[Application of robust estimator in source identification of soil pollutants: a case study at a heavy metal contaminated site].

Spatial outliers can give us useful information on locations of pollution sources. A Genton's estimator based variogram was applied to analyse the concentration of Cu, Zn, Pb, Cd, As, Se and Hg in soils at a contaminated site. Spatial outliers were identified to imply the location of the pollution sources of heavy metals. The results showed that the semivariograms of 7 elements were best described by spherical model, and the nugget variances were very small, especially for Cu, Zn and Pb, which were 0.000. Compared with Matheron's estimator, Genton's estimator can damp the effect of outliers, and so get a smaller nugget effect. The spatial outliers existed in a cluster pattern, whose locations coincided with those of copper smelters in the area. This indicated that the copper smelters were the important sources of soil heavy metals.