Hyperspectral Imaging as a Public Geoscience Tool

Dr. Philip Lypaczewski, Lead Researcher – Hyperspectral, College of the North Atlantic

Published on December 14, 2021

Figure 1A) The Hyperspectral Scanning Unit operating within its transport container. B) Early data example for three core boxes showing mineralogy and mineral chemistry.

As the mining industry progresses towards automation, digitalization, and more sustainable mining practices, reflectance spectroscopy, also known as Hyperspectral Imaging if an imaging component is present, is a technology that is bound to become routinely used in the sector. Hyperspectral imaging is a passive, contactless and non-destructive spectroscopic analysis technique that allows the rapid and objective identification of mineralogy (Table 1) on a variety of rock surfaces, without the need for any sample preparation. In the commonly used Shortwave Infrared spectral range (SWIR, 1000-2500 nm), most hydrated minerals can be uniquely identified from absorption features caused by specific cation-OH bonds. In many cases, the relative strength and exact position of these absorptions also allows to quantitatively estimate mineral chemistry. For example, the Al2-OH absorption of white mica near 2200 nm is used to estimate its AlVI content, with a position varying from 2190 nm for Al-rich (muscovitic) to 2215 nm for Al-poor (phengitic) white micas, a change that is often useful in identifying hydrothermal alteration. Additionally, as spectral response in this wavelength range is not negatively impacted by the fine-grained nature of alteration minerals, it is a particularity well suited tool to objectively characterize fine-grained hydrothermal alteration mineralogy around a potential deposit, which can often be difficult to do with the naked eye alone.

Table 1: Mineralogy detected by reflectance spectroscopy, by commodity type

Deposit type

Spectral range
Wavelength (nm)

Minerals detected by hyperspectral


(400-1000 nm)

Direct elemental detection of REEs e.g., Nd in calcite, monazite, eudialyte, etc. (805 nm)


(1000-2500 nm)

Proximal alteration (potassic, phyllic zones)

  • Muscovite (2200 nm AlVI content, crystallinity)
  • Biotite (2250 nm Mg#)

    Distal alteration (argillic, propylitic, Na-Ca zones)

  • Kaolin-group minerals (2160 nm)
  • Epidote (1550 nm Fe/Al content)
  • Prehnite (1475 nm)
  • Chlorite (2250 nm Mg#)
  • Amphiboles (2400 nm Mg#)

Epithermal (Au)

(1000-2500 nm)

Low sulphidation

  • Smectites (2200 nm AlVI content)
  • White mica (2200 nm AlVI content, illite, etc.)
  • Chlorite (2250 nm Mg#)

    High sulphidation

  • Topaz (2080 nm)
  • Pyrophyllite (2160 nm)
  • Alunite/Jarosite (1480/1740 nm)
  • Kaolin-group (2160 nm)
  • White mica (2200 nm AlVI content, illite, etc.)

Orogenic (Au)

(1000-2500 nm)

Biotite (2250 nm Mg#)
Chlorite (2250 nm Mg#)
(2200 nm AlVI content, crystallinity)

Fe deposits

(400-1000 nm)

(1000-2500 nm)

Magnetite (1100 nm)

Hematite (1100 nm)

Hydrous Fe-oxides (1100/1900 nm nm)

Carbonates (2300 nm)

Amphiboles (2400 nm)

Industrial materials

(400-1000 nm)

(1000-2500 nm)

Detection of impurities (e.g., gypsum [1530 nm], clays [2200 nm], carbonates [2330 nm], hydrated minerals)

All deposits

MWIR (2500-5000 nm)

Carbonates (+ Chemistry)

All deposits

LWIR (7500-13000 nm)

Quartz / Silicification

Feldspars (+ Chemistry)

In active exploration or exploitation sites, mobile hyperspectral instruments could readily be deployed on-site, acquire data on entire drill core boxes in under one minute, and deliver immediate, consistent, and objective mineralogical results. The integration of such same-day results by exploration geologists could allow for real-time decision making early in exploration campaigns, allowing for better vectoring towards mineralization. Within a potential deposit, hyperspectral data could provide the core logging geologist a priori knowledge on the material to be investigated, which could be used to delineate mineralized intervals more accurately, leading to reduced assay costs and minimization of cutting and rock waste. At the preliminary economic assessment stage, hyperspectral imaging data would allow the early recognition of deleterious minerals that could later affect processing of ore (e.g., the presence of talc, clays, or others).

