Analysis of ground penetrating radar data using hierarchical Markov Chain Monte Carlo simulation

Abstract

Ground penetrating radar (GPR) is a geophysical method used in highway maintenance to determine subsurface conditions within the right-of-way. GPR operates by using short-pulse radiation of radio-frequency electromagnetic energy to record dissimilarities in electrical properties of subsurface materials. As such, GPR results are susceptible to the transmission frequency used and the inherent properties of different subsurface materials. Uncertainty due to these susceptibilities can lead to ambiguity in the interpretation of GPR data. To distinguish heterogeneity from uncertainty, this paper modeled GPR data on pavement layer thickness using Markov Chain Monte Carlo (MCMC) simulation. MCMC is able to model heterogeneity within a given dataset and was employed to estimate and predict layer thicknesses obtained from GPR data. Simulated results were consistent with field data and provided statistical estimates of missing values in the original dataset. This analysis will aid relevant stakeholders to verify and determine consistency in field GPR data.