| Abstrakt: |
Long-term stream discharge data is indispensable in irrigation and drainage design. However, in Uganda, this data is poor and insufficient, limiting irrigation system design. Conversely, the rainfall monitoring network is denser than the river flow monitoring network. Therefore, we attempt to build a model that calculates river discharge from input of rainfall. In this study, the lumped parameter Tank Model was applied. The model was applied to the Namatala River catchment (155 km²) in Eastern Uganda. The study sought to ascertain the applicability of a lumped parameter model to a mid-sized catchment. Specifically, the objectives were: 1) To calibrate the numerical values of Tank Model parameters, 2) To verify the Tank Model parameters. This Tank Model required daily rainfall, evapotranspiration and river discharge data during calibration. Data for years 2015 and 2016 was used for calibration and validation respectively. During calibration, Monte Carlo simulation was used to find the numerical values of 16 Tank Model parameters. The best performing calibration parameter set had Nash-Sutcliffe (NS) efficiencies of 0.608 and 0.257 in calibration and validation respectively. However, among the 2015 calibration parameter sets, the one with a calibration NS of 0.502 performed best in validation (NS = 0.526). Equifinality was observed during parameter calibration. By using Tank Model, simulated discharge was divided into its surface runoff, interflow and base flow components. Tank Model was adaptable to Namatala River catchment. [ABSTRACT FROM AUTHOR] |
