Abstract
The convective activity over the Rift Valley lakes of East Africa, as deduced from cloud tops colder than a predefined threshold, is examined. Relationships between satellite-derived convective indices and rainfall measurements are also examined. The diurnal cycle of convective activity over Lake Victoria and over the land is analyzed. The maximum convection is found to occur during the morning between 0500 and 0800 LST over Lake Victoria, and a second maximum occurs in the afternoon. In contrast, over surrounding land, the maximum occurs generally in late afternoon and during the evening. It is also found that a linear relationship exists between satellite-derived convective indices and rainfall measurements; the correlation between the two is strong enough that the indices can be used to estimate annual and areally averaged monthly rainfall. The cold cloud indices explain more than 50% of observed variances of rainfall for the months of May through October. However, the performance is inadequate in several instances during February and March. The results show that the satellite algorithm is robust enough to estimate spatial averages of monthly rainfall with satellite estimates accounting for between 75% and 95% of observed variances of rainfall. The results further show that there is an exceedingly high correlation between convective rainfall over Lake Victoria and in the surrounding catchment. This permitted the derivation of a relationship between rainfall over the lake and its catchment.