Complexities in hindcasting models--when should we say enough is enough?

Groundwater models are routinely used in hindcasting applications to predict the past concentration levels in contaminated aquifers. These predictions are used in risk assessment and epidemiological studies, which are often completed either for resolving a court case or for developing a public-policy solution. Hindcast groundwater modeling studies utilize a variety of computer tools with complexity levels ranging from simple analytical models to detailed three-dimensional, multiphase, multispecies, reactive transport models. The aim of this study is to explore the value of using complex reactive transport models in hindcasting studies that have limited historic data. I review a chlorinated solvent exposure problem that occurred at a U.S. Marine Corp Base in Camp Lejeune, North Carolina and use it as an example to discuss the limits of hindcasting modeling exercises. The lessons learned from the study are used to reflect upon the following questions related to model complexity: How should we decide how much is enough? Who should decide when enough is enough?