The dynamics of water availability within a region can be quantitatively analyzed by partitioning the water into blue and green water resources. It is widely recognized that vegetation is one of the key factors that affect the assessment and modeling of blue and green water in hydrological models. However, SWAT-EPIC has limitations in simulating vegetation growth cycles in subtropics because it was originally designed for temperate regions and naturally based on temperature. To perform a correct and realistic assessment of changing vegetation impacts on modeling blue and water resources in the SWAT model, an approach was proposed in this study to modify the SWAT plant growth module with the remotely sensed leaf area index (LAI) to finally solve problems in simulating subtropical vegetation growth, such as controlling factors and dormancy. Comparisons between the original and modified model were performed on the model outputs to summarize the spatiotemporal changes in hydrological processes (including rainfall, runoff, evapotranspiration and soil water content) under six different plant types in a representative subtropical watershed of the Meichuan Basin, Jiangxi Province. Meanwhile, detailed analysis was conducted to discuss the effectiveness of the modified SWAT model and the impacts of vegetation changes on blue and green water modeling. The results showed that (1) the modified SWAT produced more reasonable seasonal curves of plants than the original model. E<sub>NS</sub> (Nash-Sutcliffe efficiency) and R<sup>2</sup> increased by 0.02 during the calibration period and accounted for an increase of 0.09 and 0.03, respectively, during the validation period. (2) The comparison of model outputs between the original and modified SWAT suggested that evapotranspiration was more sensitive to vegetation changes than other components of green water. In addition, vegetation presented conservation capability in the blue water. (3) The variation in blue and green water resources with different plant types after modifying the SWAT model showed that seasonal changes in vegetation led to a significant difference between forest and non-forest areas.
