LAUSR

This study explores the relationship of Digital Elevation Model (DEM) resolution and flow accumulation threshold value to watershed characteristics and streamflow in two Nepalese catchments, Kaligandaki and Bagmati basin. DEMs from three different sources: ASTER Global DEM v2, SRTM v4.1, and Contour of Nepal, with resolutions of 30 m, 90 m, and 250 m were used. Automatic watershed delineation in SWAT was used to derive the stream network, and basin and flow characteristics at different DEM scenarios. It is found that DEM source and resolution influence basin characteristics and stream network morphology. The assumption that SWAT model performance improves with an increase in DEM resolution did not hold true. All the models were equally efficient in predicting the runoff. The model performance for both basins was similar regardless of the origin of the DEM. NSE and R2 varied in the range 0.726–0.781 and 0.772–0.81, respectively, for Kaligandaki basin. Similarly, for Bagmati basin, NSE and R2 varied in the range 0.772–0.837 and 0.779–0.839, respectively. The drainage density and simulated flow got better when the threshold area was decreased. However, the model ceased to improve any further after the optimum value of the threshold area. Our results indicate that DEM of higher resolution (30 m or finer) needs to be used to model the catchment in SWAT accurately and appropriate threshold area needs to be used for best model performance based on the temporal sensitivity of the runoff.