Modem geostatistics provides advanced estimation techniques to determine the three-dimensional distribution of grade in a deposit more accurately than ever before. The working procedure comprises a series of steps to analyse the available geological information and process the sample assay data.
Preliminary processing of the sample assay data includes dividing, compositing, declustering, and/or transforming the data according to the local geology and sampling system. This creates a new data set, in which all of the data have the same support. The spatial continuity study of the data sets establishes the statistical model for each data set. Then a variography study reveals the correlation of the data in 3-D space. This correlation is the foundation for the application of advanced geostatistical estimation techniques. According to the statistical and variographic characteristics of the data sets, various estimate methods can be used to model the grade distribution over the sampled area. These methods include inverse distance power weighting (IDP), lognormal kriging (LK), indicator kriging with equiinterval cut-off thresholds (IK), and indicator kriging with truncated lognormal model (IKL).
Data provided by Mount Isa Mines Ltd for a geostatistical study included a set of diamond drilling core samples and a set of development chip samples from the 1100 copper ore body. Statistical analysis of the data indicated that copper mineralisation was multi-populational with lognormal characteristics. The grade distribution results from the applied estimators indicate that the IDP estimator is a poor technique in handling the outlier population of the data set. The other estimators (IK, IKL, LK) estimate the high grade blocks satisfactorily. Analysis of the estimated result for the Q398 Stope implies that lowering the cut-off grade from 2.0% to 1.85% copper will increase the copper content by 2099 to 4462 tonnes (3.0-6.2%) without a significant effect on the average grade.