Publications scientifiques
* (Auteur correspondant)
HAL (Le Centre pour la Communication Scientifique Directe)
The occurrence of extreme hydro-climatic events such as floods and severe droughts has adverse impacts on society and economy of African countries, especially those of south of the Sahara, because of their low levels of development, which is associated with the high vulnerability of their populations. A study that analyzes their occurrence and their evolution, for the forecasting purposes, is a necessity for the region. This thesis focuses on the occurrence of the July–September (JAS) extreme rainfall and droughts in the northern Africa. Specifically, it addresses the issue of modeling their occurrences, estimating their return periods (RP) and their return levels (RL) in the context of stationary and non-stationary climate, and to evaluate the influence of the large-scale modes of climate variability (L-SMCV) on both their occurrence and the intensity of their RL. To conduct this investigation, we used monthly rainfall data from the Climatic Research Unit of the University of East Anglia in England, over the period 1901–2014; and those of four climate indices—the El Niño southern oscillation, the North Atlantic oscillation, the Pacific decadal oscillation, and the Atlantic multi-decadal oscillation—from the Global Climate Observing System Working Group on Surface Pressure. Two classical statistical approaches derived from the extreme value theory were used to identify and model the JAS extreme hydro-climatic events in the northern Africa: the peaks-over-threshold approach, which models the excesses above the threshold according to a generalized Pareto distribution, and the block maxima approach, which supposes that the maxima follow a distribution of generalized extreme value. The results reveal a good fit between these statistical approaches and the JAS extremes—wet and dry extremes—across the northern Africa. It appears that the areas with high occurrences of JAS extreme rainfall for a fixed RP are those of Guinean coastal regions and those of highlands on the mainland. In general, over most northern Africa, the RL of JAS extreme rainfall have decreased after 1970, while there was an increase in RLs of JAS extreme droughts in the West African Sahel with respect to 1901–70 period. In Sahel as a whole, the JAS extreme rainfall will intensify in the future, whilst in its western part, the JAS dry extremes are expected to amplify after 2009. The assessment of the influence of L-SMCVs JAS extreme rainfall indicates that the Atlantic multi-decadal oscillation significantly impacts RLs in semi-arid area, and therefore, can be considered as a predictor of JAS extreme rainfall in this part of the northern Africa. The most extreme years have been detected and sometimes associated with an El Niño oscillation phase. The length of RP associated with these years depends on their intensities and it varies from one area to another. As a result, an investigation that deals with the modeling of hydro-climatic events in northern Africa requires first an analysis that takes into account the spatial heterogeneity of the rainfall distribution, and then, should be performed on distinct climatic zones.
