Seismic imaging of incomplete data and simultaneous-source data using least-squares reverse time migration with shaping regularization

Simultaneous-source acquisition improves the efficiency of the seismic data acquisition process. However, direct imaging of simultaneous-source data may introduce crosstalk artifacts in the final image. Likewise, direct imaging of incomplete data avoids the step of data reconstruction, but it can suffer from migration artifacts. We have proposed to incorporate shaping regularization into least-squares reverse time migration (LSRTM) and use it for suppressing interference noise caused by simultaneous-source data or migration artifacts caused by incomplete data. To implement LSRTM, we have applied lowrank one-step reverse time migration and its adjoint iteratively in the conjugate-gradient algorithm to minimize the data misfit. A shaping operator imposing structure constraints on the estimated model was applied at each iteration. We constructed the shaping operator as a structure-enhancing filtering to attenuate migration artifacts and crosstalk noise while preserving structural information. We have carried out numerical tests on synthetic models in which the proposed method exhibited a fast convergence rate and was effective in attenuating migration artifacts and crosstalk noise.