Interactions of regional atmospheric and land surface processes with the West African monsoon system

The West African monsoon (WAM) rainfall is characterized by a strong temporal and spatial variability. The interplay of various large- and small-scale drivers generates frequent weather extremes such as droughts or floods, making West Africa one of the most vulnerable parts of the world in terms of food security. While it is known that teleconnections to sea surface temperatures determine the overall WAM regime, it is still uncertain by which margin and at which scales regional atmospheric and land surface processes may modify it. In this data scarce region, atmospheric models are indispensable tools allowing physically consistent interaction experiments on different components of the monsoon system. This thesis uses the Weather Research and Forecasting model (WRF) driven by reanalysis data to investigate the relevance of regional moist processes, convection and vegetation patterns for the WAM regime and presents the abilities and limits of the model to properly capture the involved processes. Furthermore, the possibility to improve WRF by adjusting it to regional characteristics is analysed. This includes the choice of a favourable model set-up based on a region-specific parameterization classification, the explicit treatment of convection and the implementation of satellite-derived land surface parameters. A mixed-physics ensemble with 27 parameterization combinations is used to evaluate the effect of regional moisture distribution on the WAM for the rainy season 1999. Although all ensemble members use the same boundary forcing, the ensemble spread covers the whole range of dry to wet monsoon regimes observed in the Sahel between 1979 and 2010. The most rigorous shift from wet to dry monsoon conditions was found to be related to an increase of low- and mid-level clouds weakening the incoming solar radiation and hence the sea-land pressure gradient. In particular, significant large-scale changes in precipitation are always linked to a change in the intensity of the pressure gradient and thus of the moist Hadley-type meridional circulation that connects the monsoon winds to the Tropical easterly jet. This shows that regional moist processes may indeed alter the monsoon dynamics. A closer look at the convective processes reveals that explicit instead of parameterized convection considerably improves both the precipitation characteristics as well as the incoming shortwave radiation associated with the modelled cloud cover. This confirms convection as a key process for the monsoon circulation since it affects the water and energy balance not only in the atmosphere but also at the surface. In turn, the partitioning of surface energy and moisture fluxes may affect the location and frequency of convective systems. Land cover and vegetation play a crucial role in this partitioning. To investigate the effect of observed interannual vegetation changes between 2009 and 2010 on the WAM precipitation, novel high-resolution satellite-derived dynamical datasets for vegetation fraction, albedo and leaf area index are implemented into WRF. The two years exhibit opposing vegetation anomalies. In comparison to a climatological land surface, the vegetation changes exhibit the strongest effect on latent heat fluxes and associated surface temperatures. Moisture divergence (convergence) and a decrease (increase) of rainy hours is found over regions with higher (lower) vegetation fraction during the day and the opposite during the night. These effects cancel out when averaged over larger regions, leading to negligible changes in total precipitation amounts. An improvement of modelled rainfall through the integration of observed dynamical surface information with respect to observations was only detectable in the Sahel region. These findings suggest that both regional atmospheric and land surface processes may trigger significant changes in precipitation amounts or shifts of the monsoon rainband when they affect the large-scale temperature gradient over land, which is always connected to a change of the large-scale pressure gradient driving the WAM. Accordingly, temperature changes at smaller scales rather affect local moisture convergence or divergence and therefore the rainfall distribution instead of total amounts. The results presented in this thesis contribute to further improve our picture of factors for WAM variability, which forms the basis for any practical measures that could improve the resilience of the West African population whose livelihood still depends on rainfed agriculture