A new grade-capping approach based on coarse duplicate data correlation

Abstract

SYNOPSIS In most exploration or mining grade data-sets, the presence of outliers or extreme values represents a significant challenge to mineral resource estimators. The most common practice is to cap the extreme values at a predefined level. A new capping approach is presented that uses QA/QC coarse duplicate data correlation to predict the real data coefficient of variation (i.e., error-free CV). The cap grade is determined such that the capped data has a CV equal to the predicted CV. The robustness of the approach with regard to original core assay length decisions, departure from lognormality, and capping before or after compositing is assessed using simulated data-sets. Real case studies of gold and nickel deposits are used to compare the proposed approach to the methods most widely used in industry. The new approach is simple and objective. It provides a cap grade that is determined automatically, based on predicted CV, and takes into account the quality of the assay procedure as determined by coarse duplicates correlation. Keywords: geostatistics, outliers, capping, duplicates, QA/QC, lognormal distribution.