Uncertainty analysis in river modelling

Abstract

Uncertainty analysis is an essential step in river modelling. Knowledge of the uncertainty is crucial for a meaningful interpretation of the model results. In this chapter we describe the whole process of an uncertainty analysis in four steps: identification, prioritization, quantification and propagation. In each step the rationale behind choosing a method is described and illustrated with an example of the design water level computation of the Dutch river Waal with a 2D hydrodynamic model. The sources of uncertainty related to the case study are identified and their (relative) importance is determined using expert opinions combined with a novel uncertainty identification method. Subsequently, the sources with the largest effect on the design water levels are individually quantified and propagated using Monte Carlo analysis to yield the quantified uncertainty in the design water levels. The uncertainty analysis provided information about the reliability of the model results and about further actions to possibly reduce the uncertainty and their benefits in terms of increased accuracy.