Role of the EM clustering method in determining the geochemical background of As and Cr in soils: a case study in the north of Changchun, China.

Abstract

Determining the geochemical background for heavy metals is vital in soil management activities. Although many statistical methods for geochemical background determination have been proposed, the multi-population problem of geochemical data, primarily regional ones, derived mainly from mixing multiple populations belonging to various geological sources or processes, needs to be better addressed. In this study, the Expectation-Maximization (EM) algorithm was employed to separate multiple populations in a 1:250,000 scale regional geochemical data set of soils in a lithologically complex region in the north of Changchun, China. The data set included 3746 surface soil samples analyzed for SiO2, K2O, Al2O3, CaO, La, Rb, Y, Ti, Ce, V, Cr, and As. The potential high-risk areas of As and Cr were determined before and after the separation of multiple populations. The comparison results show that the EM clustering method can efficiently separate multiple populations and determine soil geochemical background more reasonably, thus eliminating false contamination that is easily misidentified and better revealing concealed contamination that is challenging to detect.