Techniques for calibration and validation of SWAT model in data scarce arid and semi-arid catchments in South Africa

• Published in: Journal of hydrology. Regional studies Vol. 25; p. 100621
• Main Authors: Mengistu, Achamyeleh G., van Rensburg, Leon D., Woyessa, Yali E.
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.10.2019; Elsevier
• Subjects: Arid catchments; Calibration; Hydrologic models; model uncertainty; model validation; Regionalization; rivers; Soil and Water Assessment Tool model; South Africa; SWAT model; uncertainty analysis; watersheds; South Africa; SWAT model; Regionalization; South Africa; Calibration; Hydrologic models; Arid catchments
• ISSN: 2214-5818; 2214-5818
• DOI: 10.1016/j.ejrh.2019.100621

•Calibration of hydrologic models is challenging in arid and semi-arid areas.•The regionalization with physical similarity approach was employed for calibration.•SWAT model was successfully applied by a dual calibration and validation approach.•This study suggests some best practices to minimize model uncertainties.•Satellite data became a good alternative for calibrating models in arid areas. This study was conducted in Soutloop River Catchment, Northern Cape, South Africa. Although hydrologic models play a critical role in the management of natural resources in arid areas, their application is challenged by the scarcity of data for calibration and validation. Therefore, this study aimed at to configure, calibrate and validate SWAT model in a data-scarce catchment by using the regionalization with physical similarity approach. This approach uses dual calibration and validation procedure, i.e., one in the donor catchment (by using SWAT-CUP (SWAT Calibration and Uncertainty Programs) and the other on the study catchment (by manual calibration and verification). Based on the sensitivity analysis, sixteen parameters were calibrated by SWAT-CUP. The result from the uncertainty analysis indicated acceptable values of both the R-factor (0.8***Values are the averages of the calibration and validation procedures.) and P-factor (0.7**). The model performance evaluation also showed acceptable ranges of values (e.g., NS was 0.76** and R2 was 0.78**). However, the main calibration and validation process was conducted outside the target catchment, though it was assumed that the donor and target catchments have similar hydrological responses. Therefore, the study suggested further inspection methods to minimize the model uncertainty in the study catchment. This study enables researchers to exploit the river eco-regional classifications of South Africa to apply hydrologic models to estimate the components of water balance in arid/semi-arid catchments.