Statistical analysis of stream sediments in the Mt. Freegold area of the Dawson Range, Central Yukon Territory, Canada

Abstract

The study uses data and samples from a 2009 and 2012 stream sediment sampling program, conducted in part by the author. With the aim to identify gold-bearing prospects, a data set of 417 samples was digested with aqua regia and analyzed using inductively coupled plasma mass spectrometry (ICP-MS). I had leftover material from the original samples reanalyzed by ICP-MS after 4 acid (nitric + hydrochloric + hydrofluoric + perchloric) digestion. I compared the compositional data for the two sample sets (308 samples, 45 elements). Accuracy and precision (reproducibility) for all elements of the aqua regia sample set is noticeable lower than that of the 4 acid digestion set. Accuracy is computed by the relative error that compares the assay of the standard sample to the known concentration of the standard. Precision is an estimate of the reproducibility of the sampling and analytical system. Precision is calculated by fractional uncertainty. Field duplicates are produced by splitting samples in two. Fractional uncertainty is calculated for each pair of original sample and field duplicate. Gold shows the lowest precision of all elements for both data sets, most likely due to the nugget effect. Using a paired sample T-Test I concluded that the mean concentrations of 42 elements are different for aqua regia versus 4 acid digestions. Only for three elements (sulphur, bismuth and copper) were non-distinguishable average element concentrations found for both digestion methods. A negative t-statistic (based on aqua regia vs. 4 acid digestion-based compositions) implies higher elemental recovery by 4 acid digestion. The aqua regia data set record only a portion (presented in brackets) of the values for 4 acid digestions for these elements: Na (1%), Hf (2%), Zr (3%), K (4%), Nb (5%), Sr (7%), Al (11%), Ti (12%), Rb (14%), Ba (16%) and Ca (19%). The digestion method determines the degree of element recovery from the solids (i.e., dissolution into solution) and consequently 'trustworthiness' of a given element's concentrations. This, in turn, dictates the element's usefulness for identifying prospective targets. The digestion method sets the frame for the spectrum of possible discoveries. As demonstrated 4 acid digestion has a high trustworthiness of a given element's concentrations over a broad spectrum of rock composition. Since the majority of elements have a high recovery, it is tempting to incorporate all analyzed elements in an analytical scheme for the evaluation of prospective targets. Thus I developed a new statistical method Mahalanobis Principal Component Analysis (MPCA) that considers element associations for all elements. Mahalanobis distance is used for identifying sample of odd element composition relative to the bulk of element compositions. The scheme is based on assuming that mineralization is not restricted to a few elements (pathfinders) rather than it is a multi-element affair. Principal Component Analysis is employed to determine the contribution of each element to the 'oddness' of the element composition of the anomalous sample. Hence the cause of a sample's abnormal element composition is revealed. Elements with a high contribution to the oddness provide clues about the type of abnormality and its exploration potential. Pearson correlation of samples composition is used for identifying similar samples. Samples of similar element composition share the same rock genesis and belong to the same lineage. Thus by plotting similar samples the spatial distribution of anomalies can be shown and it is used for the further evaluation of the target potential. The more reliable data set 4 acid is used to appraise the exploration potential of the Mt. Freegold sample area using MPAC. Fifteen exploration targets are identified. Not all of them are gold related.