Multivariate statistical and GIS-based approach to identify heavy metal sources in soils

Abstract

The knowledge of the regional variability, the background values and the anthropic vs. natural origin for potentially harmful elements in soils is of critical importance to assess human impact and to fix guide values and quality standards. The present study was undertaken as a preliminary survey on soil contamination on a regional scale in Piemonte (NW Italy). The aims of the study were: (1) to determine average regional concentrations of some heavy metals (Cr, Co, Ni, Cu, Zn, Pb); (2) to find out their large-scale variability; (3) to define their natural or artificial origin; and (4) to identify possible non-point sources of contamination. Multivariate statistic approaches (Principal Component Analysis and Cluster Analysis) were adopted for data treatment, allowing the identification of three main factors controlling the heavy metal variability in cultivated soils. Geostatistics were used to construct regional distribution maps, to be compared with the geographical, geologic and land use regional database using GIS software. This approach, evidencing spatial relationships, proved very useful to the confirmation and refinement of geochemical interpretations of the statistical output. Cr, Co and Ni were associated with and controlled by parent rocks, whereas Cu together with Zn, and Pb alone were controlled by anthropic activities. The study indicates that background values and realistic mandatory guidelines are impossible to fix without an extensive data collection and without a correct geochemical interpretation of the data.