Abstract
Since its inception in 1910, the Bureau of Mines has been the principal federal agency involved in ore reserve estimations using such traditional methods as polygons of influence and variations of inverse distance weighting. While adequate for estimating the average ore grades in a mining block or section, these methods do not provide a consistent way of quantifying the reliability of the estimates. This report describes a technique for enhancing ore reserve estimations using geostatistics, the methodology of which combines the spatial and random aspects of geologic phenomena into a formal theoretical framework for deriving the estimation variance of ore reserves. Correctly used, geostatistics enables confidence intervals to be placed around ore reserve estimates calculated from drill hole data. A geostatistical software system was obained from the U.S. Geological Survey and installed, after slight modifications, on the Bureau's wang computer. System testing and validation were conducted using a simple textbook-type example. This was then followed by variogram analysis, point kriging, and block kriging of data from a previously worked section of a uranium mine. Kriging provides the best linear unbiased estimator in the sense that its estimation variance (kriging variance) is a minimum. The kriging variance then determines the accuracy of a reserve estimate and the risk associated with a financial decision based on that estimate.
