Application of the SWAT hydrological model in a small, mountainous catchment in South Africa

Abstract

Natural water resources are globally under pressure due to increasing population numbers and associated global change drivers, such as land-use and climate change. South Africa has a semi-arid climate, with variable patterns of rainfall and is often referred to as a water scarce country. Much of South Africa’s fresh water originates in mountainous areas. It is important to correctly manage these mountainous areas and the fresh water resources they provide. Hydrological models could be a useful tool aiding water resource managers in accurately assessing and predicting hydrological processes in mountainous regions. Hydrological models can be used to predict the effect that changes in a catchment area, such as land use or climate change, will have on the associated water resources in the catchment. The aim of this study was to determine if the SWAT hydrological model could successfully simulate runoff from a small, mountainous catchment in South Africa. The SWAT model was applied to the B73A quaternary catchment located east of the Blyde River Canyon, close to Hoedspruit. This catchment is highly mountainous in nature. Observed stream flow data was obtained from the Department of Water and Sanitation (DWS) at a stream gauge located in the catchment. This observed data was used to calibrate and validate the model, using Sequential Uncertainty Fitting (SUFI-2) in the SWAT-CUP program. Results from calibration show good agreement between observed and simulated monthly stream flow data (NSE= 0.80). The model was able to bracket 68% of observed data in a small uncertainty band (r-factor = 0.67). The model was then validated using the same observed stream flow data during a different time period. Results for validation were again adequate (NSE= 0.46). In this case the model bracketed 71% of the data in a slightly larger uncertainty band (r-factor = 1.12). The study illustrates the potential of the SWAT hydrological model to be used in mountainous, semi-arid catchment areas in South Africa. Despite limited climate data and soil data, as well as the use of only one stream flow gauge location of observed data during calibration, which limited the incorporation of spatial variability within the catchment area into the model. Reliable rainfall data was obtained in the form of a rainfall station in the study area, highlighting the importance of this input variable in the SWAT model. Also highlighted was the need for appropriate calibration procedures to accurately represent the local characteristics of the modelled area. It was concluded that the SWAT hydrological model was able to adequately simulate the stream flow data from the B73A quaternary catchment area. This model could be a useful tool in predicting the effect of future land use and climate change scenarios on stream flow from the B73A quaternary catchment. It could be used for water resource management in this catchment.