Large-scale hydrological extremes in Europe: past and future simulations

Eight global hydrological models (GHM): Jules, Orchidee, HTessel, H08, MPI-HM, LPJml, WaterGap and GWAVA, were assessed concerning their ability to reproduce European large scale drought and high flow events when using meteorological inputs either observed (as represented by the WATCH Forcing Data WFD) or simulated by three Global Circulation Models (GCM), ECHAM5, IPSL and CNRM run under historical emission scenarios (control runs). In addition, the same GHMs were run with future projections from the same GCMs run under the A2 emissions scenario from 2001 to 2100 and the characteristics of large scale hydrological events compared to those of historical runs. The analysis of RDI and RHFI on six contrasting regions across Europe does not suggest that any particular GHM, or family of GHMs, better reproduce the spatial coherence of flow anomalies than any other. When driven by different modelled, rather than observed, climate for the 20th century, the sensitivity of different GHMs to the climate input becomes more apparent. The difference in the characteristics of future large-scale hydrological events as simulated by different GCM/GHM combinations is further increased, due to both climate projection uncertainty and GHM sensitivity. When future projections are analysed, results suggest that in temperate regions of Europe the number of large scale drought events is projected to increase by the end of the 21st century by most GCM/GHM combinations. For large scale high flow events, the signal is less clear but suggests little change in the number of events. However, in north-west Scandinavia fewer drought events and more high flow events are projected, probably due to increases in temperature causing less snowfall and more meltwater in the system. A change in the seasonality of events is also projected to occur in the High Alps with more high flows during spring, possibly again due to the influence of a warmer climate. This multi-model analysis has clearly highlighted that the uncertainty due to hydrological modelling in climate change impact studies, often assumed to be negligible compared to that of climate modelling, can be large (and sometimes larger than that of GCMs) and should not be ignored