| Three experiments were conducted to measure the ability of ground-penetrating radar (GPR) to non-invasively determine water content while simultaneously resolving depth to wetting fronts, buried objects, and stratigraphic boundaries during dynamic hydrologic conditions. This is particularly appealing as GPR can provide dense spatial coverage for vadose zone characterization where traditional invasive measurements are costly, destructive, and time-consuming. The vadose zone was replicated using a tank filled with (1) homogeneous river sand, (2) homogeneous river sand with an embedded land mine surrogate, and (3) homogeneous river sand with an embedded layer of silica flour. These systems were subjected to controlled irrigation events and monitored with GPR using automated time-lapse wide-angle reflection refraction (WARR) surveying. The unique form of data collection allowed the data to be conceptualized into a 3D data cube, providing multi-offset projections to extract wave velocities for depth and average water content measurements and transient common-offset projections to observe changes in amplitude and traveltime of arrivals over time associated with the fluctuations in average water content of the tank.;Average water content estimates from ground-penetrating radar were similar to in-situ capacitance probe measurements for the homogeneous tank experiment. Radar estimates of depth to wetting front and bottom of the tank, however, were found to have some issues associated with wave interference, causing errors in the range of 1--25%, with the largest errors occurring at times of infiltration. It was concluded that GPR has potential, through transient multi-offset imaging of the subsurface, to greatly improve vadose zone characterization by imaging the subsurface, quantifying water content, and tracking wetting fronts as they move through the media.;The layered experiment revealed that the silica flour greatly inhibits vertical flow of water causing significant changes in the GPR response through time when compared to a similar homogeneous experiment. At initial conditions, the radar data resembled that of a single layer system; however, as the water content increased, reflections and multiples from the upper layer dominated the image, degrading the interpretation of the system and clearly illustrating that interpretation of GPR data can be affected by the hydrologic state of the subsurface.;The land mine experiment showed that the unsaturated flow of water was not affected by the land mine and closely resembled the hydrologic response of the homogeneous tank. While the land mine signal was unclear on the GPR data, differences in amplitude vs. offset relationships between groundwave arrivals for the land mine and homogeneous tank indicate that significant changes in amplitude occur which may assist present methods for landmine identification. The data also showed that high water content values, such as after a rainfall event, provide a more favorable environment for landmine identification, as the groundwave is highly attenuated, reducing wave interference. While valuable data was collected, WARR surveying of the land mine may be secondary to common offset or common mid-point surveying as the land mine was not clearly visible on the WARR data, however, more robust signal processing of WARR data may also improve data interpretation. In conclusion, these experiments have illustrated that more reliable images, water content estimates, and overall characterization of the subsurface will be attained by the transient monitoring of the subsurface with surface based GPR for variable hydrologic conditions. |
