随着流域内社会经济的高速发展，人类对水资源开发利用程度不断提高，对流域水环境大范围、持续和重度的干扰，导致水环境质量日趋恶化，水资源短缺和供需矛盾极其紧张，水资源过度开发和利用；生态用水量严重不足，生态系统完整性遭到破坏，致使流域水环境承受巨大的风险压力。流域水环境是个复杂的系统，流域水环境风险也涉及多种因素并且关系复杂。小尺度、单一污染物、仅对突发事件或仅针对于水质的风险评价已无法满足流域尺度水环境可持续发展的需要，急需建立基于大尺度特别是流域尺度的风险评价模型和方法，为流域水环境安全提供防范和保障机制。本文以海河流域和其子流域滦河流域为案例开展了不同尺度流域水环境风险评价模型及其应用研究。主要研究内容和结论包括：（1）提出了耦合水质-水量-水生态的流域水环境风险评价的概念。基于流域水环境系统中水资源短缺、水质恶化和生态系统退化交互影响的复杂性，提出流域水环境风险是指在流域尺度上，综合考虑水质、水量和水生态三个方面，描述和评估环境污染、人为活动或自然灾害等引起的水环境变化对流域生态系统及其组分产生不利作用的可能性和大小的过程。（2）建立了流域水环境风险评价程序：问题形成与描述、受体和评价终点的选取、风险源分析、风险小区划分、概念模型与分析计划、暴露-危害分析、风险表征、不确定性分析和风险管理等9个部分。其中，概念模型、暴露-危害分析和风险表征是流域水环境风险评价的关键步骤。概念模型用于揭示风险源、胁迫因子、生境及评价终点之间的关系；暴露-危害分析主要针对流域水环境水质、水量和水生态三个方面展开，定量化风险评价组分之间的暴露-响应系数；风险表征即评估危害作用的大小以及发生概率的过程，是风险评价的核心部分。针对不同的流域尺度，分别建立综合指标评价模型和相对风险模型（RRM）进行风险表征，并应用蒙特-卡罗模拟法研究评价过程中的不确定性。（3）不同尺度流域水环境风险评价模型的建立① 基于现有水资源分区和管理监测方式，建立了适用于流域尺度的综合指标评价模型，从人类活动对水环境水质、水量和水生态三个方面的影响构建了流域水环境风险评价概念模型和理论指标体系，通过Pearson相关分析和主成份分析进行敏感风险因子的筛选和归类，建立了海河流域多层次的评价指标体系，采用熵值法确定权重，并利用模糊优选模型进行风险表征。② 基于流域生态水文分区方式，建立了适用于子流域尺度的相对风险模型（RRM），为了降低评价中的不确定性，考虑了风险源产生胁迫因子可能性的大小，提出了“相对强度”系数，并改进了RRM模型中不同风险小区风险源和生境相对密集度的确定方法，将原有的分段赋值法与改进的比值法相结合，使原来离散的密集度等级值变为连续的分值，实现了区域内不同评价小区间风险程度的比较。在暴露-危害分析中，为减少不确定性，增加了“终点-生境”的暴露系数。在终点的选取中同时考虑了生态终点和人体健康终点，尝试通过RRM模型的加和进行生态风险和健康风险的整合。RRM模型的评价结果表征了不同评价小区间相对风险关系，适用于基于风险源、胁迫因子、生境和评价终点之间暴露-响应关系的实验数据和资料相对贫乏的流域尺度的风险评价。（4）对海河流域水环境问题和驱动因素进行全面的诊断和分析，为风险源和胁迫因子的确定提供依据。海河水环境退化的表征可归纳为水质恶化、水体富营养化、生境的破碎化、湿地萎缩、河口退化、生物多样性减少、生态系统功能的降低和丧失等。海河流域水环境风险主要受人类强活动干扰的驱动，地表水利用程度最高达94%，地下水超采率最高可达133%，水利工程的大规模建设，破坏了自然河道的连通性，水库拦截造成下游来水减少和河道断流等一系列生态环境问题，是水环境风险的主要人类活动驱动因素；海河流域水环境的水污染范围广（受污染河长超过70%），程度重（劣五类河流超过60%），是造成水质风险的主要因素。将海河流域水环境风险源归纳为工业风险源（采矿业和污染型工业）、农业风险源（种植业、畜牧业和水产养殖业）、生活风险源（生活污水和城镇化）、水利工程风险源。（5）不同流域尺度模型应用研究① 基于现行的管理方式和监测数据，应用提出的适用于流域尺度的综合指标评价模型，对海河流域四大子流域进行了实证研究，建立了海河流域水环境风险的耦合水质-水量-水生态的多层次评价指标体系。结果表明：四大子流域的水环境风险形势由大到小的排序是：海河南系（0.8737）、海河北系（0.5032）、徒骇马颊河水系（0.3877）、滦冀沿海诸河（0.2662）。各子流域的决策优属度最大的均为水质风险，滦冀沿海诸河和海河南系其次是水量风险，海河北系和徒骇马颊河水系其次是水生态风险。多层次评价结果得出，各水系的水资源条件均是导致风险的主控因素；滦冀沿海诸河、海河南系和徒骇马颊河水系风险的主控因素还包括污染控制和水量控制；海河南系和徒骇马颊河水系污染风险最为严重，其中海河南系主要是面源污染，徒骇马颊河水系主要是点源污染；海河北系和海河南系的水资源开发利用状况和土壤地质（水土流失、地下水超采）也是导致风险的的主要因素；徒骇马颊河水系的生态治理最为薄弱。说明该模型可以基于现有水资源分区方式对流域水环境风险空间分异情况总体评价并明确不同层面风险的主控因素。② 应用建立的适合子流域尺度的相对风险评价模型进行滦河流域水环境风险评价时选择水质恶化、生态用水量不足、生物多样性减少、生态服务价值受损和景观破碎作为评价的生态终点，人体健康作为健康评价的终点，构建了滦河流域水环境风险源-胁迫因子-生境-终点的暴露和响应路径的概念模型，针对不同胁迫因子、不同风险源类型、不同生境、不同终点、不同风险小区计算出潜在的风险值。实证研究表明：滦河上游风险较低，潘家口和大黑汀水库所在小区属于中等风险，中游和下游平原城市密集区风险最高。其中畜牧业、生活污水和污染型工业为主要风险源；生境破坏、耗氧有机污染物和营养物质是生态压力的主要影响因子；评价结果通过了蒙特-卡罗模拟法的不确定性检验，证实了该模型对基于流域生态水文分区，并考虑风险源、胁迫因子、生境和评价终点之间的复杂关系的流域水环境风险评价有很强的适用性。依据RRM模型的假设，尝试对滦河流域生态风险和健康风险进行整合。本研究认为滦河流域的生态风险和健康风险具有同等重要的地位，加和系数设定为1。结果表明，滦河流域各风险小区整合后的风险等级基本保持不变，青龙河支流风险小区整合后风险等级降低，说明生态风险和健康风险的重要性程度在不同风险小区会有所不同，不同风险小区的加和系数也需要根据上下游风险转移进行调整。

With the fast development of basin social economy, the conditions of water environment directly influence the normal production and life of more than one billion people, also relate to the sustainable, healthy and rapid development of social and economic of all provinces and cities in it. For a long time, the rapid development of social economic has led to the shortage of water resources and exacerbated the contradiction between water supply and water demand. The water quality has been extremely deteriorated and the water resources have been utmostly exploited and utilized. Long-term neglect of the eco-water demand has caused serious shortage of ecological water and the integrated of eco-system has been damaged which led the basin water environment bearing tremendous risk pressure. At present, the studies have indicated that the risk assessment of small scale, single pollutant, just on emergencies or only on water quality have been not satisfied the demand of sustainable development of basin water environment. The risk assessment models and methods based on large scales (regions or basins) are urgently needed to build which can provide prevention and guarantee mechanism.This paper comprehensively considered the interactions of many kinds of sources, stressors, habitats and endpoints based on the characteristics of basin water environment. In the study, we take Haihe River Basin and sub-basin Luanhe river basin as examples to carry out basin water environment risk assessment models and application research. The main research content includes:(1) The concept of basin water environment risk assessment which coupling the water quality, water quantity and water ecology was proposed based on the summarizing the present research on environmental risk assessment theories and methods of domestic and foreign scholars, also based on the research progress of water environment risk assessment. The study emphasized the importance of carrying out the water environment risk assessment in basin scales.