The hillslope length impact on SWAT streamflow prediction in large basins

The objective of this study was to assess the impact of hillslope length on Soil and Water Assessment Tool (SWAT) streamflow predictions in large basins using three methods for hillslope length calculation (the SWAT method, L1; a 3D analysis method, L2; a constant value, L3) combined with two DEMs (pixel size of 25 and 100 m), for a total of six DEML configurations that were tested in the Upper Danube (132000 km(2)). The delineation of subbasins and HRUs were kept unchanged in all configurations, thus isolating the DEM impact on streamflow from that of subbasins delineation. The configurations were independently calibrated in 98 gauged stations located in headwater subbasins (period 1995 similar to 2006), and validated in 150 gauged stations (period 1995 similar to 2009). The analysis of streamflow prediction was extended to its components (surface runoff, lateral flow and baseflow) using performance criteria and residual analysis, and the comparison of different components of water yield was pursued. Calibration and validation showed that all configurations simulated monthly streamflow acceptably (PBIAS < 25% for more than 70% of 150 gauged stations). DEM pixel size had negligible effect of streamflow and its components. The default hillslope length (L1) resulted in large overestimations of lateral flow. L2 resulted in the best performance as well as L3 method. Given that L2 method takes into account the topographic convergence of flow, the configuration of DEM100 and L2 is recommended for SWAT application in large basins in order to obtain reliable streamflow predictions.