A 3-D Parallel Method of PSTD-Based Reverse Time Migration of GPR Data

Ground penetrating radar (GPR) is a nondestructive technique with broad applications. Reverse time migration (RTM) of GPR data plays an important role in imaging complex structure. However, the three-dimensional (3-D) RTM of GPR data still faces big challenge due to its expensive computation cost. In this article, we develop a novel RTM method based on a 3-D parallel pseudospectral time domain (PSTD) algorithm. We apply our new RTM method to the synthetic fault model. Testing results based on the above synthetic GPR data indicate that the new RTM method is robust and effective in imaging various anomalies including faulting and layered structure. We further validate our new method based on the field GPR data across an active fault. The results demonstrate its effectivity in imaging active fault from the field data. Numerical tests also show quantitatively that the computation time and efficiency of the proposed parallel PSTD-based RTM method are significantly superior to those of the RTM algorithm based on the popular parallel finite difference time domain (FDTD) method. Comparing to the FDTD-based method, our PSTD-based method can significantly weaken the numerical dispersion effects and enhance the imaging resolution. The above superiority in computation efficiency and imaging robustness demonstrates that our new method has great potential in broadening the GPR applications, especially the cases for quasi-real-time detection and complex structure imaging.