Multivariate geostatistics based on a model of geo-electrical properties for copper grade estimation: a case study in Seridune, Iran

In projects involving reserves estimation, a principal aim is to reduce the variance of the estimation and related uncertainty. This requires extensive and costly drilling. Among the variety of geostatistical-based techniques used to reduce the variance of the estimation in mineral reserves modelling, multivariate geostatistical methods can be appropriate tools when a sparse pattern of drilling boreholes exist. The present work introduces collocated cokriging and kriging with an external drift as multivariate geostatistical methods to incorporate sulphide factor as a secondary correlated variable to estimate Cu grade distribution. The study area is one of the potential zones of porphyry Cu occurrences, located in the Kerman province of southern Iran. To estimate the Cu grade distribution in this region, sulphide factor data as a dense correlated geophysical variable with the primary variable was used because this incurs less cost than drilling holes. Application of these multivariate geostatistical techniques to a specific exploration such as Seridune Copper Deposit in interpolating Cu grade measurements (primary data) using weakly correlated sulphide factor (secondary data) suggests that when Cu grade is undersampled, the secondary data can contribute substantially to identifying primary data. The results show that incorporating a secondary variable leads to better results than ordinary kriging (as a univariate method) that does not incorporate sulphide factor data. The validation of leave-out samples was used to compare the performance of the methods. Based on mean absolute error, root mean square error and the correlation coefficient of the observed and estimated values, the methods of the collocated cokriging and the kriging with an external drift outperformed grade estimation in comparison with ordinary kriging.