Optimal identification of ground-water pollution sources and parameter estimation

An optimization-based methodology for identifying unknown sources of ground-water pollution is presented. The proposed methodology utilizes an optimization model in which the flow and transport equations are embedded as constraints. A nonlinear programming algorithm is used to obtain as solution the optimal estimates of unknown source characteristics. The input to this model includes measured pollutant concentration at observation sites. The source identification methodology is further extended to the simultaneous estimation of aquifer parameters as well as identification of unknown pollutant sources. Performance of the developed methodology is evaluated for illustrative examples considering two-dimensional flow and advective-dispersive solute transport. Different cases including variability in data availability, single and multiple potential source locations, and errors in measurement data are considered. These performance evaluations demonstrate that at least for the illustrative example problems, the proposed methodology performs satisfactorily in identifying the locations, determining the magnitudes, and specifying duration of the unknown ground-water pollution sources, even when the aquifer parameters are unknown.