®®®® SIIA Público

Título del libro: Advances In Geosciences: Volume 26: Solid Earth (se)
Título del capítulo: Statistically correct methodology for compositional data in new discriminant function tectonomagmatic diagrams and application to ophiolite origin

Autores UNAM:
SURENDRA PAL VERMA JAISWAL; PANDARINATH KAILASA;
Autores externos:

Idioma:
Inglés
Año de publicación:
2011
Resumen:

In the lead presentation (invited talk) of Session SE05 (Frontiers in Geochemistry with Reference to Lithospheric Evolution and Metallogeny) of AOGS2010, we have highlighted the requirement of correct statistical treatment of geochemicaldata. In most diagrams used for interpreting compositional data, the basic statistical assumption of open space for all variables is violated. Among these graphic tools, discrimination diagrams have been in use for nearly 40 years to decipher tectonic setting. The newer set of five tectonomagmatic discrimination diagrams published in 2006 (based on major-elements) and two sets made available in 2008 and 2011 (both based on immobile elements) fulfill all statistical requirements for correct handling of compositional data, including the multivariate nature of compositional variables, representative sampling, and probability-based tectonic field boundaries. Additionally in the most recent proposal of 2011, samples having normally distributed, discordantoutlier free, log-ratio variables were used in linear discriminant analysis. In these three sets of five diagrams each, discrimination was successfully documented for four tectonic settings (island arc, continental rift, ocean-island, and midocean ridge). The discrimination diagrams have been extensively evaluated for their performance by different workers. We exemplify these two sets of new diagrams (one set based on major-elements and the other on immobile elements)using ophiolites from Boso Peninsula, Japan. This example is included forillustration purposes only and is not meant for testing of these newer diagrams.Their evaluation and comparison with older, conventional bivariate or ternarydiagrams have been reported in other papers. © 2011 by World Scientific Publishing Co. Pte. Ltd. All rights reserved.


Entidades citadas de la UNAM: