Stochastic Analysis of Cross-hole GPR Data for Subsurface Characterization

Ground penetrating radar (GPR) is a well-established geophysical technique, used in particular via specific relationships to estimate hydrological parameters in vadose zone, i.e. moisture content, both from the surface and in boreholes. The principal aim of this work is to have a more complete view of how boreholes GPR ZOP measurements are informative of the subsoil geometry and distribution of relative permittivity. For this purpose an electromagnetic (EM) wave simulator has been applied within a stochastic Monte Carlo framework. In this manner both averaging and critically refracted wave effects are taking into account. Results from synthetic and real ZOP datasets are statistically analysed to deduce what kind of subsoil εr-distributions are resolvable and well defined with a degree of uncertainty. The results illustrate how the often employed direct-wave approach is not able to take into account the complexity of the system. It usually reconstructs a smooth profile, sometimes underestimating the real permittivity: particularly a high εr layer might be underestimated, when it is enclosed between low εr media. Results show that care must be used inverting ZOP data for physical parameter estimation, subsurface stratification could be more complex than appears from direct ZOP evidences.