PNNL

Abstract

Surface-based ground-penetrating radar data were collected at the Hanford Site in Washington, U.S.A. to assess the use of radar reflection images as a means of quantifying the spatial variability of subsurface water content. Available at the selected test site were two sets of water content data derived from neutron probe measurements that had been made to a depth of ~18 m in 32 wells. The comparison of probe-derived water content data, synthetic radar data, and the acquired radar data indicated a good correspondence between the changes in probe-derived water content and the location of reflections in the radar data. Geostatistical analysis was conducted on the two sets of probe-derived water content values and the amplitudes of the reflections in the radar reflection image to determine the horizontal correlation length of water content. The experimental semivariograms for the water content data were fit with a single exponential model with a correlation length of 10 m. The semivariogram for the radar data was fit with a nested structure containing a dominant long-range structure with a correlation length of 14 m, and a smaller-scale structure with a correlation length of 0.3 m. Quantifying the scale triplet – the spacing, extent, and support – for the two forms of measurement provided a framework for comparing and assessing the derived correlation structures.