| dc.description.abstract | Water resource management is critical to economic development in Africa, but surprisingly lack of data and its quality undermines decision making in the water sector. Hydrological simulation is a powerful tool providing timely and useful information about streamflow, but requires streamflow data for model calibration and validation. The lack of data required for model calibration has brought the practitioners and scientists to come up with alternative methods namely regionalisation techniques. In this study, 17 catchments selected in the humid part of Kenya, essentially located in the mountainous areas, were calibrated in order to derive a regional model for calibration of ungauged catchments in the area. CHIRPS rainfall estimates and PET estimated from NASA POWER meteorological data have been harnessed to calibrate AWBM, the model selected for the study. Multiple linear regression was used to develop regional models relating defined catchments soil, topography, land use/cover based attributes and the model parameters obtained from calibration. Three models were developed including those for the average surface storage and the baseflow parameters BFI and Kbase. The surface recession constant’s model could not be established and default values were adopted. The successfully estimated parameter values and the default values, were used to simulate the daily streamflows. The average of the determination coefficient R2 were comparable, with R2 values of 0.64 and 0.65 over the calibration periods and 0.62 and 0.60 over the verification periods, respectively for calibrated and estimated parameters. Better quality data will substantially improve the method, however, in context of strict data scarcity, this method can be recommended for estimation of streamflow in ungauged catchments. | en_US |
