From ENSEMBLES to CORDEX: Evolving climate change projections for Upper Danube River flow

Regional Climate Models (RCMs) generate past and future climate simulations that serve as input for subsequent modelling of impact projections. The most recent coordinated regional climate modelling initiative, CORDEX, provides RCM data for Europe with increased spatial resolution (12.5 km for the CORDEX EUR-11 ensemble) and based on the new Representative Concentration Pathways (RCPs). RCM data from the previous initiative ENSEMBLES had a spatial resolution of 25 km and was based on the SRES emission scenarios. In this contribution we explore the development from ENSEMBLES to CORDEX in a hydrological impact study for the Upper Danube, focussing on the representation of past climate and projections of the future for climate and resulting river discharge. We replicated a hydrological modelling framework, including RCM downscaling and bias correction, that used data of 21 ENSEMBLES RCMs under SRES A1B emission scenario (Kling et al., 2012), and now applied CORDEX EUR-11 data of 32 RCM simulations under RCP4.5 and RCP8.5 emission scenarios. Results with CORDEX show a small improvement in the representation of historical precipitation, but a decline in the accuracy of historical temperature simulations, despite the increase in spatial model resolution. The tendency of ENSEMBLES climate projections is reproduced with the CORDEX RCMs, albeit with slightly higher precipitation in the CORDEX data, yielding a less pronounced reduction in future mean annual discharge for the Upper Danube at Vienna. The previously identified change in discharge seasonality - increasing winter discharge and decreasing summer discharge - is confirmed with the new CORDEX data. Updating climate impact simulations in the presented rigorous replication framework allows analysing specific new information in the latest generation climate model data as well as crystallizing essential expectable patterns of change that need to be addressed in adaptation strategies.