Uncertainty Reduction in River Water Quality Modelling, Application on the River Dender in Belgium

Recently the European Union has approved the EU Water Framework Directive (EU-WFD). This directive claims that by the end of 2015 a “good status of surface water“ and a “good status of groundwater” should be achieved (EU, 2000). To that end, the EU-WFD provides several guidelines for monitoring the water bodies, leaving the practical implementation to the local governments. Each region has to develop a consistent monitoring strategy and assessment methodology. To make sure that the new water policy will succeed, a profound analysis of the actual and future state of the water is necessary. In this context, the evaluation of emissions into river water will be important. The effects of a pollution load into the river can be evaluated using models. After the calibration of a model with a certain data set, the model generates predictions over a period of months or even years. However, those predictions are uncertain. Model outcome uncertainties can become very large due to: • measurement errors, • uncertainty in the estimated model parameter values, • model structure uncertainty and • the stochastic nature of the biological processes in river water. This paper presents a methodology to reduce the uncertainty on model parameters and as a result reduce the uncertainty in the final model outputs. Three main steps are described. The methodology is applied to the Dender basin in Belgium. The river water quality model used is implemented in the ESWAT simulator.