Elastic reverse time migration without wavefield decomposition

Abstract

In this paper, we provide a clear definition represented by a concise expression for seismic data migration that is a linear inversion of subsurface exploration targets using primary waves given a smooth migration model with accurate wave kinematics. Combining the definition with noncurl characteristic of P-wave and nondivergence characteristic of S-wave, we propose an elastic reverse-time migration approach without wavefield decomposition applied to heterogeneous isotropic media, in which decoupled P- and S-wave propagators independent of each other are derived. We obtain ideal imaging results when we apply the proposed migration approach to elastic synthetic seismic data of a simple four-layered model and the complex Marmousi2 model with pure P- or S-wave source exciting, demonstrating the validity of our method. Because the proposed method does not include wavefield decomposition in the backward extrapolation of recorded wavefield and demand less spatial derivative numbers in the forward extrapolation of source wavefield, it is capable of achieving high-efficiency computing. In the case of P/S composite source exciting, we suggest that although the recorded wavefield can be divided into P- and S-wave component, the P-wave component consists of primarily reflected waves PP plus primarily converted waves SP and the S-wave component consists of primarily reflected waves SS plus primarily converted waves PS. Because PP and SP or SS and PS cannot be further separated from each other by current decoupling methods, the crosstalk artifacts are unavoidable, which is verified by the migration imaging results of a two-layered model.