The digital mineralogical data obtained from hyperspectral imaging, combined with a rapid data acquisition rate (upwards of 1000 meters per shift) make it an ideal technology to employ for digitalization initiatives for existing drill core repositories, where hundreds of thousands of meters of drill core may be archived. Several public drill core scanning initiatives are currently underway to acquire reflectance data on historical and recent drill core. The Australian National Virtual Core Library (NVCL) is perhaps the earliest such digitization project and has from 2010 to the present day acquired line-scan data for over 1,000,000 meters of drill core that is publicly accessible online (AuScope), reaching both academic and industrial audiences. Similar initiatives have been undertaken in Sweden in 2014, where hyperspectral imaging data was acquired for over 200,000 meters of drill core (Geological Survey of Sweden), in Minnesota in 2019 with nearly 5,000 meters (Minnesota Department of Natural Resources) and in Finland in 2020 with 40,000 meters (Geological Survey of Finland).

In Canada, an ongoing public geoscience initiative at College of the North Atlantic (CNA) has for goal to acquire hyperspectral imaging data on up to several hundred thousand meters of drill core stored in Newfoundland and Labrador’s Department of Industry, Energy and Technology public Core Storage Libraries (DIET). This drill core digitization process will facilitate the dissemination of the geological information contained in part of the over 1.2 million meters of archival core, currently housed in six geographically distant facilities across the province (Drill Core Storage).

The Hyperspectral Scanning Unit project consists of a mobile drill core scanning system (Photon Etc.’s nCore platform, Fig. 1A) using of a full range of hyperspectral cameras spanning from the Visible-Near Infrared (VNIR, 400-1000 nm), Shortwave Infrared (SWIR, 1000-2800 nm), Midwave Infrared (2800-5400 nm) to the Longwave Infrared (LWIR, 7500-13000 nm) spectral ranges, acquiring coregistered data at 1 mm/pixel. In addition, a high-resolution RGB camera (0.1 mm/pixel) and a laser profilometer are also integrated on the platform. Data acquisition for the full suite of instruments is done in 60 seconds, allowing up to 2000 meters of core to be scanned per day, a necessary rate to acquire data on the large amount of core in the provincial archives. Ultimately, the data acquired from this core digitalization initiative will result in the release of a public geoscience database online, which will facilitate geoscience research by aggregating mineralogical data into a single database. In turn a better understanding of regional mineralogical alteration patterns could facilitate the discovery of new mineral deposits in Newfoundland and Labrador. An early dataset containing VNIR and SWIR information is presented in Figure 1B, with data processed to highlight both the presence and chemistry of white mica and chlorite, as well as the presence of carbonates. Ultimately, ancillary datasets such as Au assays will be added to downhole imagery, which will facilitate the understanding of relationships between mineralization and mineral chemistry.

About the Author 

Dr. Lypaczewski is the lead researcher on the Hyperspectral Scanning Unit project at College of the North Atlantic, after having completed his B.Sc. in geology at McGill University, and his Ph.D. in hyperspectral imaging at the University of Alberta. His main research interests are in the applied use of lab- and field-based hyperspectral instruments for the mining industry. Particular research interests include the development of fundamental spectroscopy of minerals, and the development of spectral metrics that have direct applications to the mining industry, either by allowing vectoring towards mineral deposits, or by characterizing exploited material directly at the mine site.