Prediction Uncertainty in a Multi-Scale Groundwater Flow and Transport Model

Coupling of groundwater flow and transport models represents a common challenge in groundwater modelling. The groundwater flow models are constructed at large scales to include the natural boundary conditions. The groundwater transport models are constructed at a smaller scale with finer discretization than the flow models in order to limit numerical dispersion and to accurately delineate the solute source while still maintaining reasonable computing time. The Universal inverse modelling software UCODE 2005 was used for combining groundwater models at different. When running these models in series, a comprehensive sensitivity analysis was performed, whereby the effect of a parameter perturbation in a larger scale model on (transport) predictions using a smaller scale model is quantified. A steady-state groundwater flow model was developed in Modflow and combined with a small scale transport model. Sensitivity analyses were performed to assess prediction uncertainty for hypothetical contaminant concentrations. Including all pertinent parameters into the sensitivity analysis enabled to estimate the relative contribution of different parameters to the prediction uncertainty about contaminant concentrations. Such relative importance was used to formulate recommendations for reducing the prediction uncertainty by means of further data acquisition.