Does Geostatistics Work? ? Introduction

The problem of the estimation of ore reserves is essentially a simple one to describe. Given a set of sample data, at specified locations within a deposit, how can we estimate the value (or quantity) of ore in unsampled areas. The question may be posed of a whole deposit (global) or it may be more desirable to estimate smaller volumes within the planned mine area, e.g. stopes or blocks (local). Many methods have been evolved to tackle this problem, from hand-contouring of values by the mine geologist to extremely sophisticated geomathematical analysis. The main difficulty with any estimation method in this context is that it is almost impossible to check whether the "predicted" value bears a close relationship with the "actual" value. The only reliable way to decide if an estimation method has "worked" is to test the eventual value of ore against the predicted value. This is, more often than not, impossible - even allowing for the time factor involved. To know the "actual" value of a mined block it has to be sampled intensely, and these values with their positions recorded accurately. Alternatively, blocks may be passed through the concentration process individually and these results recorded. In most producing mines neither of these alternatives are practicable or seen to be-important. Another problem with comparing in-situ (so-called) mineable reserves with production figures is the introduction of many variables into the concentration process -- dilution, wastage, possible stockpiling, mill recoveries and so on.