南水北调中线工程自通水以来使得华北地区，特别是北京等大型城市的用水安全得到了更好的保障。但近年来由于水源地丹江口水库的来流量减少以及汉江流域生态环境保护战略的提出，引江补汉工程提上日程。由于丹江口水库是南水北调中线工程的水源地，因此该水库水质对中线调水的水质保障具有重要意义。为了科学评估引江补汉工程坝上方案对丹江口水库水质的影响，本文针对丹江口水库构建了基于MIKE21的水动力-水质耦合模型以及基于BP神经网络模型的水质模型，对调水后的丹江口水库水质进行了模拟和分析。本文收集了丹江口库区2017年-2019年逐日气象数据、汉江等主要入库支流逐日流量、丹江口水库水下地形、引江补汉调水规划方案以及主要水质监测断面逐月数据等资料，分别基于MIKE21和BP人工神经网络构建了丹江口水库水质模型，利用实测水质数据进行了模型的率定和验证，较好地模拟预测丹江口水库的溶解氧、氨氮、高锰酸盐指数与总磷等指标的变化。计算所得的各水质指标纳什效率系数范围分别为0.60~0.84和0.69~0.87。BP神经网络的运行效率较高，但MIKE21模型能够更全面地反映库区浓度场的空间分布。模拟预测结果显示，调水工程对丹江口水库的水质影响范围和影响程度均不大，氨氮和高锰酸盐指数的水质空间分布情况基本保持不变；堵河口至坝上（即汉库）有可能在年内不同时期发生总磷浓度升高的情况，约有占比8%的水域面积年均总磷浓度降为III类水标准；陶岔附近的水质基本不受影响，根据MIKE21模拟结果，在最不利工况下，陶岔的氨氮与高锰酸盐指数水平受调水工程影响会有小幅波动但水质分级不变，仍能够保持I、II类水标准；总磷浓度则有所上升，预测可能会发生最长7日的水质降级，但能够满足地表水III类水标准；BP神经网络模型的结果显示氨氮仍满足I类标准，高锰酸盐指数可能发生最长8日的水质降级，总磷可能发生最长16日的水质降级，但同样能满足III类水标准。

The Middle Route of South-to-North Water Diversion Project has improved the safety of water supply in North China, especially in large cities such as Beijing. While recently, due to the decrease of the inflow of Danjiangkou Reservoir and the proposal of the ecological environment protection strategy in the Hanjiang River Basin, Danjiangkou Reservoir are facing a great challenge of water quantity, and the water diversion project from Three Gorges Reservoir to Hanjiang River has been put on the agenda. As Danjiangkou Reservoir is the water source of the Middle Route of South-to-North Water Diversion, the water quality of this reservoir is also of great significance. In order to scientifically evaluate the impact of the upper dam program of the project of water diversion project from Three Gorges Reservoir to Hanjiang River on the water quality of Danjiangkou Reservoir, this paper constructs a hydrodynamic-water quality coupling model based on MIKE21 and a water quality prediction model based on BP neural network model for Danjiangkou Reservoir, which forecasts the water quality of Danjiangkou Reservoir after water diversion.After collecting the daily meteorological data of Danjiangkou Reservoir from 2017 to 2019, the daily discharge of Hanjiang River and other main tributaries, the DEM of Danjiangkou Reservoir’s underwater topography, the specific planning schemes of water division project, and the monthly data of main water quality monitoring sections, a hydrodynamic-water quality coupling model was constructed based on MIKE21. The model was calibrated and verified by the measured water quality data, including dissolved oxygen (DO), ammonia nitrogen (NH3-N), permanganate index (CODMn) and total phosphorus (TP) of Danjiangkou Reservoir. Based on the correlation analysis of simulated results by MIKE21 model, a back propagation(BP) artificial neural network model was constructed to predict the water quality with higher computational efficiency, and the model effect was evaluated by comparing the simulated results with the measured value. The results show that both MIKE21 model and BP artificial neural network model can simulate the water quality changes of Danjiangkou Reservoir well, and the calculated Nash efficiency coefficients of each water quality index range from 0.60 to 0.84 and 0.65 to 0.87, respectively. BP neural network has higher computational efficiency, while MIKE21 model can reflect the spatial distribution of concentration field in the reservoir more comprehensively.Finally, based on MIKE21 and BP artificial neural network models, the hydrodynamic and water quality changes of Danjiangkou Reservoir before and after water diversion were predicted. The prediction results show that the water diversion project would have little influence on the water quality of Danjiangkou Reservoir, and the spatial distribution of ammonia nitrogen and permanganate index remains basically unchanged. It might be that the total phosphorus concentration increase in different periods of the year from estuary of Duhe River to Danjiangkou Dam (i.e. Hanku), and the average annual total phosphorus concentration in about 8% of the water area would drop to Class III standard. The water quality near Taocha would basically unaffected. The results of MIKE21 model show that under the most unfavorable working conditions, the ammonia nitrogen and permanganate index levels of Taocha would fluctuate slightly due to the influence of the water diversion project, but the water quality classification could remain the Class I to II standards. The concentration of total phosphorus might increase, and it is predicted that the water quality would be degraded for up to 7 days, but it could still meet the Class III standard of surface water quality. The results of BP neural network model also show that ammonia nitrogen would still meet the Class I standard, while the permanganate index might undergo water quality degradation for up to 8 days, and total phosphorus might undergo water quality degradation for up to 16 days, but the water quality at Taocha there could also meet the water quality standard of Class III for surface water.