Abstract
A nested coupled model has been developed to investigate the two‐way interactions between the regional climate of eastern Africa and Lake Victoria. The atmospheric component of the model is the North Carolina State University (NCSU) version of the National Center for Atmospheric Research (NCAR) regional climate model (NCSU‐RegCM2). The lake component of the model is based on the Princeton ocean model (POM).
Three simulations, each 4 months long, have been performed for the short rains of eastern Africa of September through to December. The control experiment is based on the standard NCSU‐RegCM2 model coupled to a one‐dimensional model of Lake Victoria. The second experiment was based on the stand‐alone three‐dimensional primitive equation POM–Lake Victoria model forced by output from the atmospheric component of the control run. The third experiment is based on the integration of the coupled system of the NCSU‐RegCM2 model where the one‐dimensional lake model in the control run has been replaced by the three‐dimensional POM hydrodynamical model for Lake Victoria.
The results confirm that adopting the traditional modelling approach, in which the lake hydrodynamics are neglected and the formulation is based entirely on thermodynamics alone, is not entirely satisfactory for the Lake Victoria basin. Such a strategy precludes the transport of heat realistically within the lake, from the heat surplus regions to the cooler regions, and thereby results in a degraded simulation of the climate downstream over the rest of the lake and the surrounding land regions. The numerical simulations show that the southwestern region of the lake is an important source of warm water because it is relatively shallower and the water column is heated up much more quickly during the day than the rest of the lake. The result is that the surface temperature anomaly field from the all‐lake area average consists of a gradient pattern with warmer water over the shallow region of the lake over the southeastern sector and a colder pool of water over the northeastern region, where the lake is relatively deeper. This pattern is also reproduced by the one‐dimensional lake model. Some of the excess heat over the southeastern region is transported to the colder and deeper region over the northeastern part of the lake by prevailing surface wind flow. Through the lake–atmosphere coupling, the resulting asymmetric lake‐surface temperature distribution modifies the overlying wind circulation, which in turn reduces the cloud cover and rainfall. This secondary feature in the surface temperature structure cannot be generated by the traditional nested climate models, such as the standard version of the NCAR‐RegCM2 model, since the simple static lake model formulation is not capable of supporting horizontal mixing of water. Comparisons show that this feature is weaker in the RegCM2‐POM coupled model than the corresponding pattern that we obtained in our previous study based on the ‘stand‐alone’ POM lake model. In contrast, from the simple classical text‐book theoretical model of the lake–land breeze phenomena, the simulated surface wind circulation and rainfall distribution are highly asymmetric across the lake. Copyright © 2004 Royal Meteorological Society