Identification of geochemical anomalies using principal component analysis and factorial kriging analysis

Abstract

Listen

Translate article

Abstract The purpose of this paper is to identify geochemical anomalies using the combined application of data analysis and geostatistical methods. Data from a local multi-element geochemical prospecting soil survey in NW Spain were used to illustrate this application. The principal component analysis was first used to establish the different associations among variables and afterwards to obtain a homogeneous group for applying the geostatistical methodology. The selected group was composed of Zn, Co, Ni, Cr, Cu, and Fe, which were strongly correlated with the first principal component (PC1). Taking advantage of this high correlation between the group of elements and the component 1, we used PC1 as a new variable for numerically identifying anomalies. The geostatistical approach to this problem involved computing variograms of PC1 and subsequently a decomposition of this variable using factorial kriging analysis. The results of the use of factorial kriging analysis demonstrate its value as a filter in geochemical prospecting when attempting to differentiate between anomalous samples and those belonging to geochemical background. The use of a variable that groups an association of elements, instead of identifying individual anomalies, is one of the advantages of applying principal component analysis.