水温是河流最基础的生境因子之一。在气候变化和水库运行的复合影响下，全球河流水温正发生不同程度的改变，给生态环境造成了重要影响。厘清气候变化和水库运行对水温的影响，阐明其相互作用机制，已成为目前亟待解决的关键科学问题之一。本文以长江流域为研究对象，从现象分析、机理解释和量化归因三个层面对长江干流朱沱-大通段水温进行了系统化定量研究，取得的主要研究成果如下：基于实测数据的时域与频域综合分析发现，近三十年来长江干流年平均水温正以0.27-0.55 ℃/10a的速率变暖；水温的多尺度波动强度整体减弱，水温过程变得更加光滑和平坦；半年及以上尺度的水温分量显著滞后，而短期分量基本未发生滞后。水温和气温的多尺度相关性随着时间尺度的增大先增加后减小，在年尺度上达到峰值，水库运行导致水温和气温的多尺度相关性降低，但整体降幅不大。基于小波变换，本文将自回归模型与神经网络模型深度耦合，开发了一种新的水温模型，相较于多个常用神经网络模型，将水温的预测精度整体提升了约15%。建立了长江干流水热动力学模型和水龄模型，量化了干流水体的输移时间，并分析了其对水温过程和水气响应关系的影响机制。研究结果表明：（1）水体输移时间改变会对水温形式产生直接影响。三峡水库运行导致库区水体输移时间延长10天（洪季）至60天（枯季）左右，库区下泄水温因此发生显著的季节性滞后，即年最低温滞后约40天，年最高温滞后并不明显。（2）水体输移时间改变会影响水气响应关系。在库区，水体输移时间增加导致水温对气温的累积响应增加，出库水温对库区气温长期变化的响应变得更加敏感；而在坝下河段，春冬季水库补水缩短了水体输移时间，增加了水体热惰性，导致水温对气温变化的响应敏感性降低。通过建立多模型联合定量归因分析框架，对长江干流水温演变进行了量化归因，并分析了各影响因素之间的叠加效应。结果表明，长江干流水温的长期演变趋势主要受气候变化控制，在不同河段的影响占比为49%-86%。上游梯级水库群和三峡水库起次要作用，两者累积影响在不同河段的占比为9%-55%。水温年内过程的演变在上游主要受水库运行控制，而在中下游气候变化渐渐占据主导地位。气候变化和水库运行的影响在秋冬季相互叠加增强，而在春夏季相互抵消减缓。上游梯级水库群和三峡水库对水温的影响整体上相互叠加增强。但在3-4月，受流量增加和水体输移滞后的影响，梯级水库群在一定程度上缓解了三峡水库对水温的影响。

The water temperature is one of the most fundamental habitat factors for rivers. Under the combined effects of climate change and reservoir operations, global river temperatures are undergoing varying degrees of change, significantly impacting the ecological environment. Clarifying the influence of climate change and reservoir operations on water temperature and elucidating their interaction mechanisms have become critical scientific issues that urgently need to be addressed. This study takes the Yangtze River Basin as the research subject and conducts a systematic and quantitative study on the water temperature of its mainstream from Zhutuo station to Datong station from three aspects: phenomenon analysis, mechanism explanation, and quantitative attribution. The key findings of this research are as follows:Comprehensive analysis in both the time and frequency domains based on empirical data has revealed that over the past thirty years, the average annual water temperature of the Yangtze River‘s main channel has been warming at a rate of 0.27-0.55°C per decade. The intensity of multi-scale fluctuations in water temperature has generally weakened, resulting in a smoother and flatter temperature profile. The water temperature components on scales of half a year or longer show significant lag, while short-term components exhibit virtually no lag. The multi-scale correlation between water temperature and air temperature increases and then decreases with the enlargement of the time scale, reaching a peak at the annual scale. The operation of reservoirs leads to a reduction in the multi-scale correlation between water temperature and air temperature, but the overall reduction is not significant. Based on wavelet transform, this study deeply coupled autoregressive model with neural network models, developing a novel water temperature model. Compared to several commonly used neural network models, this new approach improves water temperature prediction accuracy by approximately 15%.A thermodynamic model and a water age model have been established, to quantify the water transport time and to analyze the mechanisms of its influence on the water temperature process and the water-air response relationship. The research results show that: (1) changes in water transport time have a direct impact on water temperature pattern. The operation of the Three Gorges Reservoir has extended the transit time of water in the reservoir area by about 10 days during the flood season to around 60 days during the dry season. As a result, the discharged water temperature from the reservoir exhibits a significant seasonal lag, with the annual minimum temperature lagging by approximately 40 days, while the lag in the annual maximum temperature is not significant. (2) The change in the transport time of water bodies affects the water-air response relationship. In the reservoir area, an increase in the transit time of water bodies leads to an increased cumulative response of water temperature to air temperature, making the discharge water temperature more sensitive to long-term changes of the air temperature in the reservoir area. Conversely, in the river section downstream the dam, the increased discharge caused by reservoir in spring and winter shortens the water transport time, and increases the thermal inertia of the water body, which results in a decreased sensitivity of water temperature to changes in air temperature.By establishing a multi-model joint quantitative attribution analysis framework, the evolution of the water temperature in the mainstream of the Yangtze River has been quantitatively attributed, and the superimposition effect between the influencing factors has been analyzed. The results show that the long-term evolution trend of the water temperature in the mainstream of the Yangtze River is mainly controlled by climate change, with the impact ratio being 49%-86% in different river sections. The cascade reservoirs and the Three Gorges Reservoir play a secondary role, and their cumulative impact accounts for 9%-55% in different river sections. The evolution of the annual process of water temperature is mainly controlled by the operation of the reservoir in the upstream, while climate change gradually takes the dominant position in the middle and lower reaches. The impacts of climate change and reservoir operation are mutually reinforcing in autumn and winter, and mutually offsetting in spring and summer. The impacts of the upstream cascade reservoirs and the Three Gorges Reservoir on water temperature are generally mutually reinforcing. However, in March and April, due to the increase in flow and the lag in water transport, the cascade reservoirs alleviates the impact of the Three Gorges Reservoir on water temperature to a certain extent.