The upstream of Nakdong River is contaminated by heavy metals such as Cd, Cu, Zn, As, and Pb. Although the origin of the contamination is unequivocal, it is suspected that the heavy metals have been leached from several mine tailings and a refinery. Here, receptor models, absolute principal component score (APCS) and positive matrix factorization (PMF), were used to identify the contamination sources. Source markers representing each source (factor) were investigated using correlation analysis for five major contaminants (Cd, Zn, As, Pb, and Cu) and identified as following: Cd and Zn for the refinery (factor 1), As for mine tailings (factor 2). The categorization of sources into two factors was statistically validated via the cumulative proportion and APCS-based KMO test score with the values >90 % and > 0.7 (p < 0.001), respectively. High R2 values of linear regressions between the predicted data from receptor models and observed data indicate the reliability of the model prediction; moreover, the predicted initial concentrations of contaminants were validated using a sediment sample collected from near the refinery (chi-test: p > 0.200). Concentration distribution and source contribution using GIS revealed the heavy metal contaminated zones affected by the precipitation.
Keywords: Absolute principal component score; Geographical analysis; Multivariable linear regression; Positive matrix factorization; Source apportionment.
Copyright © 2023 Elsevier B.V. All rights reserved.