Ore reserve estimation at the Enterprise Gold Mine, Pine Creek, Northern Territory, Australia. Part 2: the multigaussian kriging model

"The estimation of recoverable ore reserves for feasibility in precious metal deposits such as Pine Creek must account not only f or the extreme variability of the ore grade but also for the complexity of the geology. Compatible models f or both must be adopted. Further, very selective mining demands reserve estimates be based on small selection units whose individual grades cannot be estimated reliably from the exploration drilling data.Multigaussian kriging is one method based on a multivariate normal model for estimating recoverable reserves under such conditions. In this paper, a non-mathematical description of the multigaussian model is presented together with its application to the Pine Creek Gold Deposit. The results, when compared to the first year of production, are encouraging for the model but not without surprises.Together, Part 1 and Part 2 of this study combine a detailed and geologically conditioned variogram analysis with an appropriate estimation method for a highly variable low-grade gold deposit. This and other studies of gold deposits familiar to the author show convincingly both that while the appropriate geostatistical skills and techniques are necessary to tackle the question of reserves, equally important but often discounted is the need for an appropriate and compatible geological model. IntroductionThe Pine Creek gold mineralization is associated with saddle reefs, and quartz and quartz- sulphide veining within the Enterprise Anti cline. The anticlinal structure is complex and, prior to mining, the mineralization was considered to be mainly strata controlled. Grades vary up to 160 grams in l .5rn core samples with coefficients of variation in excess of three.The estimation of the recoverable reserves for such a depo sit must account for both the geological complexity and the extreme variability of the gold grade.Part 1 of this work presented the geological model for the deposit and summary statistics 0 f the gold data. A detailed structural and variogram analysis considered both gold grades and the normal score (or gaussian transform) of these gold grades for different composite sizes."