REDUCING UNCERTAINTY ASSOCIATED WITH GROUNDWATER-FLOW AND TRANSPORT PREDICTIONS

Effective evaluation of ground-water flow and transport problems requi res consideration of the range of possible interpretations of the subs urface given the available, disparate types of data, Geostatistical si mulation (using a modified version of ISIM3D) of hydrofacies units pro duces many realizations that honor the available geologic data and rep resent the range of subsurface interpretations of unit geometry. Hydra ulic observations are utilized to accept or reject the geometric confi gurations of hydrofacies units and to estimate ground-water now parame ters for the units (using MODFLOWP), These realizations are employed t o evaluate the uncertainty of the resulting value of the response func tion (ground-water now velocity and contaminant concentration) using M T3D. The process is illustrated with a synthetic data set for which th e ''truth'' is known, and produces a striking reduction in the distrib ution of predicted contaminant concentrations. The same system is eval uated three times: first with only hard data, then with both hard and soft data, and finally with only the realizations that honor the hydra ulic data (i.e., those accepted after parameter estimation via inverse flow modeling), Using only hard data, the mean concentration predicte d for all realizations at the point of interest is nearly two orders o f magnitude lower than the true value and the standard deviation of th e log of concentration is two. The addition of soft data brings the me an concentration within one order of magnitude of the true value and r educes the standard deviation of the log of concentration to one. Afte r eliminating realizations using inverse flow modeling, the mean conce ntration is one-third of the true value and the standard deviation of the log of concentration less than 0.5.