Ground-penetrating-radar reflection attenuation tomography with an adaptive mesh

Abstract

Ground-penetrating radar (GPR) attenuation-difference analysis can be a useful tool for studying fluid transport in the subsurface. Surface-based reflection attenuation-difference tomography poses a number of challenges that are not faced by crosshole attenuation surveys. We create and analyze a synthetic attenuation-difference GPR data set to determine methods for processing amplitude changes and inverting for conductivity differences from reflection data sets. Instead of using a traditional grid-based inversion, we use a data-driven adaptive-meshing algorithm to alter the model space and to create a more even distribution of resolution. Adaptive meshing provides a method for improving the resolution of the model space while honoring the data limitations and improving the quality of the attenuation difference inversion. Comparing inversions on a conventional rectangular grid with the adaptive mesh, we find that the adaptively meshed model reduces the inversion computation time by an average of 75% with an improvement in the root mean square error of up to 15%. While the sign of the conductivity change is correctly reproduced by the inversion algorithm, the magnitude varies by as much as much as 50% from the true values. Our heterogeneous conductivity model indicates that the attenuation difference inversion algorithm effectively locates conductivity changes, and that surface-based reflection surveys can produce models as accurate as traditional crosshole surveys.