| dc.description.abstract | River flooding has been associated with severe social and economic problems throughout
the world. While deaths from most natural disasters have declined over the past two
decades, loss of lives and property from flooding has increased. Floods occasionally
cause disasters in Kenya and the Nyando River basin is among the areas adversely
affected. In hydrology, rainfall-runoff models enable users the ability to forecast the
runoff from a catchment from the amount of precipitation received by that catchment.
This study was conducted with the aim of establishing a hydrological model that can
reasonably depict the relationship between rainfall and runoff as a first step to
establishing a flood early warning system for the Nyando basin. The study involved the
selection of a hydrological model that could be suitably domesticated to the Nyando river
basin characteristics to simulate discharge for flood management. The main criteria for
the model selection were the nature of basic algorithms, process-based and deterministic
approach to input or parameter specification and the spatial representation was to be semidistributed.
In addition, the availability of the required input data and the suitability to
flood modelling were considered. HEC-GeoHMS software was used to delineate the
Nyando catchment from a 90 m Shuttle Radar Telegraphic Mission DEM and to
expediently create hydrological inputs that were used directly with the selected model,
HEC-HMS Version 3.1 to model the catchment. The HEC-HMS model was calibrated
and validated for the Nyando catchment in the period 1980 to 1983 and 1984 to 1991
respectively. In the calibration of the model, correlation coefficient and the Nush-Sutcliffe
efficiencies were same and equal to 0.64 while the BIAS efficiency was 0.002. In
validation the correlation coefficient and Nush-Sutcliffe efficiency were equal to 0.66 and
the BIAS efficiency was 0.051, values indicating that the model was well optimized. To
demonstrate the application of the model, it was used to asses the effects created by the
inclusion of two proposed reservoirs on peak discharges for the period 1984 to 1991. The
peak outflow which was predicted to be on 01 May 1988 at 104.8m3/sec was reduced to
91.0 m3/sec by the inclusion of the two dams. This translated to a reduction of the peak
discharge by 13.2%. From the modeling efficiency values, it can be concluded that HECHMS
model was successfully adapted to the Nyando catchment and can be accepted as an
important tool in operational hydrology for estimating information required for water
resources planning, design, and operation and for flood control and monitoring. | en |
