Climatology of long-term extreme precipitation using rain gauges data over France

This study presents a comprehensive analysis of extreme precipitation events (EPs) across France over the period 1950-2023, using a rigorously quality-controlled dataset of 352 rain gauges that met strict completeness criteria (≥ 99% temporal coverage). We characterized spatial and temporal variability of EPs using multiple indices including annual maximum 1-day precipitation (RX1), 99th percentile threshold (P99th), mean exceedance above the 99th percentile (ME99th), and frequency of extreme rainy days (RP99th), following the definitions set by the Expert Team on Climate Change Detection and Indices (ETCCDI), with particular focus on the Massif Central and Cévennes regions as critical hydrological hotspots. Quality control showed that 95% of stations had no significant outliers, and 86% demonstrated homogenous precipitation data. A statistically significant increasing trend in RX1 was observed (0.5 mm decade -1 , P < 0.05), while P99th exhibited no significant trend. RP99th revealed a significant decreasing trend (-1.6 days/decade, P < 0.05), whereas ME99th showed a significant increase (~0.26 mm decade -1 , P < 0.1). Seasonally, RX1, P99th and ME99th peaked in autumn, while RP99th increased from May through December. Spatially, stations located in the Cévennes and Massif Central recorded the highest values of RX1, P99th, and ME99th, with a noticeable intensification in their trends. The years 1982 and 2003 showed peak anomalies in RX1 and ME99th. Comparison with ERA5 reanalysis data showed systematic overestimation of rainy days (by 10-30 days) and underestimation of peak precipitation in mountainous regions, highlighting the superiority of ground observations for trend detection. Investigation of large-scale atmospheric drivers revealed that six climate indices (El Niño–Southern Oscillation (ENSO), Arctic Oscillation (AO), North Atlantic Oscillation (NAO), East Atlantic Oscillation (EAO), Scandinavian pattern (SCA), East Atlantic–Western Russia (EAWR)) with semiannual and annual cycles of precipitation collectively explain up to 33% of precipitation variability, though their influence was weaker for extreme indices. Nonetheless, correlations between global climate indices and precipitation were statistically significant ( P < 0.05), exceeding 0.4 at several stations. Notably, the AO and SCA had distinct negative (r = -0.26) and positive (r = 0.29) effects, respectively. Thirteen widespread extreme precipitation events (P99th) were identified, each affecting at least 15% of stations. Among them, only the event in December 2003 was classified as severely abnormal based on Gumbel return period analysis; others were within the historical range of variability. Synoptic analysis indicated that these events were often driven by low-pressure systems over the North Atlantic, with cyclonic circulation and enhanced westerlies channeling moisture via atmospheric rivers into the region.