Key points are not available for this paper at this time.
Abstract The assessment of regional climate change requires the development of reference long‐term retrospective meteorological datasets. This article presents an 8‐km‐resolution atmospheric reanalysis over France performed with the the Safran‐gauge‐based analysis system for the period 1958–2008. Climatological features of the Safran 50‐year analysis—long‐term mean values, inter‐annual and seasonal variability—are first presented for all computed variables: rainfall, snowfall, mean air temperature, specific humidity, wind speed and solar and infrared radiation. The spatial patterns of precipitation, minimum and maximum temperature are compared with another spatialization method, and the temporal consistency of the reanalysis is assessed through various validation experiments with both dependent and independent data. These experiments demonstrate the overall robustness of the Safran reanalysis and the improvement of its quality with time, in connection with the sharp increase in the observation network density that occurred in the 1990s. They also show the differentiated sensitivity of variables to the number of available ground observations, with precipitation and air temperature being the more robust ones. The comparison of trends from the reanalysis with those from homogenized series finally shows that if spatial patterns are globally consistent with both approaches, care must be taken when using literal values from the reanalysis and corresponding statistical significance in climate change detection studies. The Safran 50‐year atmospheric reanalysis constitutes a long‐term forcing datasets for land surface schemes and thus enables the simulation of the past 50 years of water resources over France. Copyright © 2009 Royal Meteorological Society
Vidal et al. (Thu,) studied this question.