Quantification of model uncertainty for WAQUA for the Upper River Area

In the context of the Wettelijke Toets Instrumentarium 2017 (WTI2017) project, the subproject Hydrodynamic modelling aims at identifying and quantify in the model uncertainty of WAQUA regarding the calculated water levels for the primary water systems in the Netherlands. For the Dutch upper rivers, this study aims at delivering the model uncertainty that needs to be added to the calculated water levels. The model uncertainty is specified as a distribution of the uncertainty on the water level, in terms of a bias and standard deviation. The model uncertainty for the other primary water systems in the Netherlands will be quantified in 2014. In the first phase of this project, a literature study was performed and an expert judgment session was organized. The outcome of the first phase revealed that the sources of uncertainty cannot be regared independently during the uncertainty analysis. Therefore, the model uncertainty should be estimated as a whole. According to the experts the most important source of uncertainty is the use of the WAQUA outside the calibration and validation data range. The performance of the WAQUA models under extreme conditions has never been quantified. Since the uncertainty due to the use of the model under extreme conditions is expected to be much larger than the model uncertainty within the calibration range, this study focuses on quantifying the model uncertainty under extreme conditions. However, a quantification of the model uncertainty within the calibration range is also given. The model uncertainty within the calibration range is derived from the calibration results. The uncertainty due to the calibration approach, morphological changes and differences between calculated and actual discharge distribution play an important role in the overall model uncertainty under extreme conditions. The uncertainty due to the calibration approach has been addressed by carrying out an investigation to identify the sensitivity of model results to using a different calibration approach, while the uncertainty due to morphological changes and due to differences between calculated and actual discharge distribution have been addressed in earlier studies 2009 and 2012. The overall model uncertainty under extreme conditions is quantified based on expert judgment, based on the results of the sensitivity analysis and the former studies and is the best estimate that the expert can give at this moment. This leads to an overall estimate of the model uncertainty for the upper river area of the Netherlands of a bias of 0 meter and a standard deviation of 0.8 meter under extreme conditions.