Sensitivity analysis of SWAT streamflow and water quality to the uncertainty in soil properties generated by the SoLIM model

<p dir="ltr">Hydrological models are essential tools for understanding natural processes, yet they often encounter with uncertainties due to limited input data and simplified assumptions. Soil data is a crucial input parameter for hydrological models, but it faces cartographic challenges due to lack of standardized soil mapping methods. To explore the applicability of soil datasets derived from the Third Law of Geography in hydrological models, the Soil Landscape Inference Model (SoLIM) based on this principle is used to generate a soil dataset. This dataset is then utilized as input data for a semi-distributed hydrological model (Soil and Water Assessment Tool, SWAT) to assess the impact of soil dataset uncertainty on performance of the hydrological model. This study, utilized the Fengyu River Watershed in Yunnan Province, China as an example, showed that: 1) Although the integration of SoLIM-generated soil data into the SWAT model introduces sensitivity in sediment yield, the overall results remain reliable. This suggests that soil datasets based on the Third Law of Geography can serve as input for semi-distributed hydrological models. 2) Uncertainty introduced by SoLIM-generated soil data affects parameters such as evaporation, total porosity of the soil layer (Φ_d), and groundwater infiltration coefficient (K_sat). These variations influence the sensitivity of SWAT’s predictions regarding streamflow and water quality. 3) Factors influencing model outcomes include steeper slopes in certain areas, which enhance streamflow velocity and flow rates, resulting in more dynamic and non-linear responses. Additionally, soil water dynamics exhibit higher temporal and spatial variability in forests compared to pasture and rice paddy fields, where vegetation tends to be more uniform, maintaining stable soil water loss pathways and reducing uncertainty in streamflow predictions. The study provides a foundation for studies addressing the difficulties in acquiring high-precision soil datasets for such models.</p>