Understanding the variability and predictability of seasonal climates over West and Southern Africa using climate models

Includes bibliographical referencesA good understanding of seasonal climate and the limit to which it can be predicted is crucial in addressing various socio-economic challenges in Africa. However, how to improve the capability of the dynamical models of the climate system in reproducing the regional seasonal climate variability and in replicating the role of various atmospheric circulation anomalies on the regional variability remains a major challenge. Thus far, understanding of seasonal climate over these regions, as well as the ability of climate models to predict them, has focused on the agreement of simulations of dynamical models of the climate system, rather than considering outliers as potentially vital contributors to understanding and predictability. This thesis uses discrepancy in a large ensemble of climate simulations as a tool to investigate variability in dominant seasonal rainfall and temperature patterns (i.e. classes) over West and Southern Africa, to examine the capability of climate models in reproducing the variability, and to study the predictability of the seasonal climates over South Africa. The dominant classes of variability (of rainfall and maximum temperature fields) in both regions are examined based on the Self-Organizing Map (SOM) classifications. The sequences in which each class occurs cannot be linked simply to a single common index of global scale atmospheric circulation anomalies, implying that the chaotic regional atmospheric circulations that modulate the global scale modes of variability are indispensable. The climate model examined adequately reproduces the dominant classes of seasonal climate over West and Southern Africa