Estimation des pluies journalières extrêmes supérieures à un seuil en climat tropical : cas de la Côte d'Ivoire

The extreme rainfalls are the weather hazards that cause much damage and many casualties. The estimation of extreme rainfall is therefore of great interest to anticipate disasters such as floods and allow thoughtful planning of the territory. This study proposes through the approach values above a threshold of analyzing the probability distribution of extreme rainfall in Côte d'Ivoire. The series excesses over a threshold of 44 rainfall stations were analyzed over the period 1942-2002. The statistical analysis highlights optimal thresholds estimation of extreme daily rainfall between 32 mm and 55 mm. The majority (34) series of daily rainfall exceeding a threshold are adjusted by the generalized PARETO distribution with a shape parameter (k) significantly negative. The uncertainty in a parameter (k) is very important and increases the estimation threshold. Moreover, the spatial analysis of decadal and centennial daily rainfall shows that the highest values are mainly on the coast between latitudes 4˚30' and 5˚30' North, in the extreme West in mountainous areas and central West.