Three-Dimensional Groundwater Contamination Source Identification Using Adaptive Simulated Annealing

Abstract

Determination of groundwater contaminant source characteristics such as release histories of unknown groundwater pollutant sources from concentration observation data is an inverse problem. Often solution to this inverse problem is nonunique, and it is an ill-posed problem. A linked simulation-optimization approach can be used to solve this problem efficiently. However, this approach is computationally intensive, and the results obtained tend to be highly susceptible to errors in the measured data and estimated hydrogeological parameters. Apart from this, accuracy of the solutions is highly dependent on the choice of monitoring locations. An adaptive simulated annealing (ASA)-based solution algorithm is shown to be computationally efficient for optimal identification of the source characteristics in terms of execution time and accuracy. This computational efficiency appears to prevail even with moderate levels of errors in estimated parameters and concentration measurement errors. Also, the contaminant concentration monitoring locations are shown to be critical in the efficient characterization of the unknown contaminant sources. Optimal identification results for different monitoring networks are presented to demonstrate the relevance of a network suitable for efficient source identification.