Assessing the impact of soil data resolution on streamflow prediction with SWAT in a mesoscale headwater basin of Sri Lanka

Study region Upper Kotmale basin, Sri Lanka. Study focus The prediction accuracy of hydrological models hinges profoundly on the quality of input data, serving as a cornerstone for reliable simulations and informed water resource management. This study evaluates how varying resolutions of soil data impact streamflow predictions when utilized as input for Soil and Water Assessment Tool (SWAT). The comparison of SWAT simulations involved input soil data sourced from International Soil Reference and Information Centre (ISRIC) with a resolution of 250 m, alongside a soil map produced at 30 m resolution using the Soil Land Inference Model (SoLIM). The catchment was delineated using a constant drainage area threshold to ensure the same number of subbasins, while other data were held constant to isolate soil data effects. New hydrological insights for the region The high-resolution data nearly doubled the number of Hydrological Response Units (HRUs), significantly improving model capacity to represent soil heterogeneity and simulate key processes like infiltration, baseflow, and soil water content, especially under flood pulses and low-flow conditions. While calibration masked some performance gaps, SoLIM-based simulations offered better fidelity in capturing hydrological dynamics. These findings highlight that soil data resolution influences not only streamflow, but also the timing and behaviour of extreme events crucial for hydrological modeling in data-scarce, topographically complex basins where planning and risk mitigation depend on reliable model outputs.