismic acquisition by using regular sampling in an orthogonal grid

Abstract

Seismic survey acquisition permits capturing subsurface data by sensing the seismic waves induced by an artificial source. Hundreds of kilometers are sensed at a sampling rate that satisfies the Nyquist/Shannon theorem to avoid signal aliasing, this means that a high-density arrangement of sensors is required. In seismic, a compressive seismic imaging (CSI) framework has been developed. To test CS theory, random sampling or simultaneous shooting techniques are applied to marine and land environments. For land, random acquisitions require creating new paths on the surface to place each source and receiver, additionally, for terrains with complex access, the artificial used sources are made of dynamite. For this reason, random acquisitions have an elevated environmental impact compared to regular acquisitions, where the same path is used to locate all the sources. This work proposes to use regular sampling (which is not a traditional sampling technique to be used with CS concepts) and to remove sources in a specific configuration present in orthogonal grids with CS concepts in order to reduce acquisition costs and environmental impact. The seismic wave data that should be induced by the removed source is reconstructed using a proposed modified iterative hard thresholding (IHT) algorithm that favors structural similarities of the data. Simulations were performed on real data to illustrate the accuracy of the proposed method, using the Curvelet transformation basis, which attains reconstructions 50% faster than Wavelets.