Identification of Groundwater Pollution Sources Using GA-based Linked Simulation Optimization Model

Abstract

The genetic algorithm (GA)–based simulation optimization approach is used for optimal identification of unknown groundwater pollution sources. Simple as well as complex scenarios of multiple unknown groundwater pollution sources are considered. A flow and transport simulation model is externally linked to the GA-based optimization model to simulate the physical processes involved. The simulation model uses potential pollution source characteristics that are evolved by the GA and simulates the resulting concentration measurement values at observation locations. These simulated spatial and temporal pollutant concentration measurement values are used to evaluate the fitness function value of the GA. The main advantage of the proposed methodology is the external linking of the numerical simulation model with the optimization model. This approach makes it feasible to solve the source-identification problems for complex aquifer study areas with multiple unknown pollution sources. The performance of the developed methodology is evaluated for combinations of source characteristics (locations, magnitudes, and release periods), data availability conditions, and concentration measurement error levels.