Incorporating Assurance into Groundwater Quality Management Models

Abstract

Recent efforts in aquifer management modelling have tried to account for uncertainty in predicted groundwater flow velocities and concentrations. Although there are many origins of model prediction uncertainty, model misspecification represents a significant one. Two types of model misspecification have been addressed. First is inadequate description of spatial variability of hydraulic conductivities. The latest approach has been to develop robust aquifer management plans that remain optimal in the presence of unknown spatial variability. The second type of model misspecification that has been looked at is model parameter uncertainty. For this problem, parameter estimation has been united with nonlinear stochastic optimization. Plans to control water quality degradation are developed that account for both flow and transport parameter uncertainty. The management model assures specified local water quality standards will be met at a specified reliability level. A brief survey of older literature shows that most studies were limited because they treated groundwater quality management models as linear systems. In the past five years problems have been formulated and solved using nonlinear systems theory. Furthermore, studies have begun to incorporate model uncertainty. These efforts, which attempt to account for both nonlinearity and model uncertainty, are critically analyzed in this review.