Automated update of geological domains using blast-hole assays and minimum covariance determinant

Abstract

Mining operations demand accurate 3D geological models for resource estimation and production. The industry-wide accepted method for constructing such domains is to use exploration-based wireframes, limiting the accuracy of the domains by the sparsity of the exploration holes. Update of the domains using blast-hole data is highly desirable to increase the accuracy of the models and enable better decision making for downstream mining processes. We discuss some of the practical and theoretical problems of using existing boundaries created using exploration drill-hole data. We also propose using the blast-hole data to adjust ore to waste boundaries. The proposed methodology is evaluated on a dataset from an iron ore deposit located in the Brockman Iron Formation of the Hamersley Province, Western Australia, demonstrating important changes in multiple boundaries in this real-world mining dataset.