An evaluation of the use of regional climate model data applied to extreme precipitation in the Meuse basin

Understanding of hydrological and meteorological extremes is essential for flood risk management and flood protection. A primary focus in these professions is adequate estimation of extreme events that correspond to large return periods. Hydrological and meteorological observations only go back several decades, complicating frequency analysis of these large extremes. Capturing the tail behaviour of extremes is particularly challenging with such short records, resulting in high uncertainty of large precipitation and discharge extreme estimates.

This study proposes an alternative strategy for hydrological and meteorological frequency analysis. Long timeseries obtained from regional climate models are used to replace short observational datasets, leading to a substantial reduction of the statistical uncertainty of meteorological and hydrological extreme estimates. The approach was tested in the Meuse basin as part of the EMFloodresilience project, evaluating meteorological extremes from 16 synthetic ensembles of 65 years from the RACMO regional climate model (forced by the EC-EARTH global climate model). Hydrological extremes are analysed in a subsequent study from Rijkswaterstaat and Deltares, by forcing the wflow discharge model with the RACMO climate model dataset.

The study results reveal that bias-corrected model data is climatologically comparable to observational averages and extremes, exhibiting similar GEV location and scale parameters. Revealing a previously unexamined range of extremes, the model data offers a more plausible method to estimate the tails of annual extremes and likely provides a better estimate of the corresponding GEV shape parameter. Spatially, the model-derived parameter shows greater consistency across different sub-catchments of the Meuse basin compared to observations, suggesting a more robust insight in the tail behaviour of extremes. Additionally, a distinct separation between GEV distributions of summer and winter events is observed, indicating a transition in magnitude dominance from winter to summer maxima and possibly the presence of a double population. The existence of such a double population is difficult to obtain from observations, but can have an enormous impact on the return values of summer extremes. This emphasizes the need for further research on this area for adequate flood management.