Resolving Infiltration‐Induced Water Content Profiles by Inversion of Dispersive Ground‐Penetrating Radar Data

Core Ideas Dispersive GPR data can resolve 1D water content distributions in waveguide layers. Blocky‐layer model inversions of the data provide good data fit. Blocky‐layer model inversions are unrealistic in a hydrologic framework. Independent piecewise‐linear model inversions provide improved data fit. Piecewise‐linear models honor smooth distributions of water content. Ground‐penetrating radar (GPR) data were collected before, during, and after a 24‐min‐long forced infiltration event in a large sand tank. High spatial and temporal resolution were achieved by automation of the radar system, thereby allowing these data to be collected during the course of the experiment while continuously changing the distance between the antennas through offsets ranging between 0.17 and 2.17 m. These multi‐offset data showed evidence of a phenomenon known as waveguide dispersion during early infiltration times (5–10 min), indicating that a shallow layer of high water content was present. The GPR data exhibiting this dispersive behavior were used to fit water content profiles for the wetting front, i.e., the waveguide, with time using either a blocky‐layer model or a piecewise linear function. Results from the separate inversions showed good agreement with in situ soil moisture measurements and a calibrated unsaturated flow model. The piecewise linear model, however, was able to honor the gradational nature of the hydrologically induced waveguide and was in better agreement with the observed soil moisture data. Furthermore, the piecewise linear model returned a water content profile that showed a consistent progression of the wetting front with time, whereas a less consistent progression of the wetting front was observed for the blocky‐layer model.