Assessment of groundwater safety surrounding contaminated water storage sites using multivariate statistical analysis and Heckman selection model: a case study of Kazakhstan

Petrochemical enterprises in Kazakhstan discharge polluted wastewater into special recipients. Contaminants infiltrate through the soil into the groundwater, which potentially affects public health and environment safety. This paper presents the evaluation of a 7-year monitoring program from one of the factories and includes nineteen variables from nine wells during 2013-2019. Several multivariate statistical techniques were used to analyse the data: Pearson's correlation matrix, principal component analysis and cluster analysis. The analysis made it possible to specify the contribution of each contaminant to the overall pollution and to identify the most polluted sites. The results also show that concentrations of pollutants in groundwater exceeded both the World Health Organization and Kazakhstani standards for drinking water. For example, average exceedance for total petroleum hydrocarbons was 4 times, for total dissolved solids-5 times, for chlorides-9 times, for sodium-6 times, and total hardness was more than 6 times. It is concluded that host geology and effluents from the petrochemical industrial cluster influence the groundwater quality. Heckman two-step regression analysis was applied to assess the bias of completed analysis for each pollutant, especially to determine a contribution of toxic pollutants into total contamination. The study confirms a high loading of anthropogenic contamination to groundwater from the petrochemical industry coupled with natural geochemical processes.