A Numerical Modeling Study of the Coupled Variability of Lake Victoria in Eastern Africa and the Regional Climate
January 3rd, 2001
The objective of this investigation was to investigate and study the coupled atmosphere-lake climate system over the Lake Victoria basin, and determine the corresponding physical mechanisms that are involved. The primary research vehicle for the investigation is a fully coupled model of the regional climate of Eastern Africa and Lake Victoria which has been developed and applied in this study. The atmospheric component of the model is the NCAR Regional Climate Model (RegCM2). The lake component of the model is based on the Princeton Ocean Model (POM) configured for Lake Victoria by replacing the open boundaries in the standard version of the model with a closed coastline and adopting the bythemetry of Lake Victoria. The horizontal resolution is 20 km for both the atmosphere and lake model components.The results show that the bythemetry and geometry of the lake play a fundamental role in determining the climatology of Lake Victoria. There exists Kelvin-like waves in the thermocline trapped along the coast and they propagate clockwise around Lake Victoria with periodicity of about 30 days. The oscillations entirely disappear in the case of the isothermal conditions. The 3-dimensional model produces a surface temperature pattern indicative of horizontal lake water mixing associated with the horizontal spiral pattern that is not present in the 1-dimensional model. Preliminary comparison of the coupled RegCM2-POM model simulation results with the observations indicates that the model produces more realistic lake surface temperatures (LST) and rainfall over and around the lake than the standard version of RegCM2 in which a simple one dimensional thermal diffusion lake model is used. Over Eastern Africa, the regional climate variability is significantly influenced by the circulation over the Lake Victoria basin. Moisture advection contribution is important but secondary to evaporation in explaining the heavy rainfall over the lake. The interaction between the lake-land breeze and the prevailing northeasterly flow accounts for the asymmetry in the distribution of the diurnal rainfall variations and the southwestward movement of the dominant bands of divergence/convergence. During the 1982 El Nino when the averaged LST over the lake was higher than that during the normal year, the LST gradient was weakened along the SW-NE axis over the lake by the strong lake circulation. This results in LST distribution whereby the southwestern region of the lake is cooled while the region of maximum LST moves to the central-eastern region of the lake from the southwestern region of the lake. The net change in rainfall distribution over the lake during the 1982 El Nino is a combination of the effect associated with the large-scale convergence pattern and the meso-scale climate changes associated with the shift of the region of maximum rainfall toward the central-eastern part of the lake from the western part of the lake in response to the LST redistribution. Conversely, the weaker lake circulation enhances the LST gradient over the western part of the lake, especially over the northwestern region, and the rainfall maxima is still found over the northwestern sector of the lake. Therefore, the hydrodynamics of the lake play an important role in determining the coupled variability of the lake circulation and the lake basin-wide climatic conditions. This outcome based on the use of the coupled 3-dimensional lake model is not reproducible from the corresponding simulations based on the coupled 1-dimensional lake model. It is therefore apparent, that neglecting the lake's hydrodynamics and basing the lake model only on thermodynamical considerations deprives the coupled regional climate model of the ability to transport heat efficiently within the lake and thereby degrades the simulation of the climate downstream over the rest of the lake and the surrounding regions. The potential climate change resulting from total clearing of the tropical rain forests in Africa was also investigated by the standard version of the NCAR CCM3 global climate model. Over Eastern and Western Africa the impact of deforestation is primarily characterized by reduction in rainfall, however the CCM3 resolution of T42 which we have adopted may not be adequate to resolve the large contrasts in terrain and vegetation types. A striking result is that the strong remote response of the Southern Africa region to deforestation over Central Africa. This may be attributed to the role of the trapped large amplitude Rossby waves which transmit the response signal a long distance away from the source region. Based on the present results we infer that, the downscaling would be highly beneficial not only for the immediate region of Eastern and Central Africa where significant removal of tropical forest vegetation cover could occur in the coming decades, but also for the region further south to infer the projected detailed response of the Southern Africa region to remote deforestation effects.