(2) The procedure of basin water environment risk assessment was developed which included nine steps: problem formulation and description, receptors and endpoints selected, risk sources analysis, risk regions division, conceptual model and analysis plan, exposure-hazard analysis, risk characterization, uncertainty analysis and risk management. The risk sources of Haihe River Bain was concluded to industrial sources (mining industry and contaminated industry), agricultural sources (farming, animal husbandry and aquaculture industry), domestic sources (domestic sewage and urbanization) and hydraulic engineering sources. The division of risk regions was based on the evaluation methods and the actual conditions of study area. The conceptual model was used to reveal the exposure relations between sources and receptors. The exposure and hazard analysis mainly focused on the water quality, water quantity and water ecology. Risk characterization was to evaluate the degree and probability of the hazard effects which was the core of the risk assessment. At present, there are two models mainly used in risk characterization which are comprehensive index assessment model and relative risk model. In the phase of uncertainty analysis, the Monte-Carlo simulation method was adopted in the assessment procedure. And proper measures were proposed in risk management phase to minimize the risk of study area. (3) Through overall analysis and diagnosis of water environment conditions in Haihe River Basin, the definite characterization of water environment degradation included water quality deterioration, water eutrophication, habitat fragmentation, shrinking of wetland, biodiversity reduced and ecosystem service function loss. The driving force of water environment risk in Haihe River Basin mainly included two aspects: natural and anthropogenic factors. The little amount of natural water resources and the decreased trend year by year was the main limiting factor. Large scale construction of hydraulic engineering was the main anthropogenic activity deriving force which damaged the connectivity of natural rivers and caused a series of eco-environmental problems because of the reservoirs interception led to downstream coming water reduced and dry river. The wide rage and high degree of pollution was the main reason which caused the water quality risk.(4) Based on the current means of administration water resources, the comprehensive index assessment model which was suitable for basin scale was proposed. The conceptual model and theoretical index system was developed from the affects on water quality, water quantity and water ecology by anthropogenic activities. The fuzzy optimization model was adopted in risk characterization. This paper used the model taking the four sub-basins in Haihe River Basin as case study, and built the multiple-level index system which coupled water quality-water quantity-water ecology. The results indicated that the order of the water environment risk situation of the four sub-basins from high to low was the Haihe south river system (0.8737), the Haihe north river system (0.5032), Tuhaimajiahe river system (0.3877) and Luanjiyanhai river system (0.2662). From the three aspects water quality risk, water quantity risk and water ecology risk, the largest decisive superiority of all sub-region was water quality risk, the secondly was water quantity risk in Luanjiyanhai and Haihe south river system, water ecology in Haihe north river system and Tuhaimajiahe river system. The three aspects were all largest in