基于贝叶斯网络拓扑结构的水环境风险溯源——以饮马河流域为例

为了解决流域水环境风险诊断过程中污染来源不清、污染贡献难以量化等问题,提出了一种基于贝叶斯网络拓扑结构的污染源追责量化方法.该方法首先通过互信息的计算实现流域水环境典型污染物的准确识别,在此基础上通过贝叶斯网络拓扑结构分析与启发式搜索算法快速辨析流域内典型污染来源及其污染贡献.本次选取吉林省饮马河流域2017~2020年水质监测数据进行分析.结果表明,氨氮为流域内的典型污染物;靠山南楼、靠山大桥、刘珍屯3个站点的污染来源分别为:杨家崴子、新立城大坝、砖瓦窑桥.其中靠山南楼有63%的污染来源于杨家崴子,靠山大桥有30%的污染来源于新立城大坝,刘珍屯有75%的污染来源于砖瓦窑桥.本次评估方法的构建可为流域水环境风险溯源及污染责任认定提供技术支撑.

Water environmental risk tracing based on the combination of Bayesian network topology:A case study of Yinma River Basin

In order to solve the problems of unclear pollution sources and difficult to quantify pollution contribution in the process of watershed water environment risk diagnosis, an accountability quantification method of pollution sources based on the combination of Bayesian network topology and heuristic search algorithm was proposed in this paper. The method can accurately identify typical pollutants in watershed water environment according to the quantitative evaluation of mutual information. In addition, Bayesian network topology analysis and heuristic search algorithm can quickly identify typical pollutant sources and their pollution contributions in the watershed. In this study, the monitoring data of Drinking Horse River Basin in Jilin Province from 2017 to 2020 were selected for the water quality analysis. Ammonia was a typical pollutant in the watershed; the three sections of Khao San Nan Lou, Khao San Bridge and Liu Zhen Tun were polluted by Yang Jia Weizi, Xin Li Cheng Dam and Zhuang Wa Yao Bridge respectively. 63% of the pollution in Khao San Lou came from Yangjia Weizi, 30% of the pollution in Khao San Qiao came from Xinlizheng Dam, and 75% of the pollution in Liu Zhen Tun came from Brick Wayao Bridge. This assessment method can be constructed to provide strong technical support for the tracing of water environment risk and pollution responsibility determination in the basin.