HR: 11:10h
AN: H21F-04    [PDF]
TI: Estimating Hydrogeological Zonation Using High-resolution Geophysical Data and Markov Chain Monte Carlo Methods
AU: * Chen, J
EM: jchen@lbl.gov
AF: Lawrence Berkeley National Lab, MS 90-1116, 1 Cyclotron Road, Berkeley, CA 94720 United States
AU: Hubbard, S
EM: sshubbard@lbl.gov
AF: Lawrence Berkeley National Lab, MS 90-1116, 1 Cyclotron Road, Berkeley, CA 94720 United States
AU: Fienen, M
EM: fienen@stanford.edu
AF: Stanford University, Department of Civil and Environmental Engineering, Stanford, CA 94305 United States
AU: Mehlhorn, T
EM: mehlhorntl@ornl.gov
AF: Oak Ridge National Lab, PO. Box 2008, MS6038, Oak Ridge, TN 37831 United States
AU: Watson, D
EM: watsondb@ornl.gov
AF: Oak Ridge National Lab, PO. Box 2008, MS6038, Oak Ridge, TN 37831 United States
AB: Although the importance of hydrogeological heterogeneity on contaminant transport is well recognized, the influence of the heterogeneity on remediation efficacy is not yet well established. In this study, we investigated the utility of high-resolution tomographic seismic data for estimating hydrogeological zonation using Markov chain Monte Carlo (MCMC) methods. The method was tested on data collected at the DOE NABIR Field Research Center (FRC) at Oak Ridge National Laboratory in Tennessee, where the subsurface consists of steeply dipping and fractured saprolite, and where ongoing studies are investigating the potential of biostimulation for uranium remediation. Our previously developed hydrogeophysical estimation approaches have been applied to several datasets collected within saturated porous environments. Those studies focused on estimating hydraulic conductivity using geophysical tomographic data, by first deriving relationships between co-located geophysical attributes and hydraulic conductivity measurements, and then using them in Bayesian models to estimate hydraulic conductivity. However, we found that developing relationships between seismic velocity and hydraulic conductivity with confidence at this fractured site proved difficult, possibly due to the difference in sampling volumes of the borehole flowmeter data and the geophysical data, which is exacerbated by the presence of fractures. For example, the wellbore flowmeter data collected at the site may sense the local fractures intersecting the wellbores, whereas the seismic data may sense fracture zones in a directionally dependent and effective manner over the distance between the two boreholes. Instead of estimating the absolute values of hydraulic conductivity as we did in other projects, we chose to estimate the hydrogeological zonation, defined as the probability of observing high permeability fracture zones, by integrating crosswell seismic and borehole flowmeter data using a Bayesian model. Within the Bayesian framework, both seismic velocity and zonation indicator at each pixel were considered as random variables, and crosswell seismic travel time and borehole flowmeter measurements were considered as data with measurement errors. Our goal was to estimate all the unknown quantities simultaneously by conditioning to the available data. We used MCMC methods to solve the Bayesian model by drawing many samples from the posterior distribution functions. Using those samples, we obtained the probability of observing high permeability fracture zones at each pixel along the tomographic cross sections. Our estimation results suggest that, over the study area, a localized high permeability fracture zone has laterally varying thickness and geological dip.
DE: 0935 Seismic methods (3025)
DE: 1719 Hydrology
DE: 1829 Groundwater hydrology
DE: 1869 Stochastic processes
SC: Hydrology [H]
MN: 2003 Fall Meeting