Use of Surrogate Models for a Groundwater Pollutant Source Characterization Problem

This paper investigates a groundwater source identification problem in which concentrations at observation wells are used to reconstruct the pollution loading scenario. This inverse problem is solved using a simulation-optimization approach using evolutionary algorithms to conduct the search. Varying levels of complexity may be modeled, leading to different levels of accuracy in predictions of source location, size, and concentration. More complex models will increase the computational effort needed to model pollutant transport as part of a search procedure. In addition, the amount of non-uniqueness will typically increase as the complexity of the problem increases. This paper describes an investigation of surrogate modeling procedures and methods to generate very different solutions to characterize groundwater pollutant source under varying degrees of problem complexity.