随着经济快速发展和人口数量不断增加，水污染问题越来越凸显，严重危害人类的生命健康、制约社会的可持续发展。防治水污染已成为我国乃至世界各地政府高度关注的课题。目前许多地区通过加强对重点企业的监督、提高污水处理设施的覆盖率和效率，“因地制宜”地不断完善水污染管控措施。然而，各区域作为一个开放的经济体系，不仅地理边界内各个产业之间交织复杂，与外部也存在大量的商品和服务交换活动，同时伴随着内嵌在商品和服务中的隐含环境影响在产业系统之间和贸易区域间的转移，只关注自身污染物排放的管理政策仍存在一定局限性。许多研究表明隐含在商品和服务中的水污染物总量呈逐年递增趋势，分析隐含水污染转移，探究经济系统和水污染之间的关系，为各区域协同防治水污染提供了一个新的视角。灰水足迹可用于隐含水污染分析，量化隐含污染转移导致的环境影响，并把所有污染物折算成统一单位。此外，灰水足迹作为一种可持续性指标，相比较根据水体中污染物浓度来评价水质的传统方法，能从水量的角度评价区域的水污染状况，体现了水质和水量的结合。 灰水足迹是以自然本底值和一定的环境水质标准为基准，将污染物负荷同化吸收所需的淡水量。但是，目前关于灰水足迹的理论体系还处于发展阶段，直接应用到隐含水污染转移分析或者水污染评价中仍存在一些问题：一方面，灰水足迹计算方法的精度还有待提高，传统方法未考虑到水体自净能力，且灰水足迹值由最关键污染物单一因素决定，不能准确量化多种污染物存在情况下的综合效应，也不能描述最关键污染物负荷一定而其他污染物排放量改变的情况下灰水足迹的对比变化；另一方面，以往的灰水足迹研究多从生产视角出发，将研究主体处于“封闭”经济中，注重单一区域或者单一产品生产过程中直接灰水足迹的简单核算，而对消费驱动的间接灰水足迹缺乏定量研究，较少涉及经济结构和灰水足迹之间的联动关系，难以为之后的污染物控制提供决策支持。 基于上述认识，本研究致力于全新构建综合考虑多种污染物的灰水足迹核算模型（Multi-Pollutants Gray Water Footprint Model, Multi-P-GWF Model），应用到城市、区域、国家等不同空间尺度，基于环境投入产出分析，同时从生产和消费视角，精准追踪水污染在城市内部产业系统之间以及区域贸易往来中的隐含流。最终目的是通过开展隐含水污染转移格局与规律的研究，识别城市产业系统水污染控制的关键节点、摸清水环境污染的区域联系，为探讨污染物减排、制定环境补偿政策提供科学依据。此外，水体上下游联系紧密，以流域为研究单元更符合其自然属性，本研究还进一步将灰水足迹扩展到流域尺度，根据流域水功能区划首次提出设定分段水质目标，更精细化地核算了流域灰水足迹。本研究主要取得了以下成果： （1）在方法学上，Multi-P-GWF Model模型实现了灰水足迹计算从单因子确定性模型到多因子模糊性模型的突破，提高了灰水足迹核算的精度和科学性。针对传统灰水足迹计算方法的不足，本研究基于质量平衡方程和模糊综合评价方法构建了改进的灰水足迹核算模型，不仅考虑水体的自净能力，还将多种污染物纳入到核算体系，更客观、合理地量化了污染物排放对水环境造成的影响。以北京市2016年为例，选取化学需氧量（COD）、氨氮（NH3-N）、总磷（TP）、总氮（TN）、石油类（Petroleum Contaminants）、挥发酚（Volatile phenol）为主要水质指标对改进的灰水足迹核算模型进行实例研究，发现当水质目标分别为V类，IV类，III类, II类和I类时，基于改进模型的灰水足迹阈值下限对比传统计算方法分别下降了17.85%、31.29%、34.03%、41.49%、54.96%。传统计算方法中，最关键污染物为TN，因此，灰水足迹的值完全由TN决定，与其他污染物的排放量无关。而在改进的灰水足迹核算模型中，虽然TN的权重系数最大（0.5874），其他污染物指标也对灰水足迹取值范围的确定做出了相应贡献。因此，改进后的灰水足迹核算模型对非最关键污染物的负荷变化更加敏感。本研究为更科学地计算灰水足迹提供了方法学支撑。 （2）本论文将改进的灰水足迹核算模型应用到城市尺度，引入环境投入产出分析，重点发展了城市产业系统灰水足迹量化方法。该方法通过分析生产视角各生产部门直接灰水足迹的去向以及消费视角最终产品灰水足迹的来源，描述了隐含水污染在城市内部产业系统间的转移路径，有助于定量划分各生产部门的减排责任，为城市制定污染物削减政策提供科学依据。以北京市2007年为例对该方法进行验证，结果表明从生产视角来看，第一产业、第二产业和第三产业直接灰水足迹分别占产业系统灰水足迹总量的84.21%、7.14%、8.65%。而从消费视角来看，第一产业、第二产业和第三产业最终产品灰水足迹分别占产业系统灰水足迹总量的35.67%、31.45%、32.87%。由此可见，隐含水污染的总体流动趋势是从第二和第三产业流向第一产业。具体到各生产部门，农业部门的直接灰水足迹是4204.59×10⁶m³，而最终产品灰水足迹下降到1842.05×10⁶m³。农业部门的直接灰水足迹仅有42.33%（1780.0×10⁶m³）用于自身最终产品，分别有23.50%（988.0×10⁶m³）和18.08%（783×10⁶ m³）用于生产中间产品提供给食品及烟草和其他服务业部门作为原材料，食品及烟草和其他服务业部门属于隐性高污染行业。因此，从整个经济系统减排的角度来看，各部门不仅应当减少自身直接灰水足迹的产生，还应节约使用其他部门提供的中间产品，控制污染物的间接排放。 （3）本论文将改进的灰水足迹核算模型应用到国家尺度，基于多区域投入产出分析方法对2007年全国整体框架下30个省市的灰水足迹进行了比较，并分析了隐含水污染在中国区域间的流动情况。中国2007年的灰水足迹总量合计15354.54×10⁸ m³，在空间分布上，各省市的灰水足迹总量差异较大，均方差系数高达74.41%，山东省灰水足迹总量最高（1593.60×10⁸ m³），而青海省灰水足迹总量最低（60.79×10⁸ m³），前者约为后者的26倍。净调入虚拟灰水量排名前五的省份分别是上海（196.10×10⁸ m³）、北京（102.39×10⁸ m³）、天津（57.98×10⁸ m³）、广东（47.70×10⁸ m³）和山西（33.59×10⁸ m³）；净调出虚拟灰水量排名前五的省份分别是河南（127.24×10⁸ m³）、河北（86.18×10⁸ m³）、山东（63.56×10⁸ m³）、黑龙江（53.24×10⁸ m³）和湖南（45.64×10⁸ m³），从整体来看，隐含水污染从中部沿海和南部沿海等发达地区通过省际贸易转嫁到华北、东北等农业占比高，重工业密集的非发达地区。这个结果表明中国现有的贸易结构一定程度上还存在不合理性，东北和华北作为缺水地区却是虚拟灰水的净调出区域，承受的水污染已超过其环境容量。而拥有丰富水资源、水污染指数较低的南部沿海地区却从其他地区调入了大量虚拟灰水，将自身消费引起的水污染转嫁给上游生产地。 （4）本论文将改进的灰水足迹核算模型扩展到流域尺度，以北运河流域（北京段）为例，结合Johnes输出系数模型估算非点源污染物的来源和数量，并根据水功能区划对北运河各子流域进行分段水质目标设定，从水量的角度评价了北运河流域的水污染程度。北运河流域2015年的灰水足迹总量为100.29×10⁸ m³，而相应的水资源总量仅为8.80×10⁸ m³，因此北运河流域整体的水污染指数高达11.53。这个结果表明北运河流域的污染物排放已经远超过该流域的水环境容量，即使本研究中多数河段水功能区划均为V类，仍不能达到设定的水质目标要求。从各个子流域来看，水污染指数最高的分别是小中河流域（54.20）、蔺沟河流域（23.00）和凤河流域（14.43）。小中河流域和凤河流域的耕地面积占比较高，农业生产造成的非点源污染是不可忽视的因素。蔺沟河流域虽然单位面积灰水足迹强度相对较低，但由于其水资源量较少，水环境容量极其有限，对污染物的排放更加敏感，容易造成水质降级。以上三个子流域都是北运河污染物总量控制需要重点关注的区域。

With economic development and population increase, the problem of water pollution becomes prominent, which seriously endangers human health and restricts the sustainable development of society. How to prevent and control water pollution is a topic worthy of attention for the governments in China and even around the world. At present, most of regions implement water pollution control measures by strengthening the supervision of key enterprises and improving the coverage and efficiency of sewage treatment facilities. However, these management policies still have some limitations due to only focusing on their own pollutant emissions. As an open economy, each region not just has a complex industrial system within its geographical boundaries, but has trade activities with other regions. At the same time, environmental impacts embodied in traded goods and services are transferred through economic activities. Some studies shown that the total amount of water pollutants embodied in goods and services are growing fast. By analyzing the transfer of embodied water pollution and exploring the relationship between economic system and water pollution, a new perspective is provided for the coordinated prevention and control of water pollution in various regions. Based on natural background concentration and reference water quality standards, gray water footprint (GWF) is defined as the amount of fresh water required to assimilate pollutant load. The advantages of GWF applied to virtual water pollution analysis are that it converts all embodied pollutants into a homogeneous unit: freshwater consumption. Furthermore, it reflects the economic activities impacts on water quality, by quantifying the fresh water required to assimilate the emissions. But the current GWF theory system is still in the initial stage, which unsuitable applied directly to analyze embodied water pollution transfer or evaluate water pollution level: firstly, the GWF calculation method needs to be improved, the conventional method without considering the water self-purification ability and the GWF value is mainly influenced by concentration value of the highest concentrated pollutant. Thus, the co-presence of other critical compounds is almost disregarded; Secondly, previous GWF studies were mostly carried out from the perspective of production, but lack the perspective of consumption. That makes it difficult to provide decision support for the subsequent pollution control. Based on the above understandings, this study is committed to establish an improved GWF calculation framework, which takes water self-purification ability and the effects of multiple pollutants into consideration. Then, the improved model is applied at city and national scale. From the perspective of production and consumption, combined with the environmental input-output framework, the flow of water pollution embodied in industrial systems and inter-regional trade can be clearly figured out. The ultimate goal is to identify the key nodes of water pollution control in industrial system and find out the connections of water environmental pollution between regions, so as provide scientific basis for formulating environmental compensation policies. In addition, it is more appropriate to conduct research at river basin scale due to the upstream and downstream water bodies are closely related. Based on the export coefficient model, GWF at a river basin scale is calculated in a more refined way. The main achievements of this study are as follows: (1) In terms of methodology, the calculation method of GWF has changed from single factor deterministic model to multi-factor fuzzy model. In order to solve the limitations existing in the conventional GWF calculation method, based on a mass-balance model and fuzzy synthetic evaluation, an improved GWF calculation model (Multi-P-GWF Model) is established and applied in Beijing 2016. Taking chemical oxygen demand (COD), ammonia nitrogen (NH3-N), total phosphorus (TP), total nitrogen (TN), petroleum, petroleum contaminants, volatile phenol as the main water quality indexes, it is found that when the water quality targets are set as class V, class IV, class III, class II, class I, the threshold lower limits of GWF decreased by 17.85%, 31.29%, 34.03%, 41.49%, 54.96% respectively, if compared with the values calculated by conventional approach. Applying the conventional method, the most critical pollutant detected would be TN. Consequently, the integrated GWF value is completely determined by TN, so there's no connection between GWF and the non-critical pollutants. In the case of Multi-P-GWF model, although the weight of TN is the largest (0.5874), also the other indexes contribute to the determination of GWF range, meaning that this method is sensitive to the presence of multiple pollutants. The improved model taken in this study enhances the scientific soundness of GWF assessments and can describe the existing pressure on surface water. (2) The Multi-P-GWF model is applied at city scale. Based on the environmental input-output framework, the destination of direct GWF and the source of final product’s GWF can be figured out. By describing the transfer path of embodied water pollution in the industrial system, it is helpful to investigate the responsibilities of sectors for emission reduction. In Beijing in 2007, from the perspective of production, the direct GWFs of primary industry, secondary industry and tertiary industry respectively account for 84.21%, 7.14%, 8.65% of the total. From the perspective of consumption, primary industry, secondary industry and tertiary industry respectively account for 35.67%, 31.45%, 32.87%. The results show that the overall trend of embodied water pollution is from secondary and tertiary industries to primary industry. At sectoral level, the direct GWF of agriculture is 4204.59 ×10⁶m³. This includes 42.33% of the total caused by final products. 23.50% are due to intermediate products provided for food and tobacco sector as raw material. 11.94% are caused by intermediate products provided for other service sectors as raw materials. The results of GWF matrix show the virtual water pollution transfer among economic sectors via material exchange. The sectors which receive intermediate products from others means that the virtual water pollution transferred from them to up-stream sectors. These down-stream sectors should share responsibility for both direct and indirect wastewater discharge. Traditional management of water pollution mostly focus on the control of direct wastewater discharged by industries and improving the wastewater treatment technology. Nonetheless, it is often to not quantify the transfer in the indirect impacts brought about by the virtual water pollution transfer. The sectors of food and tobacco, other services receive a large quantity of indirect GWF from agriculture, thus, saving raw material and promoting the raw material efficiency are indirect effective measures to alleviate local water pollution. (3) The total GWF in China in 2007 is 15354.54×10⁸ m³. In terms of spatial distribution, GWF at provincial scale varied greatly, with a mean square coefficient of 74.41%. Shandong province ranks first（1593.60×10⁸ m³）, while Qinghai province is the lowest（60.79×10⁸ m³）. Shanghai（196.10×10⁸ m³）, Beijing（102.39×10⁸ m³）, Tianjin（57.98×10⁸ m³）, Guangdong（47.70×10⁸ m³） and Shanxi（33.59×10⁸ m³）are the top five net importers in terms of virtual gray water . The top five net exporters of virtual gray water are Henan（127.24×10⁸ m³）, Hebei（86.18×10⁸ m³）, Shandong（63.56×10⁸ m³）, Heilongjiang（53.24×10⁸ m³） and Hunan（45.64×10⁸ m³）. On the whole, embodied water pollution is transferred from developed areas such as the central and southern coastal areas to non-developed areas such as the north and northeast areas. The results show that the existing trade pattern of China is unreasonable to some degree: the northeast and north as the water-deficient areas, are the net exporters of virtual gray water. However, the southern coastal with abundant water resources import a large amount of virtual gray water from other regions, transferring the water pollution caused by consumption to the upstream production sites. (4) Studying GWF at river basin scale could better reflects the impacts of water pollution on the available quantity of water resources, but the lack of pollutant emission data is a barrier to GWF assessment. The export coefficient model is introduced to estimate non-point source pollution. This model is implemented in the Beiyun basin as a test. In 2015, the total GWF of the Beiyun basin is 100.29×10⁸ m³， while the corresponding total water resource is only 8.80×10⁸ m³. Therefore, the water pollution level of the Beiyun basin is 11.53. This result shows that the pollutant discharge in the Beiyun basin far exceed the water environmental capacity. Even though most of the functional zoning in this study is of class V, it still cannot meet the water quality target requirements. In terms of sub-basins, the water pollution index of Xiaozhong sub-basin is the highest (54.20), followed by Lingou (23.00) and Fenghe sub-basin (14.43). For Xiaozhong and Fenghe sub-basin, the non-point source pollution caused by agricultural production is a non-negligible factor. Despite the GWF intensity is relatively lower, Lingou sub-basin is more sensitive to pollutant discharge due to its limited water environmental capacity, which is likely to cause water quality degradation. Therefore, the three sub-basins mentioned above need to be focused on and preferentially treated.