Assessment of flood peak simulations by Global Hydrological Models

With significant changes to flood frequency anticipated as a result of climate change it becomes important to investigate how global hydrological models process climate forcing data. Flood frequency distribution describes the relationship between flood peak magnitude and its return period, indicating the average period of time between exceedance of a certain flood magnitude. The steepness of the distribution (or of the growth curve) is a measure of the variability of the flood peak series. Analysis of variation in extreme rainfall-runoff processes between global hydrological models was undertaken by comparing the variability in extreme rainfall events of certain frequency in WATCH Forcing Data (WFD) with the resulting variability in flood events as predicted across the models GWAVA, JULES, and WaterGap. Analysis of propagation of climate model biases between global hydrological models JULES and WaterGap compared changes in predicted flood frequency growth curve steepness when applying baseline daily climate forcing simulated by the ECHAM5 climate model time series. The impact of climate change on the probabilistic behaviour of floods is assessed by considering the change in growth curve between the control run conditions and the future scenarios as predicted by ECHAM5-A2 driven time series. Spatial patterns of variation in extreme rainfall-runoff processes differ between models, particularly in regions of extreme climate, highlighting the importance of using more than one hydrological model. Flood statistics derived from simulations from two hydrological models run with the same forcing climate differ significantly, suggesting that the models are sensitive to different climate characteristics, or that they are calibrated with sub-optimal conditions. Simulated control climate might also have some characteristics different than those of historical observations. Climate change simulations indicate some general agreement between models in the emerging spatial pattern of future changes to flood variability across Europe; however, some distinct regional and sub-regional differences in magnitude of change and spatial pattern are evident. The results obtained in this research are promising, and should be extended to include a larger sample of hydrological and climate models, with more detailed investigation of hydrological model structure and validation of modelled values with observed flow series.