Reserve estimation in multivariate mineral deposits using geostatistics and GIS

Abstract

In this study, drilled core values from the Magnesite Incorporated Company—its original Turkish name being Manyezit Anonim Sirketi (MAS)—Beylikova Magnesite Open-Pit Mine in Eskisehir-Turkey provided multivariate ore grade and reserve estimates that were used to integrate geostatistical estimation methods with Geographic Information Systems (GIS) technology. GIS technology is known for with its visual and spatial query capabilities in three-dimensional (3D) environments. Using the company’s topographical maps, a digital terrain model of the mine area of interest was generated in a GIS environment. Bore hole locations, drilled core values and cutting depths were also input to the sophisticated spatial geodatabase. Ore impurity grade values were analyzed according to their lognormal distributions, then evaluated by applying their corresponding variogram and cross-variogram models — models used in ordinary cokriging. To estimate ore reserve and grade, ordinary cokriging techniques were used: a twovariable model with two different variable combinations and a three-variable model. Spatial queries were applied to estimation results in a 3D GIS platform in order to determine the location, shape, and quantity of the ore body. Subsequently, the Mean Standardized Square Error (MSSE) statistical procedure was applied to the estimation results to assess and compare their accuracy. Based on these assessments, it was determined that ordinary cokriging with three variables was the most appropriate and accurate approach for estimating ore grade distribution and reserve in Beylikova Open-Pit Mine.