Frequency of precipitation and temperature extremes over France in an anthropogenic scenario: Model results and statistical correction according to observed values

Météo-France atmospheric model ARPEGE/Climate has been used to simulate present climate (1961–1990) and a possible future climate (2071–2100) through two ensembles of three 30-year numerical experiments. In the scenario experiment, the greenhouse gas and aerosol concentrations are prescribed by the so-called SRES-A2 hypotheses, whereas the sea surface temperature and sea ice extent come from an earlier ocean–atmosphere coupled simulation. The model covers the whole globe, with a variable resolution reaching 50 to 60 km over France. Model responses on daily minimum and maximum temperature and precipitation are analyzed over France. The distribution of daily values is compared with observed data from the French climatological network. The extreme cold temperatures and summer heavy precipitations are underestimated by the model. A correction technique is proposed in order to adjust the simulated values according to the observed ones. This process is applied to both reference and scenario simulation. Synthetic indices of extreme events are calculated with corrected simulations. The number of heavy rain (> 10 mm) days increases by one quarter in winter. The maximum length of summer dry episodes increases by one half in summer. The number of heat wave days is multiplied by 10. The response in precipitation is less when only the change in the mean is considered. Such a corrected simulation is useful to feed impact models which are sensitive to threshold values, but the correction does not reduce, and may enhance in some cases, the uncertainty about the climate projections. Using several models and scenarios is the appropriate technique to deal with uncertainty.