Department of Chemical and Environmental Sciences, Microbiology Laboratory, Centre for Environmental Research, University of Limerick, Limerick, Ireland E-mail:
J Water Health. 2014 Jun;12(2):310-7. doi: 10.2166/wh.2014.178.
Determining the likelihood that groundwater contains faecal coliforms can aid water resource management in facilitating the protection of drinking water supplies. This study assesses the incidence of the faecal indicator organism Escherichia coli in 125 private water supplies (PWSs) serving individual houses in the Mid-West Region of Ireland. Two factors, aquifer type and rainfall (mm), were chosen as independent variables that can affect the vulnerability of a groundwater body. Using a geographical information system, the relative hydrogeological and climatological features unique to each sampling location were derived. Utilising this information, a logistic regression (LR) model was used to predict the probability of contamination of PWSs with E. coli. The model contained two independent variables: rainfall (mm; p < 0.001) and aquifer characteristics (p = 0.001). The full model, containing both predictors, was statistically significant at p < 0.001, indicating that the model distinguished between the independent variables' relationship to the incidence of contamination. The likelihood of E. coli contamination is greater with increased rainfall and in areas where a bedrock aquifer is dominant. The LR model explained between 27.4% (Cox and Snell R squared) and 36.8% (Nagelkerke R squared) of the variance in contamination and correctly classified 75.2% of cases.
确定地下水是否含有粪便大肠菌群，可以帮助水资源管理部门保护饮用水供应。本研究评估了 125 个为爱尔兰中西部地区个别房屋服务的私人供水系统（PWS）中粪便指示生物大肠杆菌的发生率。选择含水层类型和降雨量（mm）作为两个独立变量，它们可以影响地下水体的脆弱性。利用地理信息系统，可以推导出每个采样点特有的相对水文地质和气候特征。利用这些信息，使用逻辑回归（LR）模型来预测 PWS 受到 E. coli 污染的概率。该模型包含两个独立变量：降雨量（mm；p<0.001）和含水层特征（p=0.001）。包含两个预测因子的完整模型在 p<0.001 时具有统计学意义，表明该模型区分了自变量与污染发生率之间的关系。降雨量增加和基岩含水层占主导地位的地区，E. coli 污染的可能性更大。LR 模型解释了污染方差的 27.4%（Cox 和 Snell R 平方）和 36.8%（Nagelkerke R 平方），正确分类了 75.2%的病例。
