Time-lapse multi-offset imaging of infiltration in heterogeneous media

Reflection profiling with ground-penetrating radar has proven to be a useful and efficient tool for mapping the shallow subsurface. Though less extensively used, multi-offset GPR data can provide additional information about EM wave velocity needed for depth registration of the images and estimation of water content. Such information is particularly useful for time-lapse imaging of hydrologic events, like infiltration. We present examples where time-lapse multi-offset GPR data are used to image changes in a sand tank during controlled flux infiltration experiments for the following conditions: 1) homogeneous sand, 2) sand with a buried land mine surrogate, and 3) sand with an embedded thin layer of low permeability silica flour. The tank measures 1.5m x 1.5m x 0.8m and is filled with 0.6m of sand underlain by 0.2m of gravel for drainage. A network of 15 soil moisture probes embedded in the sand measure water content continuously throughout the experiment at 10 second intervals. In each experiment we collected over 6000 traces of 900MHz multi-offset ground-penetrating radar data using an automated motion control system, yielding a multi-offset section across the tank every 20 seconds. The data are organized into a cube with axes of elapsed time since the start of infiltration, GPR travel time, and antenna offset. Conceptualizing the data as a cube allows us to evaluate the consistency of patterns in the multi-offset and common offset gathers. Clearly visible in all data are the airwave, groundwave, and bottom of tank reflection. Also present in the radar data is an arrival associated with the wetting front. Normal move-out analysis of arrivals for the homogeneous tank allow us to evaluate water content changes throughout the experiments, which were found to be in good agreement with results from HYDRUS-1D model simulations and the in-situ moisture probes. Arrivals associated with the land mine and thin layer are also observed in the additional experiments. Radar responses associated with wetting front propagation and transient water content values in the tank were not significantly affected by the presence of the land mine surrogate. Systematic changes in the amplitude of the arrivals directly associated with the land mine surrogate are, however, observed as a function of both antenna offset and water content throughout the experiment. In contrast, the presence of the thin layer affects the hydrologic dynamics of the tank and therefore greatly alters all radar responses compared to those observed for the homogeneous tank. These experiments show that the multi-offset radar response of the subsurface depends strongly on hydrologic conditions and that data collected under different conditions could be useful for discriminating between subsurface targets.