HR: 1340h
AN: H23A-1016    [Abstracts]
TI: Comparison of Hydrologic Parameter Estimates Using Sequential and Integrated Data Fusion During a GPR Monitored Infiltration Event
AU: * Sicilia, G T
EM: tom@coffeepot.org
AF: Environmental Engineering & Earth Sciences, 340 Brackett Hall, Clemson, SC 29634- 0919, United States
AU: Moysey, S M
EM: smoysey@clemson.edu
AF: Environmental Engineering & Earth Sciences, 340 Brackett Hall, Clemson, SC 29634- 0919, United States
AB: Constraining parameters that govern variably saturated flow is important for applications ranging from quantifying water availability for ecosystems to constraining recharge rates and contaminant fluxes to groundwater. In this study we explore the effectiveness of sequential versus integrated data fusion for estimating unsaturated flow parameters using ground penetrating radar (GPR) data. In Sequential Data Fusion (SDF), geophysical imaging is used to create a map of the geophysical properties of the subsurface. Subsequently these properties are transformed to hydrologic properties that can be used to constrain an independent hydrologic inverse problem. In contrast, Integrated Data Fusion (IDF) uses the geophysical data to directly constrain hydrologic properties of interest without performing the intermediate geophysical imaging step. Our comparison of SDF and IDF is performed for a synthetic study of 2D infiltration into a homogeneous soil from a constant flux point source located at the ground surface. Here we focus on results for the estimation of intrinsic permeability (k) from cross- borehole GPR travel times collected throughout the duration of the infiltration event. The target permeability (k=7.4x10^-12m²) is uniform over the 20 meter by 20 meter area modeled in this study; though the soil is homogeneous, we emphasize that water content is both spatially variable and transient. We use TOUGH2 to simulate infiltration, MATLAB to simulate GPR travel times, and PEST to perform the parameter estimation. To quantitatively compare SDF and IDF, we calculate the normalized error in estimated permeability for each method. In our study, we investigated the performance of the data fusion methods under varying survey geometries by changing the antenna spacing. In all cases we have found that IDF significantly outperforms SDF. For large antenna separations (1.7-6.7m) SDF produces an average error in estimated permeability of 78.6% while IDF errors are only 35.4%. As ray density is increased for antenna separations of 1.0-1.5m, average estimation error for SDF drops to 77.3%, but is drastically reduced to only 5.8% for IDF. Also, SDF estimates are consistently biased lower than the target value, while IDF results are unbiased. Our results suggest the IDF is a powerful new approach for hydrologic characterization of the subsurface using geophysical measurements.
DE: 1835 Hydrogeophysics
DE: 1838 Infiltration
DE: 1875 Vadose zone
DE: 3260 Inverse theory
DE: 5114 Permeability and porosity
SC: Hydrology [H]
MN: 2007 Fall Meeting