Sparse least-squares reverse time migration of small scatters in seismic exploration

Seismic exploration has a great potential in exploration of mineral resources in the deep subsurface, while also faces many challenges. Due to the steep structure, small-scale scatters and small differences of physical properties between the background and the orebody, the conventional migration method has a low imaging resolution for small-scale orebodies. In this paper, we propose a least-squares reverse time migration based on a sparsity-promoting constraint. First, the non-uniformly distributed orebodies are assumed to be equivalent to a random medium, and a multiscale orebody model for such a medium is established. Second, we improve the existing Least-Squares Migration (LSM) by using the sparse constraint as the priori information, and then compress the imaging space by Curvelet transform. After iterative calculations, the imaging resolution of small-scale scatters can be enhanced. Third, the random source encoding is used to reduce the number of gathers for LSM, which speeds up the computation. By the sparsity-promoting constraint, the imaging error caused by crosstalk noise is reduced greatly. Numerical calculation on the Luzong metallic deposit model indicates that the proposed method can image the metallogenetic geological model containing small-scale scatters with a relatively high resolution.