Effects of Land-use / Land Cover Changes on Flow and Sedimentation from the Metsimotlhabe River Catchment Using Soil and Water Assessment Tool (SWAT) Model

Abstract

This study was undertaken to assess the effects of Metsimotlhabe River catchment Land-use / Land Cover dynamics (2006 and 2018) on the inflow and sedimentation into the Bokaa Reservoir using the SWAT model implemented as ArcSWAT. The ArcSWAT simulation was performed for 13years of recording periods starting from 2006 to 2018 with the first 4years used as a warm-up period, 2010 to 2012 as model calibration period, and 2013 to 2015 as model validation period. The calibration was performed against flow quantity on the model run with 2006-Land-use Land Cover (LULC) (Scenario 1) as input and the best parameter values were applied to the ArcSWAT through the manual calibration tool in ArcSWAT. The calibrated parameters were also applied to the ArcSWAT project created with 2018-LULC (Scenario 2). The statistical model performance measures were computed for the separate periods using the SWAT-CUP model. As a result, the ArcSWAT model was well-calibrated against the flow as indicated by the statistical analysis. With Scenario 1 the R 2 obtained for the calibration, validation, and overall periods were; 0.72, 0.89 and 0.82 respectively, whereas in Scenario 2 the R 2 were 0.69, 0.86 and 0.78 for calibration, validation and overall periods respectively. The change in LULC that occurred between 2006 and 2018 produced larger peaks of flow and total sediment yield from the catchment. The total sediment yield produced for the entire simulation period increased by 117.2% when the most recent LULC was used (Scenario 2, 2018). These changes could be attributed to a 2.27% increase in built-up areas and an 8.0% and 7.89% decrease in forest and shrubland areas respectively due to the 2% increase in the annual population of people in the catchment. A positive correlation with R 2 of 0.965 was achieved between the sediment yield simulated by the flow-calibrated ArcSWAT and sediment yield derived from applying remote sensing spectral analysis