Intra-annual link of spring and autumn precipitation over France

• Main Authors: Hirschi, Martin, Seneviratne, Sonia
• Language: English
• Published: 2018
• Subjects: North Atlantic
• Online Access: http://doc.rero.ch/record/313609/files/382_2009_Article_734.pdf

In a previous study, an intra-annual relationship of observed precipitation, manifested by negative correlations between domain-averaged spring and autumn precipitation of the same year, was found in two domains covering France and Central Europe for the period 1972-1990 (Hirschi etal., J Geophys Res 112(D22109), 2007). Here, this link and its temporal evolution over France during the twentieth century is further investigated and related to the atmospheric circulation and North Atlantic/Mediterranean sea surface temperature (SST) patterns. Observational datasets of precipitation, mean sea level pressure (MSLP), atmospheric teleconnection patterns, and SST, as well as various global and regional climate model simulations are analyzed. The investigation of observed precipitation by means of a running correlation with a 30-year time window for the period 1901-present reveals a decreasing trend in the spring-to-autumn correlations, which become significantly negative in the second half of the twentieth century. These negative correlations are connected with similar spring-to-autumn correlations in observed MSLP, and with negatively correlated spring East Atlantic (EA) and autumn Scandinavian (SCA) teleconnection pattern indices. Maximum covariance analyses of SST with these atmospheric variables indicate that at least part of the identified spring-to-autumn link is mediated through SST, as spring precipitation and MSLP are connected with the same autumn SST pattern as are autumn precipitation, MSLP and the SCA pattern index. Except for ERA-40 driven regional climate models from the EU-FP6 project ENSEMBLES, the analyzed regional and global climate models, including Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (AR4) simulations, do not capture this observed variability in precipitation. This is associated with the failure of most models in simulating the observed correlations between spring and autumn MSLP. While the causes for the identified relationship cannot be fully established its ...