LEAST-SQUARES REVERSE TIME MIGRATION OF PURE QP-WAVE IN ANISOTROPIC MEDIA USING LOW-RANK FINITE DIFFERENCE

The pseudo-acoustic least-squares reverse time migration (PA-LSRTM) is often used for imaging of anisotropic media. Due to acoustic approximation, it, however, shows severe instability in the forward simulation, strong quasi-SV (qSV) wave residual in the demigration record, and terrible numerical dispersion in tilted transversely isotropic (TTI) media. The low-rank finite-difference (LFD) approach can effectively overcome these problems, but existing research only focuses on forward modeling, and no examples are found in LSRTM. For the first time in this paper, we derive the pure qP-wave linearized forward modeling and migration operators in TTI media with the help of LFD. Then, we implement pure qP-wave least-squares reverse time migration (LFD-LSRTM) in the inversion scheme. To improve the inversion efficiency, the plane-wave encoding technique is used, and to increase its robustness, the prestack parameterization is adopted. Finally, we obtain the prestack plane-wave least-squares reverse time migration (LFD-Pre-PLSRTM). Examples demonstrate that our method provides significant advantages in imaging TTI media, yielding satisfactory results with less expensive computation and more stable convergence compared to PA-LSRTM. More importantly, the proposed method can successfully avoid troubles caused by the acoustic approximation, and reasonably allow errors in the parameter model and noise in the data, making it possible to deal with real data.