Haihe south river system and other sub-regions were relatively lower. The results also showed that the main controlling factors were pollution control, water resources condition and water amount control in Luanjiyanhai, Haihe south river system and Tuhaimajiahe. The pollution condition was the most serious in Haihe north river system and Tuhaimajiahe, which was non-point pollution in Haihe south river system and point pollution in Tuhaimajiahe. The development and utilization of water resources and soil geology (including soil erosion and groundwater overexploitation) were also main controlling factors in Haihe south and north river system. The ecological control of Tuhaimajiahe was the most weak and the water resources condition was limited factor in all sub-regions.(5) The Relative Risk Model (RRM) was improved and introduced to the basin water environment risk assessment. To decrease the uncertainty in assessment, the amount of stressors which sources produced was considered and proposed the “relative intensity” coefficient. We also improved the method of determining the relative dispersion in different risk regions sources and habitats. The former piecewise valuation method was combined with the improved ratio method which made the discrete dispersion become the continuous scores. The combined method realized the comparison of risk degree in different risk regions and decreased the uncertainty. In exposure-hazard analysis phase, we added the “endpoint-habitat” exposure coefficient to decrease the uncertainty. In endpoint selection, the ecological endpoints and human health endpoints were both considered and preliminary attempted to integrate the ecological risk and health risk through RRM model.(6) Case study of RRM model application:Water environment risk assessment of Luanhe River basin:The water quality deterioration, eco-water demand shortage, biodiversity reduced, ecological service value damaged and landscape fragmentation were selected as the ecological risk assessment endpoints and human health as the health risk assessment endpoint. Based on the analysis of sources, receptors, habitats and endpoints, the conceptual model was built to reflect the exposure and response routes. Then, the RRM model was built to calculate the different stressors, different sources, different habitats, different endpoints and different regions potential risk values. The results of the case study showed that there were three low risk regions, one medium risk region and two high risk regions. The results indicated that habitat destruction was the largest stressor for the Luanhe water environment, the second was oxygen consumption organic pollutants and the third was nutrients. The three stressors were the main influencing factors of the ecological pressure in the study area. Animal husbandry was the biggest source and next was domestic sewage and polluted industries. For habitats, waters and farmlands were bearing the bigger pressure and should take considerable attention. Water deterioration and ecological service value damaged were faced the biggest risk pressure, and secondly was biodiversity and landscape fragmentation. Although the RRM model was characterized the relative risk relations, the rating of different kinds of sources and habitats in evaluated regions could be realized by grading system. The assessment results not only passed the uncertainty test by Monte-Carlo simulation, but also agreed well with the field investigation and experiment data.Integrated ecological risk and health risk:According to the assumption of the RRM model, the ecological risk and health risk level could integrate through addition principle. In the study, we considered the ecological risk and health risk had the same important position, so the addition coefficient was determined 1. The results indicated that the risk level was not changed after integrated except for Qinglonghe river risk region. The condition illustrated that the importance of ecological risk and health risk might be not equal in different risk regions, and the risk transfer from the upper to the lower river should be considered.