Answers are grounded in this paper and cite their sources.
Accurately quantifying mineralogy from drillhole to deposit scales remains a major challenge in exploration and mining, where mineralogical information underpins alteration mapping, mineral vectoring, and orebody characterization. Quantitative X-ray diffraction (QXRD) provides robust estimates for crystalline phases but is costly and requires specialized modeling for poorly crystalline alternation minerals, while routine multi-element geochemistry alone cannot reliably distinguish mineral species. This work presents an operationally scalable geochemical and spectral workflow that integrates routine multi-element assays (four-acid ICP-MS supplemented by pXRF Si and whole-rock analyses) with visible near infrared-shortwave infrared (VNIR-SWIR) spectroscopy to predict modal mineralogy using a dynamic linear-programming (LP) optimization model. VNIR-SWIR interpretations constrain the active mineral library for each sample, while bulk chemistry controls predicted modal abundances. Testing on sixteen paired drillcore sample sets from the Camp Creek Cu–Au–Mo porphyry system (British Columbia, Canada), representing ~10 cm core samples and corresponding 2 m drillhole intervals, shows that VNIR–SWIR-informed dynamic libraries improve agreement with QXRD relative to chemistry-only models. This results in superior elemental mass-balance closure, achieving better elemental mass-balance closure than QXRD-derived mineralogy alone. Per-sample root mean square error (RMSE) decreases from ~10% to ~6% on average, with improved quartz, feldspar, and hydrous alteration mineral abundances; most VNIR-SWIR-informed samples differ from QXRD by <10% RMSE (nearly half by <5%). A Diebold-Mariano test per mineral (α = 0.05) reveals statistically significant improvements using the VNIR-SWIR in the LP model for predicted quartz, white mica, diaspore, albite, K-feldspar, carbonate, biotite, chalcopyrite, apatite, and pyrite. Ten minerals showed non-significant negative changes, and two minerals showed no directional change when considering the VNIR-SWIR mineralogy. These results demonstrate that fusing routine geochemical assays with VNIR-SWIR mineralogy constraints enables operationally scalable, cost-effective mineralogical prediction from standard industry datasets, with the potential to support deposit-scale mineral vectors and alteration models applicable to exploration and orebody characterization in complex hydrothermally altered systems.
No takes yet. Share an insight, caveat, or question.
Harraden et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: