目前，气候变化相关研究主要集中于流域气象和水文要素趋势分析以及气候因子对水文循环的影响研究，而对流域水环境污染尤其是农业非点源污染的影响研究较少，研究气候变化下流域非点源污染的变化趋势对流域水环境的管理、规划具有重要意义。相对于气候平均态，由极端降水事件引发的气象灾害对人类经济活动、社会生活和自然环境造成的影响更为明显。流域非点源污染随着降雨过程产生，降雨通常会导致土壤侵蚀的集中发生，并伴随着吸附态营养物的大量流失及土壤中的可溶态污染物大量淋溶，而小尺度流域上的快速径流过程会产生短期高峰值效应，对水环境的影响更大。在气候变化背景下，华北地区降水强度偏高的区域范围扩大，大暴雨事件频数增加，会加剧洪涝灾害发生的危险。本文在实际运用上以密云水库流域为例有针对性地开展非点源污染控制方面的相关研究，在分析历史基准期气候要素和流域水文要素变化规律的基础上，研究历史气候变化和未来气候变化趋势对流域非点源污染输出的影响，构建基于气候变化的污染控制措施评估及优化技术，提出兼顾极端降水的流域尺度水污染控制方案，为水环境保护和非点源污染综合防治提供必要的决策支持和科学参考，对流域未来非点源污染控制和管理、水源地保护、城市饮用水安全有着重要的理论和实际意义。 主要研究结论包括： （1）非点源污染的时间分布上表现出枯水期污染负荷强度最低，丰水期污染负荷强度较高的特点；污染强度在空间分布上表现出高海拔区域较低、低海拔区域较高、林业密集区较低、农业密集区较高、高径流区较高等特点。降雨特征、土地利用类型、地表植被覆盖度、土壤特性等因素直接影响土壤侵蚀和营养物流失。农田的土壤扰动程度最为严重、植被覆盖率较低、施肥活动较为频繁，属于污染高发区域；裸地的植被覆盖率较低，在降雨径流冲刷下易发生土壤侵蚀和营养物流失。高污染强度的土壤类型主要有淋溶土、粗骨土等，这些土壤作为农业用土，肥力高、径流潜力大，更易发生土壤侵蚀和营养物流失的现象。 （2）在非点源污染控制措施多目标优化系统中引进了多评估点、多主体、多水质保证率的优化方法。多评估点优化方法获得的流域非点源污染控制措施方案相较于单评估点方法可节约8%的成本，多评估点的优化方法在考虑流域内河道的拓扑关系的基础上，保证所有进入水库的污染物总量达标，能更经济有效地控制非点源污染。承担非点源污染控制的主体不同会导致控制措施优化结果的差异。以政府为主体的情景设置为：措施成本由政府支付，而措施实施过程中产生的经济效益归土地使用者所有；以土地使用者为主体的情景设置为：措施的成本由自己承担且措施实施产生的经济利益也归自己所有，在此情景中，非工程措施带来的经济利益要高于工程措施。在保证水质达标的前提下，优化算法趋于寻找成本较低的措施组合方案，因此，不同情景对措施的偏好差异较大。未来在指导流域综合管理的过程中需要权衡利益相关者的利益关系，推广更能被接受的控制技术，实现环境效益和经济效益最大化。 （3）采用HSPF模型对典型小流域降雨过程的非点源污染进行精细模拟，发现径流和污染物达到峰值的时间一致，相对于降雨峰值有一定的延迟。暴雨径流短期污染的峰值极高，与《地表水环境质量标准》（GB3838-2002）的相关标准对比，径流中污染物浓度峰值严重超标，TN、TP和粪大肠菌群超标比例分别是84.47%、15.0%和97.17%，为此，本文依据水功能区的水质标准建立控制暴雨径流短期高峰值污染的技术方案。暴雨径流过程产生的污染强度较大，水质控制目标较为严格，优化的流域污染控制措施方案中要求在所有子流域上配置两种以上的非工程措施和一种以上的工程措施。 （4）不同温室气体浓度情景下的气候特征对非点源污染的影响差异较大，严格减缓温室气体排放情景（RCP2.6）对流域未来水环境污染的发生起到一定的抑制作用。温室气体高浓度情景（RCP8.5）驱动的气候条件下降水量和降雨强度增大，进而导致非点源污染呈现加重的趋势。降水为影响未来非点源污染污染变化的主要因子，雨滴直接冲刷土壤表面及对土壤表层的淋溶作用是导致土壤侵蚀和营养物流失的主要途径。温度变化主要影响蒸散发过程，从而影响流域径流过程，对土壤侵蚀、营养物流失的影响较小。根据未来不同时期的非点源污染水平，运用全局多目标优化方法进行措施的空间配置，发现措施种类、数量和空间位置随着污染水平的变化而变化。 （5）温室气体高浓度情景RCP8.5的结果表明未来各降雨类型发生频率均有所增加，其中大雨、暴雨增加趋势较为明显，且其发生频率的年增长率波动较大。降雨量越大，污染程度越大，对下游水体的水环境健康和饮用水安全造成严重的威胁。降雨对非点源污染的影响不仅仅体现在降雨量的大小上，更有降雨前期干旱累积天数、降雨强度、降雨历时等因素的共同作用。降雨事件的前期干旱天数越长，降雨径流造成的水质超标率越高。主要通过调整措施的种类、数量和空间布局来控制不同降雨类型和不同干旱特征下的非点源污染，对前期干旱天数越长的（特大）暴雨过程的污染控制需要更多数量的工程措施和更大推广面的非工程措施。

Currently, a great deal of research focus on the impacts of climate change on the water quantity rather than water quality in a watershed. It is more important to predict the effect of future climate on the nonpoint source (NPS) pollution at watershed scale to ensure the drinking water security and to inform the valuable information to assess the efficiency of management practices and help to plan watershed management strategy. Comparing to the impacts of normal climate conditions, the extreme climate has more adverse effects on the economic development, environmental security and human health. The NPS pollution is driven by the rainfall-runoff, along with severe soil erosion and nutrient losses from soil layers. Under the background of climate change, the rainfall intensity will increase and the frequency of storm will increase, resulting in serious flood disasters and water pollution. This thesis selects the Miyun Reservoir Watershed as the study area to assess the impacts of climate change on the NPS pollution and to figure out a watershed management system that can select and place the BMP schemes in every subbasin under different climate condition. This study is useful to provide information for the protection of drinking water, the control of NPS pollution and the water resources management at watershed scale. The results of this study including: 1) There was spatial-temporal heterogeneity in the distribution of NPS pollution. Runoff and pollutant loads were lineally related to precipitation volume. Minimum values appeared in the dry periods and the dry years. The highly polluted areas were located in the areas where elevation is low and agricultural land accounts for a large area. 2) This study applied the SWAT model and NSGA-II to explore BMP solutions as influenced by different optimization assumptions such as government vs. farmer preferences, individual vs. coordinated optimization of the two upstream watersheds draining into the MRW, and