Historical Reconstruction of Groundwater Contamination at Contaminated Sites and Uncertainty Analysis

The purpose of this study is to demonstrate the use of a linear control theory model. The model is used to reconstruct the contamination history in an aquifer characterized by limited property information as well as limited contaminant concentrations data at several observation points (Guan, et al., 2009). The data that may be available to reconstruct the historical groundwater contamination at the contaminated sites, which are utilized in this approach, might contain various uncertainties such as measurement errors or data gaps. In addition to these data errors, the model used to reconstruct the historical contamination may also introduce numerical errors to the overall analysis. Evaluation of the effects of these uncertainties and numerical errors is a challenging topic. In this study, numerical and field data errors are embedded into the linear control theory model resulting in a linear stochastic control system. The Kalman filtering method coupled with Monte Carlo simulation is then used to analyze the effect of these errors on concentration distributions at observation sites that are reconstructed. Using this procedure, the temporal confidence "corridor" of concentration distributions at observation sites can be constructed for a selected confidence interval. For demonstration purposes, the proposed methodology is applied to recover contamination history at the Tarawa Terrace area of U.S. Marine Corps Base Camp Lejeune, North Carolina. Numerical results obtained show that the methodology provides an effective and robust tool to reconstruct groundwater contamination history with uncertainties. The confidence corridors obtained for a given confidence interval can provide the necessary information on the uncertainty of the predicted results.