Application of mixed (bimodal) distribution to human health risk assessment of Cu and Ni in drinking water collected by RDT sampling method from a large water supply zone

作者：Hyk Wojciech^*; Swiecicka Dorota; Garbos Slawomir

来源：Microchemical Journal, 2013, 110: 465-472.

DOI：10.1016/j.microc.2013.06.002

• Metals
• Drinking water
• Random daytime sampling
• Bimodal distribution
• Deconvolution method
• Health risk assessment

The random daytime (RDT) sampling method was employed for the assessment of average exposure on Cu, Ni and Pb, connected with the consumption of drinking water in a large water supply zone. Among three data sets of Cu, Ni and Pb concentrations determined in ROT samples (n = 99) collected from the detailed study area situated in Warsaw (Poland), these related to Cu and Ni cannot be simply described by normal or log-normal distributions (in opposite to the results of Pb determinations). The general numerical method designed for the detection and calculation of mixed (bimodal) Gaussian distribution (PDF-MIX) that usually describes data taken from heterogeneous population is presented. The method is capable to extract two normal or log-normal components (PDF I and PDF II) from the mixed distribution of a given experimentally determined data set. Thus, their statistical characteristics (central values and standard deviations) can be established. Furthermore, the particular result in the obtained data set can be assigned to the individual distribution either PDF I or PDF II). The method was applied to the results of Cu and Ni determinations in the ROT samples. The constituents of the mixed distributions PDF(Cu) 1, PDF(Cu) H, PDF(Ni) 1 and PDF(Ni) II, allowed one to estimate the geometric means: 17.7 mu g/L, 362 mu g/L, 3.9 mu g/L and 12.1 mu g/L, respectively. On the basis of the above-mentioned data, the human health risk was assessed. The hazard quotient indices estimated with the use of central values of PDF(Cu) 1, PDF(Cu) II, PDF(Ni) land PDF(Ni) 11 log-normal distributions were: 0.013, 0.26, 0.0055 and 0.018, respectively (HQ%26lt; 1 no essential human health risk can be expected). Further statistical analysis of the data sets described by individual log-normal distributions allowed one to identify the presence of copper pipe sections in the domestic distribution systems (PDF(Cu) II) and characterize the groups of consumers who use drinking water frequently (PDF(Ni) I) or occasionally (PDF(Ni) II).

• 出版日期2013-9