Academic Journals Database
Disseminating quality controlled scientific knowledge

Improved concentration data in two international geochemical reference materials (USGS basalt BIR-1 and GSJ peridotite JP-1) by outlier rejection

Author(s): Surendra P. Verma

Journal: GeofĂ­sica Internacional
ISSN 0016-7169

Volume: 37;
Issue: 3;
Start page: 215;
Date: 1998;
VIEW PDF   PDF DOWNLOAD PDF   Download PDF Original page

Keywords: Outlier statistical tests | geochemistry | normal distribution | international geochemical reference materials

Concentration data for major and trace elements were examined in two international geochemical reference materials, basalt BIR-1 from the United States Geological Survey (USGS) and peridotite JP-1 from the Geological Survey of Japan (GSJ). After outlier rejection (based on fourteen different statistical tests with thirty-five variants), concentration means and other statistical parameters were computed in these two difficult mafic materials, which have low concentrations of most trace elements, including the rare-earth elements. Two distinct methods were used: (A) All available concentration data for a given element were treated as a statistical sample of a single population; (B) Data were grouped according to the analytical method used and tested for bias among method groups using ANOVA hypothesis test, at a confidence level of 99%.For USGS basalt BIR-1 using Option (A), it was possible to derive high-quality data for four major elements (Ti, Fet, Ca, and FeO) and four trace elements (Cu, Eu, Sc, and Sr), characterized as cve and intermediate-quality data for five major and eighteen trace elements as rv. Similarly in GSJ peridotite JP-1, three major (SiO2, Fe2O3t, and CaO) and three trace elements (Co, Ni, and Sc) are at present of cve quality, and three major and nine trace elements are rv. The application of Option (B) resulted in cve quality for two elements (Ti and Cu) and rv for eleven elements in BIR-1. For JP-1 two major elements (SiO2 and CaO) are cve quality and three trace elements (As, Yb, and Zn) rv. The remaining elements compiled here are still only provisional (pv) values.A comparison of these results with literature values obtained by the original compilers using a different statistical method shows that the new statistical procedure involving a set of different tests for normal univariate data provides a better perspective of application to such geochemical databases.
Save time & money - Smart Internet Solutions      Why do you need a reservation system?