Surface And Borehole Radar Monitoring Of A Dnapl Spill In 3D Versus Frequency, Look Angle And Time

As part of the University of Waterloo continuing experiments at Canadian Forces Base Borden (CFB), a controlled spill of perchloroethylene, a dense nonaqueous phase liquid (DNAPL), was monitored by a variety of geophysical methods. Surface ground penetrating radar (GPR) at 300,500, and 900 MHz and hole-to-hole GPR at 160 MHz were periodically measured at the site for 340 hours. This report presents the results of the first 66 hours of 500 MHz surface data and an example of the 900 MHz data at 14 hours. The surface data were acquired on a one meter grid in both directions across a nine meter by nine meter experimental cell. The borehole radar data were acquired between 8 wells circling the spill point on a 3 meter radius, using all 28 combinations of non-repeating holeto-hole pairs, each at 14 different depths in the cell at 25 cm intervals. The resulting multidimensional (x, y, z, time, frequency, look angle) dataset clearly outlines the movement of the DNAPL horizontally and vertically, and the interactions of the DNAPL with the heterogeneous sand matrix. Subtle changes in grain size distribution and the resultant capillary forces in the sand caused the DNAPL to spend more time and distance in horizontal travel than vertical travel. INTRODUCTION In July and August 1991, the U. S. Geological Survey (U.S.G.S.) participated in an experiment conducted by the University of Waterloo at Canadian Forces Base Borden, in which approximately 770 liters of perchloroethylene were injected over a period of 70 hours into an undisturbed, natural sand aquifer, and then monitored for several weeks. Perchloroethylene (PCE, also called tetrachloroethene, a common dry cleaning solvent) is the third most commonly occurring organic pollutant (Plumb and Pitchford, 1985). The PCE spill was contained in a 9 meter by 9 meter experimental cell constructed of steel sheet piling. The main objective was to determine if various geophysical methods could track movement of the DNAPL in the subsurface (Redman, 1992). Finding and monitoring the movement of DNAPLs are difficult because of their physical properties (Schwille, 1988). DNAPLs have low solubility in water and are denser than water, so they will pool at the base of an aquifer, slowly releasing a toxic dissolved phase into the groundwater (Plumb and Pitchford, 1985). As DNAPLs are generally less viscous than water, movement is generally unrelated to ground water flow and more tightly coupled to geological heterogeneity (Schwille, 1988, Huling and Weaver, 1991; USEPA, 1992). The difficulty in mapping these contaminants is two-fold, lying in the low level of geophysical contrast that these contaminants provide against background soil and rock, and in the low level of contaminant concentration considered to be of regulatory concern (typically parts per billion; USEPA, 1992). Methods used by the University of Waterloo to monitor the spill included surface and borehole resistivity, neutron logging (Schneider and Greenhouse, 1992), surface GPR at