Geochemical Data Mining by Integrated Multivariate Component Data Analysis: The Heilongjiang Duobaoshan Area (China) Case Study

Abstract

The Heilongjiang Duobaoshan area is located at the confluence of the Great Xing’an Range and the Lesser Xing’an Range, and the area has undergone a complex magmatic and tectonic evolutionary history resulting in a complex and diverse geological background for mineralization. As a result of this geological complexity and the multi-period nature of mineralization, the geochemical data of the area are usually not satisfied with a single statistical distribution form, so traditional statistical methods cannot adequately explore and identify the distribution of deep-seated information in the geochemical data. Based on the above problems, this paper adopts a multivariate component data analysis method to process 14 mass fraction data elements, namely Ag, As, Au, Bi, Cu, Fe, Hg, Mn, Mo, Ni, Pb, Sb, W, and Zn, in the 1:50,000 soil geochemical data from the Duobaoshan area of Heilongjiang. The spatial distribution and internal structural characteristics of raw, logarithmic transformation and isometric logarithmic ratio (ILR) transformed data were compared using exploratory data analysis (EDA); robust principal component analysis (RPCA) was applied to obtain the PC1 and PC2 principal component combinations associated with mineralization, and a spectrum–area (S–A) fractal model was further used to decompose the geochemical anomalies of the PC1 and PC2 principal component combinations as composite anomalies. The results show the following: (i) The data transformed by the isometric logarithmic ratio (ILR) eliminate the influence of the original data closure effect, and the spatial scale of the data is more uniform; the data are approximately normally distributed, based on which RPCA can be applied to better explore the correlation between elements and the pattern of co-associated combinations. (ii) The S–A method was further used to decompose the composite anomalies of the PC1 and PC2 principal component combination in the study area. The anomalous and background fields of the screened-out PC1 and PC2 principal component combinations reflect anomalous information on mineralization dominated by Au mineralization. Moreover, the anomaly and background information after extraction were in good agreement with the known Au deposits (points), and many geochemical anomalies with prospecting potential were obtained in the periphery, providing a theoretical basis and exploration focus for the next step in the searching and exploring of the study area.