Erratic Highs -- A Perennial Problem In Resource Estimation

Most statistical and geostatistical methods of re-source and reserve estimation work extremely well for Normal (Gaussian) well-behaved information. However, most interesting mineral deposits exhibit far from this idealistic distribution of values. It seems that the more valuable the mineral, the less likely the data are to have a Normal distribution. Values tend to be "positively skewed" with a long tail into the high values. Using averaging techniques, such as inverse distance or ordinary kriging may result in consider-able overestimation of grades and tonnages unless such values are cut or otherwise modified. This paper presents some real case studies where distributions are (a) highly skewed and (b) the product of multiple phases of mineralisation. Practical methods are presented for solving such problems including combinations of indicator and log normal kriging methods. The emphasis of the presentation is on developing operational methods which can be used on project evaluation or grade control in producing mines.