Investigation and application of suitable model for rainfall-runoff modelling in the catchments of lake Victoria Basin.

Different rainfall-runoff models ranging from the system type, lumped conceptual and distributed models such as the HEC-HMS were applied and tested on selected catchments of Lake Victoria basin from the Kenyan side of the Lake. The selected catchments for the study were Sondu and Gucha river basins, which have catchment areas of 3260 km2 and 6755 km2 respectively. The missing river flow data was filled using procedures described in Kachroo et. al. (1988). The LPM was used in filling the missing river flow data. Rainfall and evaporation data were filled using the inverse distance squared method and seasonal means respectively. Lumped conceptual models, namely the NAM, SMAR and XINANJIANG were also applied to the two catchments under study. The models performances were improved to 84% and 63% for Sondu and Gucha respectively. A semi-distributed type of model known as the HEC-HMS was also applied to the Sondu catchment and gave an efficiency of R2 of 65%. Based on the investigation and application of the various hydrological models on the region, it has been proved that for the catchments under study, lumped conceptual models can be used as a better river flow forecasting tools in operational hydrology.