Integrating remote sensing is beneficial for watershed model but the effects are spatially and temporally heterogeneous

Watershed processes exhibit notable temporal and spatial variations under climate change, which can be effectively captured by remotely-sensed datasets with global coverage and high spatiotemporal resolution. These datasets precisely capture spatiotemporal dynamics of vegetation and evapotranspiration, providing constraints and corrections for watershed simulations. However, the effects of integrating remote sensing datasets on hydrological and nutrient variables, and their interactions in watershed simulations, have not yet been fully investigated due to their complexity and spatiotemporal heterogeneity. This study integrates remote sensing leaf area index (LAI) and potential evapotranspiration (PET) datasets into a watershed model and evaluates the effects of different integration scenarios. Compared with the MODIS dataset, the original model underestimated the LAI and PET data by over 20%. The simultaneous integration of LAI and PET resulted in the greatest improvement in model performance, with NSE increasing by 19%, 26%, and 25% for streamflow, nitrogen, and phosphorus, respectively. Additionally, the simultaneous integration of the LAI and PET caused partial offsetting effects, indicating that the improvement from integrating additional datasets into the watershed model is not linear. This study investigates the spatiotemporal heterogeneity of the effects derived from dataset integration and proposes optimizing strategies, which can enhance watershed simulation accuracy and exhibit potential for broader applicability in humid subtropical monsoon climate regions.