Application of Data Fusion Modeling (DFM) to Site Characterization
D. W. Porter, B. P. Gibbs, W. F. Jones, J. R. Fairbanks, L.
L. Hamm, and G. P. Flach
Subsurface characterization is faced with substantial uncertainties because the earth is very heterogeneous, and typical data sets are fragmented and disparate. DFM removes many of the data limitations of current methods to quantify and reduce uncertainty for a variety of data Apes and models. DFM is a methodology to compute hydrogeological state estimates and their uncertainties from three sources of information: measured data, physical laws, and statistical models for spatial heterogeneities. The benefits of DFM are savings in time and cost through the following: the ability to update models in real time to help guide site assessment, improved quantification of uncertainty for risk assessment, and improved remedial design by quantifying the uncertainty in safety margins. A Bayesian inverse modeling approach is implemented with a Gauss Newton method where spatial heterogeneities are viewed as Markov random fields. Information from data, physical laws, and Markov models is combined in a Square Root Information Smoother (SRIS). Estimates and uncertainties can be computed for heterogeneous hydraulic conductivity fields in multiple geological layers from the usually sparse hydraulic conductivity data and the oft n more plentiful head data. An application of DFM to the Old Burial Ground at the DOE Savannah River Site will be presented. DFM estimates and quantifies uncertainty in hydrogeological parameters using variably saturated flow numerical modeling to constrain the estimation. Then uncertainties are propogated through the transport modeling to quantify the uncertainty in tritium breakthrough curves at compliance points.
AAPG Search and Discover Article #91019©1996 AAPG Convention and Exhibition 19-22 May 1996, San Diego, California