Estimation of Stream flow in ungauged catchments of the upper Tana Basin , Kenya

Abstract

A regionalisation methodology was applied to catchments of the Upper Tana basin to enable estimation of daily stream flows for any catchment within the region for which physical attributes data and records of rainfall and temperature are available. IHACRES, a lumped conceptual rainfall-runoff model was calibrated at a daily time step to six catchments ranging in size from 49km2 to 600 km2 within the upper Tana basin to obtain a set of model parameters that characterise the hydrological behaviour of the catchments. Physical catchment characteristics representing topography, soil and land use were derived from spatial data using GIS. By correlating these two sets of parameters, equations were developed which related the conceptual parameters to catchment characteristics and which enabled the estimation of model parameters from catchment characteristics. The estimated parameters were used to simulate stream flows in two validation catchments within the same region and the goodness of fit evaluated using statistical and graphical methods. Sensitivity tests were carried out at one of the catchments in which flow response to variations in model parameters was assessed in order to analyse calibration errors. The calibration R2 ranged from 0.57 to 0.85 while the simulation R2 ranged from 0.55 to 0.77 for all catchments. The Nash-Sutcliffe (1970) efficiency ranged from 0.78 to 0.91 for calibration and from 0.77 to 0.88 for simulation. The stream flows simulated using the estimated parameters agreed well with the observed stream flow series. The R2 values were 0.21 and 0.67 while the Nash-Sutcliffe efficiency values were 0.21 and 0.68 respectively. Although varied, the results indicated that regionalisation has the potential to generate stream flow data in ungauged catchments for purposes of water resources assessment.