干旱区生态环境知识资源中心

Geochemical data used in this paper to identify geochemical signatures associated with Fe mineralization in eastern Tianshan mineral district, China are compositional data. In order to eliminate the spurious relationships between compositions, isometric logratio (ilr) transformation is currently employed to deal with the closure effects. The opened geochemical data are further analyzed by principal component analysis. By back-transforming to the clr space, PCA results of ilr transformed data with geological meanings of their counterparts assist in recognition of spatial distributions of both intermediate-felsic igneous rocks and fault systems in the study area. In comparison with log-transformed geochemical data, the ilr transformed ones are theoretically reasonable to apply standard statistics. In addition, multivariate outliers are identified to investigate their associations with the formation of these two geological features. At the end, this paper suggests that the ilr transformation is necessary to be applied routinely before applying statistical treatments to raw geochemical data. (C) 2013 Elsevier B.V. All rights reserved.