Abstract:
Floods are the most common and widespread climate-related hazard in the Lake Victoria region. However, significant delays in ground data availability have made it unfeasible to use traditional flood forecasting systems. Satellite rainfall estimates have been identified as readily and economically available data that can be used as input to run hydrologic models and produce flood-warning systems. The aim of the study therefore is develop a simple and locally viable alternative approach to circumvent the absence of reliable ground measurements by using satellite rainfall estimates for forecasting and management of floods in the study area. The satellite-derived rainfall estimates (RFE) were first evaluated using historical rainfall data for the Nyando basin corresponding to the locations of 35 gauging stations in the basin for the period 1995 to 2005. A Digital Elevation Model (DEM) of the basin was used to generate the drainage patterns of the basin. The land cover of the study area and the digital soil map are incorporated in the system. The study applies daily driven satellite-derived rainfall and the pixel based Curve Number method for spatially distributed hydrologic streamflow modelling and flood forecasting. Rainfall–runoff relationships results of the area obtained in a spatial scale are then tested on their capabilities as a flood early-warning system by comparing them with historical streamflow. The approach was further tested using RFE for the period 2006 to 2012. The results for comparisons at daily accumulations of RFE with observed rain gauge data are not satisfactory but they performed reasonably well in detecting the occurrence of rainfall. The products show significant results for 10-day accumulation where regression analysis yielded on average, a correlation coefficient (r) of 0.78. While graphical plots of daily-observed stream flow against simulated streamflow show a poor agreement, there is indication that when there is a drastic surge in the runoff volume, a rise in the river level is to be expected and it enables the prediction of the occurrence of floods and the issuance of early warnings.
