Abstract

Ore/waste classification and economic evaluations of mineral deposits rely on the grade of elements of interest, which must be predicted as accurately as possible to minimise misclassifications. Ordinary kriging is commonly used for such a purpose, but non-linear predictors such as disjunctive kriging may improve the results. In this context, this work presents two case studies, in one of which (gold grades with heavy-tailed distribution) disjunctive kriging outperforms ordinary kriging, while in the other case study (copper grades with a moderately skewed distribution), it turns out to be as accurate as ordinary kriging, although with less conditional bias.