基于贝叶斯模型的地下水风险源污染概率估计方法研究

我国地下水环境风险源点多面广，但风险源周边地下水监测水平较低，尤其是在单个监测点指标异常时，监测数据异常值的来源及风险源造成污染概率的判定方面存在较大不足.为了解决此类问题，提出了基于贝叶斯模型的地下水风险源污染概率估计方法，并以石家庄市某工业集聚区下游一个农灌井中Cr6+含量和CHCl₃含量异常事件为研究案例，计算了指标异常来源于工业集聚区内8个风险源的污染概率.结果表明:①通过结合风险源的建成时间、废水排放量等软数据及对流弥散方程，优化先验概率、似然度以及后验概率求解方法，提出了基于贝叶斯模型的地下水风险源污染概率估计方法.②该工业集聚区下游农灌井中Cr6+含量和CHCl₃含量异常事件的案例应用结果显示，Cr6+含量异常来源于S6风险源的后验概率为76.2%，即Cr6+含量异常最有可能由某无机盐制造业污染源造成；CHCl₃含量异常来源于S1和S3风险源的后验概率分别为32.7%和23.6%，监测点CHCl₃含量异常最有可能由一个或两个化学农药制造业污染源造成.研究显示，建立的地下水风险源污染概率估计方法初步解决了监测数据不足时指标异常的来源识别问题，可用于未开展详细调查前地下水污染来源的快速锁定，也可使后期的地下水污染调查更具有针对性，对地下水污染风险防控具有重要科学意义. 

Contamination Probability of Groundwater Risk Sources by Bayesian

Groundwater risk sources are widely distributed in China. However, the groundwater monitoring level around the risk sources was low. Especially when the index of a single monitoring point was abnormal, there were major deficiencies in the determination the source of the abnormal monitoring data and the probability of pollution caused by the risk source. In order to solve this problem, a method for identifying the probability of groundwater risk source based on the Bayesian formula was proposed. Taking the abnormal events of Cr⁶⁺ and CHCl₃ in an agricultural irrigation well downstream of an industrial agglomeration area in Shijiazhuang as the research object, the probability of contamination for 8 risk sources was calculated. The results showed that: (1) Through the combination of the groundwater risk source soft data and the convection dispersion equation, the posterior probability of the abnormal observation point caused by the risk source was obtained, and the identification of contamination sources with insufficient observation data in multiple groundwater risk sources was solved. (2) Based on inversion calculations of the probability of 8 different industry categories of groundwater risk sources which could be caused Cr⁶⁺ and CHCl₃ observation anomalous. The probability of the S6 was 76.2%, indicating that the anomaly of observation data was mostly caused by the salt manufacturing industry. For the abnormal values of CHCl₃, the probability of the S1 and S3 points was 32.7% and 23.6% respectively, which was mostly caused by the chemical pesticide industry. The research shows that this method can solve the problem of source identification of indicator outliers when the observation data is insufficient. It can be used to quickly identify the sources of groundwater contamination before conducting a detailed investigation. It can also make the following groundwater contamination investigation more targeted. This method has important scientific significance for the prevention and control of groundwater pollution.