Kriging, Logarithmic Kriging, And Conditional Expectation: Comparison Of Theory With Actual Results - Introduction

According to statistical theory, the geostatistical methods of ore reserves estimation are greatly superior to any other method. However acceptance of geostatistics by the mining industry has been slow and not as wide spread as one would expect. One of the reasons for this cautious approach is that it is often questioned whether the geo- statistical assumptions are satisfied in real world situations. When blocks of ore are estimated in a mineral deposit, geostatistical theory can be used to determine which method of estimation should give the best results. Using this theory, the relationship between estimated value and true block value can be quantified. There is a need to verify that this theoretical relationship is actually observed in practice, and that the block estimates ' obtained using the theoretical ly best method of estimation, are actually most similar to the true block values. In this paper the results expected according to the theory are compared with the results actually obtained. Sample values from a large bedded gold deposit have been used. Blocks of ore have been estimated by a number of methods, including the polygone, inverse power of the distance, kriging, logarithmic kriging and other geostatistical methods. The block estimates have been calculated from samples on a large grid, and compared with the true block values obtained from additional samples on a much smaller grid. The results of this comparison show extremely good agreement between theory and actual results.