optimization to meet year-round vs. flood season only water quality standards. The results showed substantial discrepancies between these two scenarios with respect to the types and placements of BMPs and associated economic costs, highlighting the need for reconciling concerns from different stakeholders in order to arrive at a BMP plan that all parties will agree upon. In addition, we found that cross-subwatershed coordination and targeting flood season instead of year-round water quality standards could pronouncedly reduce economic costs of BMP implementation while not substantially degrading water quality. 3) In this study, the HSPF model was used to simulate rainfall-runoff process that occur in a short period and for identify the impacts of rainfall on the quantity and quality of runoff, which accounts for sub-daily hydrological and water quality process and spatial and temporal variation of meteorological factors. The model revealed that pollution in the Tumenxigou catchment during a rainstorm event was serious. An average reduction of 82.47%, 15% and 97.17% of nitrogen, phosphorus and fecal coliforms loads, respectively, are necessary to comply with water quality standards. Accordingly, watershed management efforts for water quality restoration and protection should be considered during rainstorm events or summer months when the pollutant concentrations far exceed the allowable limits (water quality standards). Based on the assessment of BMP efficiency, a combination of land-use zoning practices, nutrient management strategies, soil management practices and structural management practices was optimized by the modified NSGA-II method to achieve the water environmental control targets. 4) The future climate data simulated by GCMs under the four greenhouse gas concentration trajectories of Representative Concentration Pathways (RCPs) were applied as the weather input in the calibrated SWAT model to predict the runoff and pollution loads in the future. This study predicted the impacts of precipitation volume, intensity, period, temperature and other climate factors on the NPS pollution loads in the future, and based on this result, the different combinations of watershed management practices are configured under different climate conditions. The pollution loads are lineally related to precipitation volume in future. The pollution level under the climate condition based on RCP8.5 (highest greenhouse gas concentration scenario) was higher than the lower greenhouse gas concentration scenarios (RCP2.6, RCP4.5 and RCP6.0) due to a larger increase in precipitation volume. The temperature has an important influence on the runoff in the Miyun Reservoir watershed through increasing evapotranspiration (ET), but has little effects on pollutant loads. Considering the different pollution levels under different climate conditions, the types, number and placement of BMPs should be configured in according to the different water quality improvement targets. 5) According to climate data based on RCP8.5, the precipitation in the future are higher than that in the reference period (1990-2014), and the frequencies of rainstorm and heavy rainstorm (extreme climate) increase significantly, which will result in great NPS pollution. In this regard, the dynamic management systems should be applied in relation to the varying pollution levels under different rainfall patterns. The discussion on the relationship of precipitation and pollutant loads revealed that the pollution loadings during rainfall-runoff were coordinately affected by precipitation volume, rainfall intensity, non-rainfall days before a rainfall event and the soil moisture capacity before rainfall. The different pollution levels and improvement targets of water quality under different rainfall patterns need dynamic management systems, such as more structural BMPs (filter strips and detention basins) and more implementation area of non-structural BMPs to control pollution in the process of (heavy) rainstorm runoff than that of the heavy rain.