Reconstructing Groundwater Source Release Histories Using Hybrid Optimization Approaches

Abstract

Reconstructing release histories of contaminant sources in groundwater is an important environmental forensics problem for identifying responsible parties and allocating remediation costs in a contamination incident. Such problems typically require the solution of an inverse problem. This paper investigates and compares several hybrid optimization approaches that combine genetic algorithms with a number of local search approaches for solving these problems. Previous studies in this area have focused on simple two-dimensional and/or single-source release history problems. The problems solved in this article address both single- and multiple-source releases in three-dimensional heterogeneous flow fields. A parallel computing environment is used to handle the heavy computational needs of these problems. The results indicate that hybrid optimization methods, especially those that combine an initial global heuristic approach (e.g., genetic algorithms) with a subsequent gradient-based local search approach (e.g., conjugate gradients) are very effective in solving these problems.