Approche régionale pour l’estimation des distributions ponctuelles des pluies journalières dans le Languedoc-Roussillon (France) / Regional approach for the estimation of low-frequency distribution of daily rainfall in the Languedoc-Roussillon region, France

Abstract

Abstract This paper aims to show the benefit of a regional approach for the estimation of rare daily rainfall. The studied region is Languedoc-Roussillon (south of France), where recent exceptional storms necessitate the revision of the statistical distributions, particularly their asymptotic tails over extreme values. The example of a large single-site time series of maximum daily rainfall at Marseille (1864–2002), very close to the studied region, shows a hyper-exponential behaviour for extreme events. At the regional scale, the homogenization process of daily maximum rainfall has been performed by considering that the coefficients of variation of the yearly maximum daily rainfall are stationary over the study zone. Two regional sample studies have been carried out on 15 and 23 gauges, randomly distributed in space, and a similar distribution could be fitted to both samples. As in the case of Marseille, the regional distribution shows a hyper-exponential asymptotic behaviour at the extreme values. The obtained regional distribution provides a systematic method for computation of rare daily rainfall that may be applied in every part of the studied region and, when compared with previous estimations, leads to a significant increase in the depth of rare rainfall.

Keywords:

• hyper-exponentiel
• Languedoc-Roussillon
• période de retour
• précipitations extrêmes
• régionalisation
• station-années
• hyper-exponential
• frequency estimation
• extreme precipitation
• regionalization
• station-year

Additional information

Notes on contributors

Christophe Bouvier

[email protected]