Inverse Methods to Improve Accuracy of Water Content Estimates from Multi-offset GPR

Abstract

Ground penetrating radar (GPR) is a powerful hydrogeophysical tool for estimating porosity and water content of geologic materials using radar wave velocities and appropriate petrophysical relations. In substrates with more than one layer of interest, surface-based multi-offset measurements require careful analysis to accurately retrieve physical properties for each layer. Frequently, Dix inversion is used to calculate interval velocities, however the assumptions and limitations of this approach are widely known. In particular for survey geometries and targets encountered with GPR, the assumptions inherent to Dix inversion are readily violated, and therefore inverse modeling is required to avoid velocity error. While the impact on velocity incurred by violating the assumptions of Dix inversion is well understood, the effects on water content estimates have not been widely reported and are therefore the subject of this work. In a subsurface representative of an unsaturated zone overlying an aquifer, error in excess of 50% in water content due to violating the assumptions of Dix inversion is demonstrated. Examples are shown using raytracing inversion to solve for subsurface water content structure that avoids the errors inherent to Dix inversion. These results are intended to underscore the importance of minimizing assumptions and using more correct physics when analyzing multi-offset GPR data, particularly due to the large potential errors that may be encountered if water content estimation is the main objective.