A High-Efficiency Spectral Element Method Based on CFS-PML for GPR Numerical Simulation and Reverse Time Migration

Improving the accuracy and efficiency of numerical simulation ground penetrating radar (GPR) becomes a pressing need with rapidly increased amount inversion data growing demand for migration imaging quality. In this article, we present spectral element time-domain (SETD) procedure GPR forward modeling further apply it to reverse time (RTM) complex geoelectric models. This approach takes into account flexibility finite methods high precision methods. Meanwhile, in procedure, frequency shifted perfectly matched layer (CFS-PML) is loaded effectively suppress echo at truncated boundary, per-element GPU parallel framework used can achieve up 5.7788 times compared CPU calculation. The experiments on SETD spatial convergence CFS-PML optimal parameter selection showed that, under same degree freedom, offered substantially better traditional FETD. RTM different profiles orders via model verify universality algorithm. results indicate that effect has been significantly improved increase order. It fully proves great potential efficient high-precision algorithm direction shows certain guiding significance underground target structure exploration.