HR: 1340h
AN: H13D-1356 [Abstracts]
TI: Identifying Spatially Variable Sensitivity of Model Predictions and Calibrations
AU: * McKenna, S A
EM: samcken@sandia.gov
AF: Geohydrology Department,
Sandia National Laboratories, PO Box 5800 MS 0735, Albuquerque, NM 87185-0735
United States
AU: Hart, D B
EM: dbhart@sandia.gov
AF: Geohydrology Department,
Sandia National Laboratories, PO Box 5800 MS 0735, Albuquerque, NM 87185-0735
United States
AB:
Stochastic inverse modeling provides an ensemble of stochastic property fields, each calibrated to measured steady-state and
transient head data. These calibrated fields are used as input for predictions of other processes (e.g., contaminant
transport, advective travel time). Use of the entire ensemble of fields transfers spatial uncertainty in hydraulic
properties to uncertainty in the predicted performance measures. A sampling-based sensitivity coefficient is proposed to
determine the sensitivity of the performance measures to the uncertain values of hydraulic properties at every cell in the
model domain. The basis of this sensitivity coefficient is the Spearman rank correlation coefficient. Sampling-based
sensitivity coefficients are demonstrated using a recent set of transmissivity (T) fields created through a stochastic
inverse calibration process for the Culebra dolomite in the vicinity of the WIPP site in southeastern New Mexico. The
stochastic inverse models were created using a unique approach to condition a geologically-based conceptual model of T to
measured T values via a multiGaussian residual field. This field is calibrated to both steady-state and transient head data
collected over an 11 year period. Maps of these sensitivity coefficients provide a means of identifying the locations in the
study area to which both the value of the model calibration objective function and the predicted travel times to a
regulatory boundary are most sensitive to the T and head values. These locations can be targeted for deployment of
additional long-term monitoring resources. Comparison of areas where the calibration objective function and the travel time
have high sensitivity shows that these are not necessarily coincident with regions of high uncertainty. The sampling-based
sensitivity coefficients are compared to analytically derived sensitivity coefficients at the 99 pilot point locations.
Results of the sensitivity mapping exercise are being used in combination with other techniques to identify well locations
for a long-term groundwater monitoring network
This research is funded by WIPP programs administered by the Office of Environmental Management (EM) of the U.S. Department
of Energy. Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United
States Department of Energy's National Nuclear Security Administration under contract DE-AC04-94AL85000.
DE: 1829 Groundwater hydrology
DE: 1846 Model calibration (3333)
DE: 1848 Monitoring networks
DE: 1869 Stochastic hydrology
SC: Hydrology [H]
MN: Fall Meeting 2005