Identifying and Dealing with Outliers in Resource Estimation

In skewed distributions seen in many geochemical data sets, samples at the extreme ends of the distribution are often termed ‘outliers’. Samples considered as outliers can often make exploratory data analysis and resource estimation more difficult and less reliable. Such outlier samples may be legitimate because they are part of a natural high- or low-grade tail in the distribution. How these samples are identified and dealt with in resource estimation can often be a material issue for the Competent Person, particularly under the latest edition of the JORC Code (JORC, 2012). This paper discusses the identification and characterisation of outliers and high- and low-grade tails and proposes methods to deal with them in estimation.For dealing with outliers and high-grade tails ,very simple approaches, such as top-cutting through to more advanced geostatistical approaches, are discussed. The strengths and weaknesses of each approach are also outlined. Some examples of how the underlying geological understanding of a deposit influences the view of outliers and low-/high-grade tails is provided. The importance of documenting, communicating and challenging a given geological model is also discussed. Finally, some ideas for dealing with the uncertainty created by outliers are outlined.CITATION:De-Vitry, C, 2014. Identifying and dealing with outliers in resource estimation, in Proceedings Ninth International Mining Geology Conference 2014 , pp 159–168 (The Australasian Institute of Mining and Metallurgy: Melbourne).