Effects of Digital Elevation Model data resolution on hydrological modelling for assessing land use/cover changes using SWAT in the Swartkops river catchment

Abstract

Digital Elevation Models (DEMs) play a critical role in hydrological modelling, with resolution significantly influencing simulation accuracy. This study investigates the effect of DEM resolution on the SWAT model's performance in simulating hydrological responses to land-use and land-cover (LULC) changes in the Swartkops River catchment, South Africa. Supervised classification of Landsat images from 1990 and 2022 using the Maximum Likelihood algorithm in IDRISI TerrSet achieved overall accuracies of 87.11% and 94.89%, respectively. Results revealed notable LULC changes, including a 9.49% decrease in bare areas and a 5.97% increase in built-up areas. Agricultural and vegetated areas increased slightly, while water bodies declined. SWAT simulations using 20m, 30m, and 90m DEMs indicated that the 30m DEM provided the best performance, with NSE values of 0.49 (calibration) and 0.67 (validation). Runoff contributions in 1990 were highest from built-up and bare areas (36% each), while vegetated areas contributed 28%. In 2022, increased runoff was observed from agricultural land. Field validation with Gerlach troughs confirmed the highest runoff in built-up (9.3 L) and bare (6.6 L) areas. These findings emphasise the importance of DEM resolution in SWAT modelling and the influence of LULC changes on runoff, aiding effective catchment management.