Geostatistical Simulation of Cross-Correlated Variables: a Case Study through Cerro Matoso Nickel-Laterite Deposit

Abstract

Geostatistical methods have been increasingly used as powerful techniques for predicting spatial attributes and modelling the uncertainty of predictions in un-sampled locations, especially through multi-element deposits. Independent Gaussian simulation constructs precise outputs over each variable, in most cases by simulating using the multi-Gaussian assumption. However, this approach does not consider the underlying correlations between the variables. Spatial uncertainty can also be quantified by co-simulation, where the relationship of the co-regionalized variables is accounted for and the spatial relationships between variables are reproduced. In this study, we apply the two aforementioned approaches (independent simulation and co-simulation) for modelling two correlated elements (Fe & MgO) at Cerro Matoso S.A. Nickel laterite deposit located in Colombia. Results show that co-simulation provides a reasonable outcome in regards to the correlation coefficient parameter and relative error as expected.