Navigating Groundwater Model Uncertainty Analysis

Abstract:

The design of a groundwater model is based on the model objective, the management or research question that the model seeks to address. Designing the uncertainty analysis of a groundwater models likewise needs to consider the objective of the uncertainty analysis; how the uncertainty in model predictions will be used.

In this paper a framework is presented to consider the various dimensions of groundwater modelling and uncertainty analysis. The discussion is structured around four main types of groundwater model objectives: characterisation and conceptualisation, design and optimisation, scenario analysis and impact analysis. The differences in model objective affect (1) the choice of a quantity of interest, (2) the balance between honouring observations, being consistent with system knowledge and estimating the range of model outcomes, (3) which sources of uncertainty to emphasize and (4) the trade-off between model runtime, number of parameters and number of model evaluations. These differences will ultimately affect which uncertainty analysis technique is most suited for a practical groundwater model investigation. The concepts are illustrated with a case study, designing the uncertainty analysis for an update to a groundwater model for the Angas-Bremer aquifer system in Australia.

The framework provides a way to systematically discuss, document and justify the various choices inherent to a groundwater model and uncertainty analysis and helps navigating the various uncertainty analysis approaches currently available.