3D multi-source least-squares reverse time migration based on L1 norm regularization

Compared with the conventional migration method, least-squares reverse time migration (LSRTM)has many advantages including, for example, higher imaging resolution, amplitude preservation and amplitude balance. LSRTM algorithm based on first order velocity-stress wave equation is able to handle the medium with variable density and has certain advantages in suppressing numerical dispersion, but is mainly applied to two-dimensional media. In order to extend the scope of the method, the first-order velocity stress equation LSRTM algorithm is extended to three-dimensional. Giventhat multi-source method will introduce high frequency crosstalk noise, L1 norm sparse regularization constraints are used to suppress the crosstalk noise caused by phase encoding. Numerical tests on synthetic data demonstrate that the phase encoding algorithm can significantly reduce the computational effort and at the same time improve the computational efficiency. The L1 norm regularization can effectively suppress the low and high frequency noise, improving imaging resolution and the image quality. Lastly, the linear Bergman solution reduces the dependence of the inversion results on model parameters and is therefore suitable for processing of field data. © 2019, Periodical Office of China University of Petroleum. All right reserved.