Deconstructing the Iron Boomerang—Quantitative Predictions of Hematite, Ochreous, and Vitreous Goethite Mixtures

Abstract

The so-called iron boomerang is constructed by using a combination of spectral metrics derived from the 900 nm 6A1→4T1 Fe3+ crystal field absorption. Previous work showed the iron boomerang provides qualitative information about the mineralogical type of spectral datasets of Western Australian iron ore deposits. That work is expanded to demonstrate how the shape of the iron boomerang is driven by a linear mixing regime between hematite and ochreous and vitreous goethite. The iron boomerang enveloping shape is defined by mixing pathways between 3 different 2-endmember systems of hematite–ochreous goethite, hematite–vitreous goethite and ochreous–vitreous goethite and a 3-endmember mixing regime in the interior of the boomerang. This provides a novel methodology of quantifying the relative hematite, ochreous, and vitreous goethite content from spectrally derived data. A combination of 4 spectral metrics as input features to a random forest regression model results in modelling root mean squared errors (RMSE) of approximately 4% when all endmembers are known and in a system where the endmembers are not known the RMSE is approximately 10%. Additionally, a means of identifying potential hematite endmembers from a spectral dataset is presented. Lastly, the regression models are applied to 3 iron ore diamond drillcore from the Hamersley Province in Western Australia and demonstrates that the proportions of hematite, ochreous, and vitreous goethite can be estimated downhole which, in turn, provides ore zone delineation and hardcap and hydrated zone identification.