基于复合模型的河流沉积物重金属污染源解析及不确定性研究

Due to their environmental persistence and biogeochemical recycling and ecological risks, heavy metal pollution become a major problem related to aquatic environmental safe. Thus, according to the contamination features of heavy metals, using quantitative methods to accurately identify natural and anthropogenic sources and apportion their contributions is one of the challenging and hot issues in the fields of pollution prevention and control of heavy metals in the aquatic environment. Receptor model is a general approach for source apportionment. However, due to the openness of water environment system, as well as the complexitiess of the chemical behavior of heavy metals, the adaptabilities, accuracies and uncertainties of receptor model for source apportionment of heavy metals in aquatic environment are worthy of to further study. In the cross-view point of environmental science, geochemistry, hydrogeology, statistics probability and artificial intelligence, the present research will identify the emission and transport mechanism of heavy metals in river sediments, and establish a methodological framework based on factor analysis with nonnegative constraint and chemical mass balance combined support vector machines for source apportionment of heavy metals. The critical problems of this study will concentrated on the research of logic mechanisms and uncertainties analysis of the combined model. The method will also be applied for source apportionment of heavy metals in the Le'An River within the Poyang Lake Basin. These results will provide policy and decision makers with a useful help for supporting the pollution control, ecological repairmen, pollutants mitigation and environmental forensics of water system. Meanwhile, this study will provide a useful direction for control of heavy metals to support the management of water pollution for Poyang Lake.

面对河流重金属污染的严峻形势，开展污染源解析研究，准确识别重金属的自然和人为来源及其贡献程度，是当前流域重金属污染防控领域具有挑战的热点问题之一。受体模型是一类较为通用的源解析技术，然而，鉴于河流环境系统的开放性，以及重金属化学行为的复杂性，单一受体模型应用于河流重金属源解析的适应性、可靠性及不确定性值得深入研究。本研究拟从环境科学、地球化学、水力学、概率统计学以及人工智能的交叉视角，在辨识河流沉积物重金属污染的源排放特征及迁移转化的基础上，开展基于"非负约束因子分析-化学质量平衡/支持向量机"复合模型的河流沉积物重金属污染源解析研究，系统阐明复合模型源解析的逻辑机理，深入揭示影响复合模型解析不确定性的关键因素，并以鄱阳湖流域乐安河沉积物重金属源解析为例进行研究验证。项目成果将为河流重金属污染控制、生态修复、总量减排、污染事故调查提供科技支撑，也能为乐安河重金属污染防控提供科学依据。

针对河流沉积物中重金属的污染问题，开展污染源解析研究，识别重金属的污染来源及其贡献程度，是当前流域重金属污染防控领域具有挑战的热点问题之一。本课题提出开展基于复合受体模型的河流沉积物重金属源解析研究，其主要内容包括：阐明复合模型的源解析机理，分析复合模型源解析不确定性，以鄱阳湖流域乐安河为例，辨识河流沉积物中重金属的污染特征，识别影响重金属分布的主要来源，解析其贡献比例。通过三年的资助研究，我们完成了上述研究内容的所有工作，达到了预期的研究目标。通过研究，验证了利用受体模型复合应用开展污染源解析的适宜性，建立了基于FANNC复合模型的河流沉积物重金属污染源解析技术框架，识别了影响复合模型源解析不确定性的主要因素，提出了对应的不确定性分析方法，研发了自动化软体程序。这些研究成果对提升污染源解析可靠性具有一定的参考价值，能为环境污染溯源理论体系的完善提供适当的基础研究，也将为流域的水源保护、污染防治、生态修复、事故调查与损害赔偿提供科技支持。而利用复合模型开展的应用研究结果表明，乐安河流域沉积物中重金属污染较为严重，具有较高的生态风险，主要污染因子包括Au, Cd, Cu, Mo, Hg, Ag和Zn，影响这些重金属分布的主要污染源来自铜尾矿（~40%）、采矿废水（~30%）以及农业活动（~18%）。为保护乐安河及鄱阳湖流域的生态安全，亟需进一步规范区域的矿产开采活动，对于污染较重的德兴段，需要考虑开展河流修复工程以清除淤积在沉积物中的高含量重金属污染物。