Ground Penetrating Radar Optimization in Southern Arizona Soils

Abstract

Ground Penetrating Radar is an effective, non-destructive subsurface imaging system that has gained popularity since its inception. However, the efficacy of this method is significantly dampened when used in conductive, high-loss environments such as those commonly found in the clay rich alluvium of southern Arizona. In this thesis different survey parameters are tested to optimize the depth of investigation of GPR in these conditions, as well as determining what parameters should be used to calculate both GPR wave velocity and attenuation. To calculate velocity and attenuation the conductivity and dielectric permittivity must be known. Data from soil samples in southern Arizona indicate that, at common GPR frequencies, conductivity is a complex term, and this complex value must be used for calculations. Using the complex conductivity values in these calculations yield accurate velocities, and therefore are assumed to produce accurate attenuation values. Additionally, it is shown that areas with similar geologic features which have similar DC resistivity values will have similar complex conductivity values. In test sites located geologically close to mountain ranges, i.e. relatively shallow portions of valleys, it is shown that frequencies 200MHz and below are able to resolve a 3m deep aluminum sheet target across a variety of system if appropriate numbers of stacks are used. Higher frequencies may be able to resolve these targets depending on site conditions. Further away from mountains, i.e. deeper portions of valleys, GPR systems struggle to resolve targets even at low frequency and high numbers of stacks. As technology improves and systems are capable of even more numbers of stacks, this may change. Additionally, while collecting data in a 3D grid may slightly reduce the number of stacks needed to resolve a target, 3D data still significantly benefits from increased numbers of stacks. Comparing the Average Trace Amplitudes of datasets containing GPR lines of different stacks, approximate depths of investigation can be calculated for any number of stacks. This method is useable even if a target cannot be resolved with a system as the overall characteristics of each line are compared to one another. This is useful in high-loss environments as depths of investigation with different antennas can be estimated without knowledge of the electrical properties of the site without the requirement of a known target to estimate these properties.