Multi-model projections of water resources in Europe under two degree global warming

Summary

One of the main objectives of the EU-FP7-project IMPACT2C is to develop projections of water fluxes and stores
in Europe under two degree global warming. For this purpose, a multi-model assessment was carried out using
eleven CORDEX climate change simulations, which were carried out with five different GCM/RCM combinations
driven by three different RCPs (2.6, 4.5 and 8.5). After making bias corrections, the output from the eleven climate
simulations was used to force five pan-European hydrological models (E-HYPE, Lisflood, LPJmL, VIC and
WBM), resulting in an ensemble of 55 simulations. The ensemble of climate changes (the plus-two-degrees climate
relative to 1971-2000) was evaluated in terms of the median, the standard deviation (measure for uncertainty)
and significant changes. The latter are defined as those changes for which the absolute value of the median exceeds
the standard deviation. We also performed a flood analysis for two return periods (10 and 100 years) fitting a GEV
distribution on the data. Changes in water resources and largely driven by changes in precipitation. Precipitation is
projected to increase in most parts of Europe with decreases confined to Southern Europe. Generally, the patterns
of changes in evapotranspiration and runoff mimic the precipitation change pattern. As a result river discharge is
projected to increase in the majority of Europe in the plus-two-degrees climate. The largest increases occur in the
east and the far north while discharge decreases in parts of the Mediterranean. Due to a large spread in model
outcome only in half of Europe the projected changes in discharge are significant. Changes (mostly decreases) in
soil moisture are significant only in parts of the Mediterranean. It was found that uncertainty in runoff change was
to a larger extent due to the climate models than to the hydrological models whereas uncertainty in soil moisture
changes was mainly due to the hydrological models.