Machine Learning for Mineral Identification and Ore Estimation from Hyperspectral Imagery in Tin–Tungsten Deposits: Simulation under Indoor Conditions

This study aims to assess the feasibility of delineating and identifying mineral ores from hyperspectral images of tin–tungsten mine excavation faces using machine learning classification. We compiled a set of hand samples of minerals of interest from a tin–tungsten mine and analyzed two types of hyperspectral images: (1) images acquired with a laboratory set-up under close-to-optimal conditions, and (2) a scan of a simulated mine face using a field set-up, under conditions closer to those in the gallery. We have analyzed the following minerals: cassiterite (tin ore), wolframite (tungsten ore), chalcopyrite, malachite, muscovite, and quartz. Classification (Linear Discriminant Analysis, Singular Vector Machines and Random Forest) of laboratory spectra had a very high overall accuracy rate (98%), slightly lower if the 450–950 nm and 950–1650 nm ranges are considered independently, and much lower (74.5%) for simulated conventional RGB imagery. Classification accuracy for the simulation was lower than in the laboratory but still high (85%), likely a consequence of the lower spatial resolution. All three classification methods performed similarly in this case, with Random Forest producing results of slightly higher accuracy. The user’s accuracy for wolframite was 85%, but cassiterite was often confused with wolframite (user’s accuracy: 70%). A lumped ore category achieved 94.9% user’s accuracy. Our study confirms the suitability of hyperspectral imaging to record the spatial distribution of ore mineralization in progressing tungsten–tin mine faces.

A Review of Tungsten Resources and Potential Extraction from Mine Waste

Tungsten is recognized as a critical metal due to its unique properties, economic importance, and limited sources of supply. It has wide applications where hardness, high density, high wear, and high-temperature resistance are required, such as in mining, construction, energy generation, electronics, aerospace, and defense sectors. The two primary tungsten minerals, and the only minerals of economic importance, are wolframite and scheelite. Secondary tungsten minerals are rare and generated by hydrothermal or supergene alteration rather than by atmospheric weathering. There are no reported concerns for tungsten toxicity. However, tungsten tailings and other residues may represent severe risks to human health and the environment. Tungsten metal scrap is the only secondary source for this metal but reprocessing of tungsten tailings may also become important in the future. Enhanced gravity separation, wet high-intensity magnetic separation, and flotation have been reported to be successful in reprocessing tungsten tailings, while bioleaching can assist with removing some toxic elements. In 2020, the world’s tungsten mine production was estimated at 84 kt of tungsten (106 kt WO3), with known tungsten reserves of 3400 kt. In addition, old tungsten tailings deposits may have great potential for exploration. The incomplete statistics indicate about 96 kt of tungsten content in those deposits, with an average grade of 0.1% WO3 (versus typical grades of 0.3–1% in primary deposits). This paper aims to provide an overview of tungsten minerals, tungsten primary and secondary resources, and tungsten mine waste, including its environmental risks and potential for reprocessing.

Machine-Learning for Mineral Identification and Ore Estimation From Hyperspectral Imagery in Tin-Tungsten Deposits

This study aims to assess the feasibility of delineating and identifying mineral ores from hyperspectral images of tin-tungsten mine excavation faces using machine-learning classification. We compiled a set of hand samples of minerals of interest from a tin-tungsten mine and analyzed two types of hyperspectral images: 1) images acquired with a laboratory set-up under close-to-optimal conditions; and 2) scan of a simulated mine face using a field set-up, under conditions closer to those in the gallery. We have analyzed the following minerals: cassiterite (tin ore), wolframite (tungsten ore), chalcopyrite, malachite, muscovite, and quartz. Classification (Linear Discriminant Analysis, Singular Vector Machines and Random Forest) of laboratory spectra had a very high overall accuracy rate (98%), slightly lower if the 450 – 950 nm and 950 – 1780 nm ranges are considered independently, and much lower (74.5%) for simulated conventional RGB imagery. Classification accuracy for the simulation was lower than in the laboratory but still high (85%), likely a consequence of the lower spatial resolution. All three classification methods performed similarly in this case, with Random Forest producing results of slightly higher accuracy. The user’s accuracy for wolframite was 85%, but cassiterite was often confused with wolframite (user’s accuracy: 70%). A lumped ore category achieved 94.9% user’s accuracy. Our study confirms the suitability of hyperspectral imaging to record the spatial distribution of ore mineralization in progressing tungsten-tin mine faces.

Methylobacterium nonmethylotrophicum sp. nov., isolated from tungsten mine tailing

A novel pink-pigmented strain, designated 6HR-1T, was isolated from tungsten mine tailings in Jiangxi Province, PR China. Cells were Gram-stain-negative, aerobic, non-spore-forming, rod-shaped and motile with a polar flagellum (monotrichous). It could not utilize methanol, methylamine, formaldehyde or formate as a sole carbon source. The methanol dehydrogenase mxaF gene was absent but the xoxF gene was present. Phylogenomic and 16S rRNA gene phylogenetic analyses clearly showed that strain 6HR-1T was affiliated to the genus Methylobacterium and closely related to ‘Methylobacterium terrae’ 17Sr1-28T (98.6 %), Methylobacterium platani JCM 14648T (97.7 %), Methylobacterium variabile DSM 16961T (97.7 %) and Methylobacterium currus KACC 19662T (97.4 %). The average nucleotide identity and digital DNA–DNA hybridization values between strain 6HR-1T and its closely related type species were 87.4–88.7 and 33.2–36.3 %, respectively. It had summed feature 8 (C18 : 1 ω7c and/or C18 : 1 ω6c) as the major fatty acid and ubiquinone 10 as the predominant respiratory quinone. Polyphasic characterization supported that strain 6HR-1T represents a novel species of the genus Methylobacterium , for which the name Methylobacterium nonmethylotrophicum sp. nov. is proposed with the type strain 6HR-1T (=GDMCC 1.662T=KCTC 42760T).

Tailings microbial community profile and prediction of its functionality in basins of tungsten mine

In a circular economy concept, where more than 300 million tons of mining and quarrying wastes are produced annually, those are valuable resources, supplying metals that are extracted today by other processes, if innovative methods and processes for efficient extraction of these elements are applied. This work aims to assess microbiological and chemical spatial distribution within two tailing basins from a tungsten mine, using a MiSeq approach targeting the 16S rRNA gene, to relate microbial composition and function with chemical variability, thus, providing information to enhance the efficiency of the exploitation of these secondary sources. The tailings sediments core microbiome comprised members of family Anaerolineacea and genera Acinetobacter, Bacillus, Cellulomonas, Pseudomonas, Streptococcus and Rothia, despite marked differences in tailings physicochemical properties. The higher contents of Al and K shaped the community of Basin 1, while As-S-Fe contents were correlated with the microbiome composition of Basin 2. The predicted metabolic functions of the microbiome were rich in genes related to metabolism pathways and environmental information processing pathways. An in-depth understanding of the tailings microbiome and its metabolic capabilities can provide a direction for the management of tailings disposal sites and maximize their potential as secondary resources.