Comparison of Two Multivariate Grade Simulation Approaches on an Iron Oxide Copper-Gold Deposit

Abstract

Multivariate geostatistics takes advantage of the autocorrelation observed between variables and is a valuable tool for mining applications. This study presents a comparison of a traditional multivariate simulation technique based on a model of coregionalization against another simulation in which decorrelation is first achieved by ``projection pursuit,{''} where independent variables are simulated, then correlated back by the inverse of the decorrelation transform. The deposit used for the comparison is an iron oxide copper-gold (IOCG) ore body located in the Atacama Desert in the north of Chile. The project value is sensitive to grade, and simulation would help in assessing the grade uncertainty. The correlations observed among the different economic variables provide an opportunity to improve the simulation results. This paper starts with a brief geological description of the ore body including some of the significant geological risks. A short summary of different multivariate simulation techniques is given. The process and results of multivariate simulation of total copper, acid-soluble copper, gold, and iron within the deposit using the two methodologies are presented. The focus is on the validation and comparison of processes and results to confirm whether projection pursuit can be used as an alternative for comprehensive uncertainty analysis for resource classification.