GPU-acceleration 3D rotated-staggered-grid solutions to microseismic anisotropic wave equation with moment tensor implementation

Abstract

To improve the accuracy of microseismic inversion, seismic anisotropy and moment tensor source should be carefully considered in the forward modelling stage. In this study, 3D microseismic anisotropy wave forward modelling with a moment tensor source was proposed. The modelling was carried out based on a rotated-staggered-grid (RSG) scheme. In contrast to staggered-grids, the RSG scheme defines the velocity components and densities at the same grid, as do the stress components and elastic parameters. Therefore, the elastic moduli do not need to be interpolated. In addition, the detailed formulation and implementation of moment-tensor source loaded on the RSG was presented by equating the source to the stress increments. Meanwhile, the RSG-based 3D wave equation forward modelling was performed in parallel using compute unified device architecture (CUDA) programming on a graphics processing unit (GPU) to improve its efficiency. Numerical simulations including homogeneous and anisotropic models were carried out using the method proposed in this paper, and compared with other methods to prove the reliability of this method. Furthermore, the high efficiency of the proposed approach was evaluated. The results show that the computational efficiency of proposed method can be improved by about two orders of magnitude compared with traditional central processing unit (CPU) computing methods. It could not only help the analysis of microseismic full wavefield records, but also provide support for passive source inversion, including location and focal mechanism inversion, and velocities inversion.