Evaluation of the uncertainty of groundwater model predictions associated with conceptual errors: A per-datum approach to model calibration

The effect of systematic model error on the model predictions varies in space and time, and differs for the flow and solute transport components of a groundwater model. The classical single-objective formulation of the inverse problem by its nature cannot capture these characteristics of model error. We introduce an inverse approach that allows the spatial and temporal variability of model error to be evaluated in the parameter space. A set of solutions for model parameters are obtained by this new method that almost exactly satisfies the model equation at each observation point (per-datum calibration). This set of parameter estimates are then used to define a posterior parameter space that may be translated into a probabilistic description of model output to represent the level of confidence in model performance. It is shown that this approach can provide useful information regarding the strengths and limitations of a model as well as the performance of classical calibration procedures.