Statistical levelling of multi-element geochemical data

• Published in: Applied computing and geosciences Vol. 10; p. 100060
• Main Author: Williams, Peter M.
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.06.2021; Elsevier
• Subjects: Compositional analysis; Data levelling; Geochemical surveys; Yukon geochemistry; Geochemical surveys; Data levelling; Compositional analysis; Yukon geochemistry
• ISSN: 2590-1974; 2590-1974
• DOI: 10.1016/j.acags.2021.100060

Regional compilations of multi-element geochemistry can show shifts of level and dynamic range between component surveys, for example when analyses have been made using different laboratories or procedures or at different times. To create a unified composite picture of the geochemistry over a region, some form of relevelling of individual surveys may be needed. Existing treatments have focused on individual elements, independently, across individual pairs of map sheets. Such approaches, however, may fail to preserve the important covariance structure between elements of interest, and may risk dependency of the final blend on the order of levelling of sheets. The paper proposes a method for levelling all elements, and their covariances, simultaneously across all component sheets, taking full account of the compositional nature of the data. The method is shown to be computationally feasible, requiring manageable execution time even for large numbers of elements and large numbers of component surveys.