GEOSTATISTICAL ANALYSIS OF HISTORICAL GROUNDWATER CONCENTRATION DATA FOR CALIBRATION OF FLOW AND TRANSPORT MODELS

Large amounts of historical data are available on the concentration of selected contaminants in groundwater at the Hanford Site, and this data can be used to evaluate fate and transport model performance used for decision making at the Site, including a System Assessment Capability (SAC) model recently developed for the Hanford Site. The historical groundwater contaminant concentration data can also be used to constrain initial inventories for sources of particular plumes, which can be highly uncertain and are important inputs to the SAC and other risk assessment models of the Hanford Site.

Geostatistical methods applied to the historical concentration data were used to generate several hundred stochastic simulations of four radioactive contaminants: tritium, technetium-99, iodine-129, and uranium, for two time points, 1992 and 2001. The simulations included all major plumes for each of these radioactive contaminants at the Site and were generated using a 50-m grid covering 781 square kilometers. Post-processing of the simulated contaminant concentrations on the fine grid provided several quantitative metrics that will be used to evaluate the overall performance of the SAC model. One metric was the total area for which the contaminant concentration was above the drinking water standard (DWS) for each realization. Analysis of the suite of realizations provided a measure of uncertainty about the area above the DWS. The concentration simulations were also converted to estimates of the contaminant mass (or activity) in each grid cell. Mass estimates were based on probability distributions for the porosity of each geologic unit and a model of the thickness of the geologic units present in each cell of the grid. The post-processed simulations provided probability distributions of the total mass or activity for each contaminant, as well as estimates of the center of mass of each plume. The approach provides a best estimate of the metrics, as well as estimates of the uncertainty in the metrics. Output from the SAC model will be compared to the geostatistical results and used to help calibrate the model. General results of the analysis indicate that the contaminant mass within a plume is only known to within a factor of about four, even when the sampled concentration data are assumed to be without error.