Statistical Evaluation of Geochemical Au Sample Quality

Abstract

Exploration and mining activities are based on anomalous occurrence of minerals. The basic concept governing this high risk venture is ‘no ore, no mining’. Hence, the quality of anomalous sample or its accurate determination is of great concern to exploration and mining operations. Regarding erratic mineralisation such as gold (Au), anomalous pattern of no significance whatsoever may appear in geochemical sample data as a result of poor sampling, improper sample handling or error in analytical techniques among other causes. To prevent the frequency of these occurrences, quality control checks coupled with classical statistical probe can form an integral part of the checklist to eliminate these errors. Although duplicate results have often accompanied original Au assays in most analytical reports submitted by laboratories, it is not immediately known upon what b asis the results need to be accepted or rejected. Often, some geologists accept results upon quick sight comparison. A total of three hundred and ninety (390) geochemical soil samples from the Sefwi-Bibiani belt of Ghana together with some blanks and standards were subjected to statistical analysis after following rigorous quality control sampling protocols. The statistical models employed include outlier test, distribution and correlation analysis. The original and duplicate samples were then statistically compared using simple nested One –Way Analysis of Variance (ANOVA), the Chi Square Test and the Student’s t –Test.
The ANOVA and the t–Tests revealed no significant analytical error. However, the other tests indicated multimodality of the populations as well as batch effect which culminates into significant procedural error. The investigation concludes that these systematic procedural errors if unchecked could mask true geochemical distribution.