Health risk analysis can predict and control the risks posed by heavy metals, especially in drinking water, which is a highly sensitive environmental receptors. In order to evaluate heavy metal pollution in drinking water, the monthly average concentrations of As, Cd, Cu, Hg, Ni, and Zn were used to assess the health risk between January 2015 and December 2018 in a drinking water source. Furthermore, Spearman rank correlation coefficient and the ARIMA model were used to analyze temporal variations. The results showed that the monthly average concentrations of heavy metals exceeded the class Ⅲ values as specified by Chinese environmental quality standard for surface water(GB 3838-2002), especially Hg with a minimum monthly average four times more than that set by the standard limits. Overall, the order of carcinogenic risk of As and Cd was decreased; the non-carcinogenic risk of Zn, Cu, Ni, Pb, and Hg was increased. Further, the comprehensive non-carcinogenic risk for adults was lower than 1 throughout the study period except February 2015, when the comprehensive non-carcinogenic risk for children was lower than or close to 1 after October 2017, and the comprehensive carcinogenic risk for children was more than 10-4. Meanwhile, the children's health risks are higher than that for adults, with the main health risk characteristic factors of As, Cd, and Hg. The Spearman rank correlation coefficient were -0.714069, -0.773122, and -0.62234, indicating the significant downward trend from 2015 to 2018. However, the children's comprehensive carcinogenic risk, whose average value was 0.000234 much more than 10-4, had significant upward trend in 2018 with Spearman rank correlation coefficient 0.902098. The ARIMA(3,1,3) model was able to predict the comprehensive carcinogenic risk for children from heavy metals in drinking water, and the result indicated the children comprehensive carcinogenic risk should be monitored to ensure levels between 0.000200 and 0.000302. The study has positive significance for risk warning and environmental management compared to the analysis and prediction of health risk from heavy metals in drinking water sources based on time series models.
Keywords: ARIMA model; Spearman rank correlation coefficient; drinking water source; health risk; heavy metal.