HR: 13:40h
AN: H43J-01    [Abstracts]
TI: On application of ground-penetrating radar tomography in shallow subsurface hydrological parameter estimation
AU: * Hou, Z
EM: hou@berkeley.edu
AF: U.C.Berkeley, CEE, U.C. Berkeley, Berkeley, CA 94720 United States
AU: Chen, J
EM: JChen@lbl.gov
AF: LBL, 1 Cyclotron Road, Berkeley, CA 94720 United States
AU: Rubin, Y
EM: rubin@ce.berkeley.edu
AF: U.C.Berkeley, CEE, U.C. Berkeley, Berkeley, CA 94720 United States
AB: The tomographic ground-penetrating radar (GPR) methods are believed to have the potential to improve our estimation of hydrological parameters such as hydraulic conductivity and water retention parameters in the shallow subsurface. However, before the application of a tomographic GPR approach, it is necessary to evaluate the sensitivity of the GPR responses to changes of the flow field properties, and the performance of the GPR forward model under various conditions in the shallow subsurface. Unfortunately, such analyses are not well documented to our knowledge. In this study, we propose a stochastic approach to explore the sensitivity of tomographic GPR responses to many factors such as the statistical moments and the spatial integral scale of hydraulic conductivity, the water retention parameters, the flow field boundary conditions, the field scales and aspect ratios, the ratio of infiltration rate to saturated hydraulic conductivity, the infiltration time, and the GPR transmitter/receiver locations. A flow simulator based on the Bresler-Dagan (BD) model is developed to simulate water flow in the upper soil layer of spatially variable fields. Then we obtain GPR traveltimes from the flow fields by employing two different tomographic GPR forward models: 1) the straight-ray model, which assumes straight ray paths between sources and receivers; and 2) the curved-ray model, which solves the Eikonal equation in the celerity domain and considers all the transmitted, diffracted and head waves in a local traveltime computation scheme. Random hydraulic conductivity fields are generated given various parameters including the moments and the spatial integral scale of hydraulic conductivity. Then by varying the other aforementioned factors, we simulate numerous transient flow fields during infiltration and redistribution. Next, the two GPR forward models are applied to selected snapshots of the flow fields to obtain first-arrival traveltimes, corresponding to different GPR transmitter/receiver settings. Statistics of the traveltimes from a single model and the model differences are summarized for model evaluation and comparison under various conditions. Conclusions are made regarding to how the aforementioned factors affect the tomographic GPR responses, under what conditions the two forward models give significantly different results, and which model is more appropriate to use under such conditions. We also give suggestions on how to determine the field scales and how to choose the dynamic snapshots of a flow field during an infiltration experiment, such that the GPR data is beneficial to hydrogeological inversion in shallow subsurface.
DE: 1829 Groundwater hydrology
DE: 1835 Hydrogeophysics
DE: 1869 Stochastic hydrology
DE: 1875 Vadose zone
DE: 1894 Instruments and techniques: modeling
SC: Hydrology [H]
MN: Fall Meeting 2005