Changement climatique - Recherche française (avec laboratoires CNRS) - 2010-2015

Mountain regions are highly vulnerable to climate change and are likely to be among the areas most impacted by global warming. But climate projections for the end of the 21st century are developed with general circulation models of climate, which do not present a sufficient horizontal resolution to accurately evaluate the impacts of warming on these regions. Several techniques are then used to perform a spatial downscaling (on the order of 10 km). There are two categories of downscaling methods : dynamical methods that require significant computational resources for the achievement of regional climate simulations at high resolution, and statistical methods that require few resources but an observation dataset over a long period and of good quality. In this study, climate simulations of the global atmospheric model ARPEGE projections over France are downscaled according to a dynamical method, performed with the ALADIN-Climate regional model, and a statistical method performed with the software DSClim developed at CERFACS. The two downscaling methods are presented and the results on the climate of the French mountains are evaluated for the current climate. Both methods give similar results for average snowfall. However extreme events of total precipitation (droughts, intense precipitation events) are largely underestimated by the statistical method. Then, the results of both methods are compared for two future climate projections, according to the greenhouse gas emissions scenario A1B of IPCC. The two methods agree on fewer frost days, a significant decrease in the amounts of solid precipitation and an average increase in the percentage of dry days of more than 10%. The results obtained on Corsica are more heterogeneous but they are questionable because the reduced spatial domain is probably not very relevant regarding statistical sampling.