Open AccessLog‐ratio transformed major element based multidimensional classification for altered H igh‐ M g igneous rocks
Author(s) -
Verma Surendra P.,
RiveraGómez M. Abdelaly,
DíazGonzález Lorena,
QuirozRuiz Alfredo
Publication year - 2016
Publication title -
geochemistry, geophysics, geosystems
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.928
H-Index - 136
ISSN - 1525-2027
DOI - 10.1002/2016gc006652
Subject(s) - igneous rock , linear discriminant analysis , geology , classification scheme , discriminant , geochemistry , algorithm , database , mineralogy , computer science , pattern recognition (psychology) , artificial intelligence , machine learning
A new multidimensional classification scheme consistent with the chemical classification of the International Union of Geological Sciences (IUGS) is proposed for the nomenclature of High‐Mg altered rocks. Our procedure is based on an extensive database of major element (SiO 2 , TiO 2 , Al 2 O 3 , Fe 2O 3 t, MnO, MgO, CaO, Na 2 O, K 2 O, and P 2 O 5 ) compositions of a total of 33,868 (920 High‐Mg and 32,948 “Common”) relatively fresh igneous rock samples. The database consisting of these multinormally distributed samples in terms of their isometric log‐ratios was used to propose a set of 11 discriminant functions and 6 diagrams to facilitate High‐Mg rock classification. The multinormality required by linear discriminant and canonical analysis was ascertained by a new computer program DOMuDaF. One multidimensional function can distinguish the High‐Mg and Common igneous rocks with high percent success values of about 86.4% and 98.9%, respectively. Similarly, from 10 discriminant functions the High‐Mg rocks can also be classified as one of the four rock types (komatiite, meimechite, picrite, and boninite), with high success values of about 88%–100%. Satisfactory functioning of this new classification scheme was confirmed by seven independent tests. Five further case studies involving application to highly altered rocks illustrate the usefulness of our proposal. A computer program HMgClaMSys was written to efficiently apply the proposed classification scheme, which will be available for online processing of igneous rock compositional data. Monte Carlo simulation modeling and mass‐balance computations confirmed the robustness of our classification with respect to analytical errors and postemplacement compositional changes.