Using Ground Penetrating Radar Signal Reflection to Model a Highly Transmissive Bedrock Aquifer

The use of numerical modeling software to characterize groundwater contaminant fate and transport requires an understanding of the distribution of hydraulic properties of the aquifer. As the complexity of this distribution increases, a more sophisticated understanding is required. Modeling contaminant transport in bedrock aquifers requires that the spatial distributions of highly conductive geologic features are characterized along with the hydraulic transmissivities of those features (Shapiro, 2003). The goal of this research is to refine the numerical model of a highly transmissive crystalline aquifer by characterizing the spatial variability of the hydraulic transmissivity using the attenuation of the ground penetrating radar (GPR) signal. Slug tests were performed in ten bedrock wells, located within and surrounding a highly conductive zone. The transmissivities measured in these wells ranged from 1.5 E-1 ft2/day to 4.2 E+2 ft2/day; values greater than 100 ft2/day were measured in four of the bedrock wells. A GPR survey was conducted over the well network and five transects were selected for further analysis. The location of the ten wells were projected onto the nearest GPR transects and the GPR data from each location were isolated for further analysis. A visual inspection GPR signal data indicates that the signal reflection increases at locations where high transmissivity values were measured. This relationship was attributed to an increased volume of fluid present within the fractures; a characteristic previously used to identify productive bedrock aquifers at multiple sites (e.g., Porsani, 2005, Halihan, 2008). The correlation between the transmissivities and the analyzed signal was evaluated at the four locations where higher transmissivities were measured. No correlation was observed when all four points were included. When the data collected from well BW-18 were excluded from the analysis, a near perfect correlation was observed for the remaining three points. The transmissivity measured in BW-18 was the highest value measured onsite. This suggests that the aperture thickness, a characteristic which is not identified using GPR, may play a significant role in the distribution of transmissivities at this location. The remaining GPR data were transformed to the corresponding hydraulic conductivity using a linear regression fitted to the three correlating points. The values calculated were used as the conductivity values for the model layer corresponding to the shallow bedrock. The resulting model was no more accurate than an identical model created using only the measured transmissivities. Consistent with previous findings, this investigation determined that GPR data collected by surficial geophysical methods is capable of characterizing highly transmissive regions. Since the spatial variations in the aperture thickness are not identified using this method, it is not appropriate for sites where significant variation is expected. The high correlation suggests that it may be possible to relate signal attenuation to transmissivity over relatively small scales; however an attempt to demonstrate the method’s ability to improve modeling at this site has not been successful