Mapping Hydrogeologic Properties Using Ground Penetrating Radar

Abstract

Ground penetrating radar (GPR) images provide information about the location, scale and geometry of subsurface heterogeneities. With well control or geologic intuition, this information can be used by hydrogeologists to fix the location of boundaries in fluid flow and contaminant transport models. While the location of boundaries is important, the hydrogeologist also requires estimates of the physical properties within the boundaries, for instance the hydraulic conductivity and porosity of different stratigraphic units. In essence, what the hydrogeologist really wants are the input parameters to his or her model. If the model is deterministic, these inputs are a discretized map of hydrogeologic properties. If the model is stochastic, these inputs are geostatistical parameters such as the correlation length, distribution, mean, variance and trend of hydrogeologic properties. With our goal being to help the hydrogeologist, we are investigating the capabilities of GPR to map hydrogeologic properties and characterize the heterogeneity of shallow aquifers. This paper presents results of a detailed 3-D GPR survey conducted at the U.S.G.S. Toxic Waste Hydrology Research Site on Cape Cod. The survey was completed in an area encompassing two well arrays where the spatial heterogeneity of aquifer properties is being investigated. Survey parameters were selected to cover the relevant scales of hydraulic conductivity variability thought to control contaminant migration in the sand and gravel aquifer. Trace density is one trace per square foot in an area measuring 370 feet by 100 feet, making this one of the most detailed 3-D GPR data sets every collected. The data were processed to remove the effects of topography and acquisition geometry. Common midpoint soundings conducted at 16 locations and water table depths measured in 20 wells were used to construct a model of subsurface electromagnetic propagation velocities. This velocity model was used for depth conversion and to estimate the average dielectric properties of the saturated and unsaturated zones. An empirical relationship between permittivity and porosity was then used to estimate the average porosity of the aquifer material. Small-scale features in the GPR data were examined using a sophisticated visualization program. With this program, 3-D images may be sliced, rotated and magnified, allowing us to construct profiles coincident with the well arrays. Individual amplitude ranges may be specified to construct isosurfaces and highlight events of interest; this is useful for reducing the opacity of 3-D images and for examining the topology of different events such as the water table and apparent stratigraphic reflections. Using these techniques, a map of the water table was produced that was accurate to within several tens of centimeters. In addition, correlations were observed between apparent radar stratigraphy and hydraulic conductivity variations determined from borehole flowmeter measurements and a large-scale tracer test conducted previously at the site. Seven continuous cores were acquired to confirm radar interpretations and provide material for laboratory measurement of dielectric and hydrogeologic properties. By comparing GPR images with other field and laboratory measurements of material properties, a better understanding is gained of the capabilities and limitations of this shallow remote sensing tool.