Abstract

The Cévennes-Vivarais region in southern France frequently suffers from Heavy Precipitation Events (HPEs), especially during the autumn season. The northwestern Mediterranean Sea is a source of heat and moisture for these HPEs, with strong air sea exchanges, which are mainly controlled by the near-surface wind intensity and the Sea Surface Temperature (SST). The aim of this study is to characterize the SST structures, location and variability related to HPEs. Indeed, the Gulf of Lion is characterized by a cyclonic circulation with three main oceanic features: (1) the Northern Current (NC), (2) the Balearic Front (BF), (3) and the deep Convective Cell (CC). The MEDidterranean ReanalYsiS (MEDRYS1V2), an ocean reanalysis at 1/12˚ resolution was used over the 2000-2011 period to identify the NC, BF and CC oceanic feature's locations and for the calculation of an SST index. Then, an unsupervised cluster analysis method, using Principle Component Analysis for dimension reduction and the Silhouette for clustering, was used 20 in order to determine the most typical periods. The Local Moran's I (LMI) spatial statistical method, was used to highlight the areas of significant temporal variability of SST, considering the periods defined with the clustering method. For each period [early autumn (August-September), October and late autumn (November-December)], the LMI values, only considering the HPE initial stage, are extracted and compared to the averaged LMI values. The results highlight that in August-September, the Rhone river outflow have the most significant effect on SST variability on average, same as during HPEs. In October, for the HPEs initial stage, the largest variability seems to be related to the effects of the Mistral and Tramontane wind. In later autumn, there is a southward displacement of the BF and CC patterns and during the HPEs initial stage, the most significant SST variability is found near the center of the CC.

Keywords: Heavy precipitation events; Sea surface temperature; North-Western Mediterranean Sea circulation; Statistical clustering; Local Moran's I; Silhouette