Key points are not available for this paper at this time.
Hyperspectral drill-core sensors provide a fast and non-destructive technique for collecting reflectance spectral data from drilling products. The spectral response allows proportions of a wide range of minerals to be estimated. This information can be used for supporting accurate drill-core logging and alterations mapping that are essential components of reporting the geology of a drill hole. The AuScope National Virtual Core Library (NVCL) contains more than 5900 drill cores from across Australia with HyLogger3™ hyperspectral data collected by the Australian geological surveys. These data include the visible to near infrared, shortwave infrared and thermal infrared wavelength ranges. HyLogger3™ data can be pre-processed using various algorithms in The Spectral Geologist software (TSG™), and the results are available to the public through NVCL.In this paper, we present a workflow for HyLogger3™ data, called MyLogger, that translates the thermal infrared hyperspectral drill-core data into a first-pass geological log using the main rock-forming mineral groups. The workflow requires a few simple decisions by the user, and decisions are facilitated by the provision of interactive plots. The workflow is provided as a web app using Python™ and Streamlit, and uses the wavelet tessellation method for multiscale drill-core analysis as well as using Scikit-learn machine learning. This workflow is demonstrated using HyLogger3™ data from drill hole from South Australia’s Mineral Systems Drilling Program as a case study for the generation of a geologically meaningful and depth accurate first-pass geological log. The ability to automatically extract a geological log from hyperspectral drill-core data alleviates the time-consuming process of manual interpretation of spectra by subject-matter experts to allow a rapid assessment of drill-hole geology. This will make the geological information contained in the NVCL database as well as other hyperspectral drill-core data more accessible to a broader range of geoscientists.
Stromberg et al. (Mon,) studied this question.