Chetty, Prinavan Carl2023-04-112023-04-112022https://hdl.handle.net/10539/34941A dissertation submitted in partial fulfilment of the requirements for the degree of Master of Science in Engineering to the Faculty of Engineering and Built Environment, School of Civil and Environmental Engineering, University of the Witwatersrand, Johannesburg, 2022The estimation of areal rainfall from point measurements by rain gauges are key inputs into various water resource modelling applications including streamflow simulation. A major uncertainty in these estimates arises from the inadequacy of rain gauges in capturing the variation of rainfall between rain gauges. This study aimed to quantify monthly areal rainfall uncertainty and assess the impacts of this uncertainty on streamflow simulation. This was achieved by: i) quantifying areal rainfall uncertainty and generating monthly stochastic areal rainfalls incorporating this uncertainty; ii) using the stochastically generated areal rainfalls to generate stream flows, and iii) comparing variability and quality of these stream flows with those obtained using historical rainfall. A stochastic areal rainfall generator that has been used in previous studies was selected to quantify areal rainfall uncertainty on seven catchments in South Africa; while the assessment of streamflow simulation was carried for two of the catchments using the Pitman model. A manual-automatic calibration approach using the SCE-UA optimizer was applied. The assessment of areal rainfall uncertainty with alternate omission of 4 rainfall stations obtained an average monthly areal rainfall difference equal to 18.3% of the average monthly rainfalls. The generated stochastic rainfalls were unbiased with the average mean, standard deviation and skewness matching the respective historic values closely. The streamflow simulation revealed that incorporating areal rainfall uncertainty by using stochastic areal rainfalls lead to a significant increase in the range of simulated stream flows without significant loss of simulation performance in calibration and validation runs. The areal rainfall uncertainty quantification applied is therefore considered realistic.enDissertation