USE OF CLUSTER ANALYSIS TO IDENTIFY WEATHERING PATTERNS IN ROCKS OF VARIOUS COMPOSITIONS

Abstract

The authors cluster analysis to rank the ten most important oxides occurring in rocks and minerals (SiO2, TiO2, Al2O3, Fe2O3, FeO, CaO, MgO, Na2O, K2O, and H2O+) in order (from 1 to 10). The first is the oxide with the lowest concentration by weight and the last is that with the highest. The rankings were than used to determine the characteristics LM, KN and OK: the first is the discriminator and the other two are classifiers. While the ranges of the igneous rocks show significant similarities in their chemistry and minerals, the ranges of the most important sedimentary rocks (LM/KN and LM/OK diagrams) indicate their closely correlation mainly with the origin of the rocks. The ranges of the original rocks and the end products of their weathering show little differentiation on the LM/KN diagrams, suggesting some inheritance of their overall chemical composition. General comparison of the chemical compositions of original rocks of different composition and of the end products of their weathering can also be represented on LM/KN and LM/OK diagrams as weathering vectors. These vectors become particularly informative when they are shown on the diagrams together with the ranges of the original rocks (LM/KN diagram) and the end products of weathering (LM/OK diagram). An analysis of the study results is discussed.