GPU-accelerated decoupled P-wave reverse time migration for 3D ellipsoidal anisotropic media

The elastic wave equations can accurately describe the kinematic and dynamic characteristics of seismic waves in subsurface formations, but it requires a lot of calculation time and memory usage. To address the challenges of low computational efficiency, large memory consumption, and strong P-S wave coupling in solving the elastic wave equations, we start from the elastic wave equations for 3D VTI media and solve the eigenvalues of the elastic Christoffel matrix. By applying elliptical approximation, we derive a set of first-order velocity-stress equations for decoupled P-, SV-, and SH-waves in ellipsoidal anisotropic media. The forward modeling results of the decoupled equations show that this approach is suitable for both weakly and strongly anisotropic media as P-, SV-, and SH-waves can be completely decoupled and propagate independently. Based on GPU acceleration, a high-efficiency reverse time migration algorithm is developed for decoupled P-waves in ellipsoidal anisotropic media. Numerical tests verify the computational efficiency and imaging accuracy of the proposed method.