Evolution of the rainfall regime in the West African Sahel : detection, elements of attribution and projections

Abstract

Anthropogenic global warming has consequences on the hydrological cycle and in particular on rainfall at regional scales. Rainfall in the Sahel, driven by the West African monsoon, is characterized by significant variability over a wide range of spatial and temporal scales as well as by high sensitivity to global climate fluctuations. The socio-economic stakes of the region are particularly high, with, on the one hand, an essentially rain-fed agriculture that is vulnerable to droughts and, on the other hand, an increase in flood flows associated with strong demographic growth and little or unplanned urbanization, exacerbating the risk of flooding. In this context, this thesis aims to document and understand past changes in the sahelian rainfall regime in order to anticipate its future evolution.Existing tools allowing a fine description of the statistical properties of rainfall -- especially extremes -- over the region are adapted to a context of temporal non-stationarity. This framework has allowed to robustly demonstrate the increase in intensity of extreme rainfall events over the region, at time steps ranging from sub-hourly to daily. It was also shown that the strongest events are those whose frequency increases the most. These trends have been expressed in terms widely used in the field of hydrological engineering in order to promote the appropriation of decision support tools and the implementation of adaptation practices.The latest numerical climate simulations are then used to better understand the factors responsible for these changes. These simulations, which represent remarkably well the rainfall regime evolution observed since 1950 in the region, have highlighted the major role of anthropogenic climate forcing factors in the rainfall intensification: aerosols seem to be the main contributors to this trend, with an additional role of greenhouse gases. These forced signals were modulated in time and space by the internal variability of the o