Abstract: To protect the water quality of the Xinfengjiang Reservoir and monitor the eutrophication risk，this study developed an analytical model for remote sensing inversion of chemical oxygen demand （COD_Mn） based on the underwater radiative transfer process. This model comprehensively takes into account three water quality parameters that influence the underwater light field：chlorophyll a，total suspended matter，and COD_Mn. The model was applied to conduct multi-temporal monitoring of eutrophication in the reservoir and its surrounding rivers. Through accuracy verification，the model achieved a root mean square error of 0.68 and a mean absolute percentage error of 25.22%，demonstrating its reliability in complex water environments. The spatiotemporal analysis of the water quality in Xinfengjiang Reservoir revealed the consistent good quality of the main body over the long term. However，due to extensive aquaculture and anthropogenic discharges，the Zhongxin River exhibited frequent eutrophication，which may pose a potential threat to the overall water quality of the reservoir. It is recommended to enhance monitoring of the Zhongxin River，promptly address illegal discharges，and implement ecological engineering measures such as vegetative drainage ditches in the watershed. These efforts can effectively reduce agricultural non-point source pollution，contributing to the restoration and improvement of the ecological environment of Xinfengjiang Reservoir.