Analysis of decadal variability and climate change in West Africa using CMIP5 products - Application to estimate agricultural yields at the end of the century

Abstract

The West African monsoon is characterized by high decadal and multi-decadal variability whose impacts can be catastrophic on local populations. The factors advanced to explain this variability put in competition the role of sea surface temperatures and atmospheric dynamics related in particular to Saharan Heat Low. In addition, the emergence of the climate change footprint on the West African Monsoon, linked to the increase in greenhouse gas emissions,involves regional effects (radiative forcing on the Saharan atmospheric circulation) and global effects (radiative forcing on sea temperatures). This thesis addresses these questions by comparing the sets of control and historical simulations of climate models carried out in the CMIP5 project with observational data from the 20th century.Through multivariate statistical analyses, it has been established that decadal modes of ocean variability (AMO, IPO and IDV) and decadal variability of Saharan atmospheric dynamics significantly influence the decadal variability of monsoon rainfall. These results also suggest the existence of an external anthropogenic forcing that is superimposed on the natural decadal variability inducing signal intensification in historical simulations compared to control simulations. In addition, we have shown that the decadal variability of rainfall in the Sahel, once the influence of oceanic modes has been eliminated, appears to be driven mainly by the activity of the Arabian Heat Low in the central Sahel and by the meridional temperature gradient structure in the western Sahel over the intertropical Atlantic.Moreover, the long-term evolution of the West African Monsoon over the period 1901-2099 is reflected in climate simulations by precipitation patterns over West Africa that are quite different from one model to another, which have been grouped into five categories, which may be very different from the multi-model average. We have also shown in these climate projections an increa