3D Multi-Offset, Multi-Polarization Acquisition And Processing Of Gpr Data: A Controlled Dnapl Spill Experiment

Dense non-aqueous phase liquid contaminants (DNAPL) typically have much lower electric<br>conductivity and electric permittivity than water. The bulk electric properties of the subsurface can be<br>significantly altered when these contaminants replace water in the pore space. Ground-penetrating radar<br>(GPR) is sensitive to permittivity contrasts and provides the potential to identify zones of low permittivity<br>associated with the presence of DNAPL. To test 3D multi-fold GPR techniques for quantifying DNAPL<br>induced permittivity anomalies, my research team conducted a small (107 cm x 122 cm), controlled DNAPL<br>spill experiment. The model was confined within a cylindrical polyethylene tank; model material consisted<br>of medium to coarse grained sand with a thin gravel layer near the base. My team injected twenty liters of<br>a chlorinated solvent solution into the vadose zone just below the surface, and monitored contaminant<br>migration into and through the water saturated zone to the bottom of the tank. I compiled a comprehensive<br>dataset for testing a variety of data processing and analysis techniques including 900 MHz, multi-offset, 3D<br>surface datasets in both TE and TM polarizations, 2D GPR transmission data, downhole TDR probe data,<br>and post-injection soil samples for chemical analysis. Both reflection tomography from TE polarized surface<br>data and crosswell tomography from transmission data reveal significant velocity anomalies associated with<br>pooled DNAPL that approaches a saturation of 40%. Further, thinbed offset-dependent reflectivity analysis<br>of TM surface data suggests the formation of a thin, highly saturated (80-100%) DNAPL zone at the top of<br>the main DNAPL pool. This work demonstrates that detailed analysis of multi-offset, multi-polarization GPR<br>data can significantly improve our ability to quantify subsurface permittivity anomalies.