2018, Huang, Shaochun, Kumar, Rohini, Rakovec, Oldrich, Aich, Valentin, Wang, Xiaoyan, Samaniego, Luis, Liersch, Stefan, Krysanova, Valentina

This study assesses the flood characteristics (timing, magnitude and frequency) in the pre-industrial and historical periods, and analyzes climate change impacts on floods at the warming levels of 1.5, 2.0 and 3.0 K above the pre-industrial level in four large river basins as required by the Paris agreement. Three well-established hydrological models (HMs) were forced with bias-corrected outputs from four global climate models (GCMs) for the pre-industrial, historical and future periods until 2100. The long pre-industrial and historical periods were subdivided into multiple 31-year subperiods to investigate the natural variability. The mean flood characteristics in the pre-industrial period were derived from the large ensemble based on all GCMs, HMs and 31-year subperiods, and compared to the ensemble means in the historical and future periods. In general, the variance of simulated flood characteristics is quite large in the pre-industrial and historical periods. Mostly GCMs and HMs contribute to the variance, especially for flood timing and magnitude, while the selection of 31-year subperiods is an important source of variance for flood frequency. The comparison between the ensemble means shows that there are already some changes in flood characteristics between the pre-industrial and historical periods. There is a clear shift towards earlier flooding for the Rhine (1.5 K scenario) and Upper Mississippi (3.0 K scenario). The flood magnitudes show a substantial increase in the Rhine and Upper Yellow only under the 3.0 K scenario. The floods are projected to occur more frequently in the Rhine under the 1.5 and 2.0 K scenarios, and less frequently in the Upper Mississippi under all scenarios.