Long-wave infrared (LWIR) spectra can be interpreted using a Random Forest machine learning approach to predict mineral species and abundances. In this study, hydrothermally altered carbonate rock core samples from the Fourmile Carlin-type Au discovery, Nevada, were analyzed by LWIR and micro-X-ray fluorescence (μXRF). Linear programming-derived mineral abundances from quantified μXRF data were used as training data to construct a series of Random Forest regression models. The LWIR Random Forest models produced mineral proportion estimates with root mean square errors of 1.17 to 6.75% (model predictions) and 1.06 to 6.19% (compared to quantitative X-ray diffraction data) for calcite, dolomite, kaolinite, white mica, phlogopite, K-feldspar, and quartz. These results are comparable to the error of proportion estimates from linear spectral deconvolution (±7–15%), a commonly used spectral unmixing technique. Having a mineralogical and chemical training data set makes it possible to identify and quantify mineralogy and provides a more robust and meaningful LWIR spectral interpretation than current methods of utilizing a spectral library or spectral end-member extraction. Using the method presented here, LWIR spectroscopy can be used to overcome the limitations inherent with the use of short-wave infrared (SWIR) in fine-grained, low reflectance rocks. This new approach can be applied to any deposit type, improving the accuracy and speed of infrared data interpretation.

Mineral distributions can be determined in drill core samples from a Carlin-type gold deposit, using micro-X-ray fluorescence (µXRF) raster data. Micro-XRF data were collected using a Bruker Tornado µXRF scanner on split drill core samples (~25 × 8 cm) with data collected at a spatial resolution of ~100 µm. Bruker AMICS software was used to identify mineral species from µXRF raster data, which revealed that many individual sample spots were mineral mixtures due to the fine-grained nature of the samples. In order to estimate the mineral abundances in each pixel, we used a linear programming (LP) approach on quantified µXRF data. Quantification of µXRF spectra was completed using a fundamental parameters (FP) standardless approach. Results of the FP method compared to standardized wavelength dispersive spectrometry (WDS)-XRF of the same samples showed that the FP method for quantification of µXRF spectra was precise (R2 values of 0.98–0.97) although the FP method gave a slight overestimate of Fe and K and an underestimate of Mg abundance. Accuracy of the quantified µXRF chemistry results was further improved by using the WDS-XRF data as a calibration correction before calculating mineralogy using LP. The LP mineral abundance predictions were compared to Rietveld refinement results using X-ray diffraction (XRD) patterns collected from powders of the same drill core samples. The root mean square error (RMSE) for LP-predicted mineralogy compared to quantitative XRD results ranges from 0.91 to 7.15% for quartz, potassium feldspar, pyrite, kaolinite, calcite, dolomite, and illite. The approaches outlined here demonstrates that µXRF maps can be used to determine mineralogy, mineral abundances, and mineralogical textures not visible with the naked eye from fine-grained sedimentary rocks associated with Carlin-type Au deposits. This approach is transferrable to any ore deposit, but particularly useful in sedimentary-hosted ore deposits where ore and gangue minerals are often fine grained and difficult to distinguish in hand specimen.

No abstract

No abstract

No abstract