Spatiotemporal variability analysis of nutrient and sediment loads in surface water using improved hydrologic modeling for depression-dominated watersheds.

Identifying pollutant hotspots and key influencing factors is crucial for mitigating nutrient and sediment pollution, which often involves collinearity among multiple independent variables. Spatial autocorrelation and Geodetector help address this collinearity challenge. The objective of this study was to examine the spatiotemporal variability of nutrients and sediment hotspots and identify the key driving factors. A recently developed Dynamic Partial Contributing Area (DPCA) model was coupled with the Soil and Water Assessment Tool (SWAT) and applied to a watershed in North Dakota, USA. Simulation results were used as inputs for Moran's I spatial autocorrelation test to explore spatiotemporal variations of nutrients (total nitrogen (TN) and total phosphorus (TP)) and sediment hotspots. Additionally, Geodetector was employed to identify the key factors inducing these pollutions. Flood frequency analysis based on the 90-year peak discharge dataset was performed to observe variations in hotspots. The spatial autocorrelation results indicate a stronger correlation during dry periods compared to wet periods. Flood frequency analysis further underscored these findings, revealing that hotspots were more pronounced during periods of lower flood frequency. The Geodetector results showed that the interaction of precipitation and land use land cover (LULC) is the key indicator responsible for TN and TP pollution during wet periods, while the interaction of precipitation and cropping management factor emerges as the most significant influencing factor for sediment pollution during wet periods. This study demonstrates effective methods for identifying pollutant hotspots and their key driving factors, which will be helpful as theoretical guidance for best management practices (BMPs) for nutrient and sediment pollution.