Remote Sensing for Mineral Exploration

Mineral exploration begins with a simple problem: how do you decide where to spend limited money on ground work, drilling and geochemistry in a large, often remote and poorly mapped area? Remote sensing, the acquisition of information about the Earth’s surface from satellites and aircraft, helps answer this question cheaply and quickly.

Satellite sensors record the sunlight reflected or the heat emitted by the ground in many wavelength ranges, called bands. Different minerals, rocks, soils and vegetation interact with light in characteristic ways. By analysing these signals, geologists can map rock units, detect hydrothermal alteration associated with mineral deposits, identify faults and fractures, and prioritise targets before setting foot on the ground.

Remote sensing does not find gold, copper or lithium directly. It highlights geological features and alteration patterns that may be associated with mineralisation, so that field teams can focus on the most promising areas. It reduces cost, speeds up reconnaissance and is especially valuable in large or inaccessible regions such as parts of northern Kenya, the Horn of Africa and the wider African continent.

The Physics Behind It: Spectral Signatures

Sunlight covers visible light (roughly 0.4 to 0.7 micrometres), near-infrared (NIR, about 0.7 to 1.3 µm), shortwave infrared (SWIR, about 1.3 to 2.5 µm) and, beyond that, thermal infrared (TIR, roughly 8 to 14 µm). Materials absorb light at specific wavelengths because of the electronic transitions and molecular vibrations in their structure.

Iron oxides and hydroxides (hematite, goethite, jarosite) absorb strongly in the visible and near-infrared, giving rocks reddish, yellow or brown colours. They are common in weathered, oxidised sulphide zones, often called gossans.
Clay minerals and micas (kaolinite, alunite, illite, muscovite) have diagnostic absorption features in the SWIR near 2.2 µm, related to the aluminium-hydroxyl (Al-OH) bond.
Carbonates, chlorite and epidote show absorption features near 2.3 to 2.35 µm, related to the magnesium-hydroxyl (Mg-OH) and carbonate bonds.
Silicates (quartz, feldspar) have characteristic emissivity features in the thermal infrared.
Vegetation reflects strongly in the near-infrared and absorbs red light, which makes it easy to identify but can mask the geology below.

Hydrothermal systems that form many ore deposits, including porphyry copper, epithermal gold and volcanogenic massive sulphide deposits, alter the surrounding rocks and produce zones of clay, silica, iron oxide and other minerals. These alteration minerals have signatures that sensors with SWIR bands can detect.

The Main Sensors: ASTER, Landsat and Sentinel-2

ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer). ASTER is a NASA and Japanese instrument on the Terra satellite, launched in 1999. It has 14 bands: three in the visible and near-infrared at 15 m resolution, six in the shortwave infrared at 30 m, and five in the thermal infrared at 90 m. Its six SWIR bands were designed for mapping clays, carbonates and other alteration minerals, which made ASTER the most popular satellite for geological mapping and mineral exploration for many years. An important limitation is that the SWIR detectors stopped delivering usable data in 2008 because of a sensor problem, so ASTER SWIR data exist for the period from 2000 to 2008 and the thermal and VNIR bands continued. The archive is large, and in many regions the existing SWIR coverage remains highly valuable. Check the current status and data availability from the data providers before planning work.

Landsat (Landsat 8 and 9, and the earlier Landsat 5 and 7 archive). Landsat is a long-running programme providing free, global, consistent data since the 1970s. Landsat 8 and 9 carry the Operational Land Imager (OLI) with nine bands, including two SWIR bands (about 1.6 and 2.2 µm) at 30 m resolution, a panchromatic band at 15 m and a thermal sensor. Landsat has fewer SWIR bands than ASTER, so it cannot separate as many alteration minerals, but its band 6 and band 7 combination is effective for detecting clay and hydroxyl-bearing alteration, and the long archive supports time series and change analysis. Landsat 9 continues the record from 2021.

Sentinel-2. Sentinel-2 is a European Space Agency mission delivering free multispectral data with 13 bands at 10, 20 and 60 m resolution and a revisit time of a few days with its multiple satellites. It includes visible and near-infrared bands at 10 m, red-edge and SWIR bands at 20 m (two SWIR bands near 1.6 and 2.2 µm) and bands for atmosphere correction at 60 m. The higher spatial resolution and frequent coverage make Sentinel-2 excellent for geological and structural mapping, iron oxide detection, and general mapping in areas with variable cloud cover. Like Landsat, it has limited SWIR band count compared with ASTER, so mineral discrimination is more limited.

Other useful sources.

Sentinel-1 and other radar sensors (synthetic aperture radar) penetrate cloud and respond to surface roughness and structure, helping map faults, lineaments and geomorphology, including in tropical or cloudy regions.
Commercial high-resolution satellites (such as WorldView and others) provide sub-metre to few-metre imagery for detailed mapping, and some offer additional SWIR bands.
Hyperspectral sensors (for example from airborne surveys and spaceborne missions such as EnMAP and PRISMA) record hundreds of narrow bands and can identify minerals with much finer detail. Availability and cost vary.
Digital elevation models (SRTM, ALOS and others) for terrain and structure.

Common Remote Sensing Techniques in Exploration

1. Band ratios. Dividing one band by another highlights spectral features while reducing the effect of illumination and topography. Widely used examples include:

Iron oxide ratio: red divided by blue (for Landsat, band 4 over band 2; for Sentinel-2, band 4 over band 2) highlights ferric iron oxides.
Clay or hydroxyl ratio: SWIR1 divided by SWIR2 (for Landsat 8, band 6 over band 7; for Sentinel-2, band 11 over band 12) highlights clay and hydroxyl-bearing minerals.
Ferrous iron ratio: SWIR divided by NIR (for example Landsat 8 band 6 over band 5) highlights ferrous minerals.

Ratios are often combined in a false-colour composite, with each ratio assigned to a colour channel, so that alteration zones stand out visually. Exact band numbers depend on the sensor, so always check the sensor’s band definitions.

2. Principal component analysis (PCA) and the Crosta technique. PCA transforms correlated bands into uncorrelated components. The Crosta (feature-oriented principal component selection) method applies PCA to selected band subsets to map iron oxide and hydroxyl alteration separately. It is a well-known approach with Landsat and ASTER data.

3. ASTER-specific mineral indices. Researchers have developed band-ratio indices for ASTER’s SWIR and TIR bands to map specific minerals, such as an alunite-kaolinite index, a sericite-muscovite index, carbonate and chlorite indices and a quartz index from thermal bands. These offer more mineral specificity than Landsat or Sentinel-2.

4. Spectral classification and mapping.

Spectral Angle Mapper (SAM) compares each pixel’s spectrum with reference spectra and measures the angle between them, which is relatively insensitive to illumination.
Matched filtering and linear spectral unmixing estimate the abundance of target minerals in each pixel.
Supervised and unsupervised classification group pixels with similar spectra into classes for lithological mapping. Reference spectra can come from spectral libraries (such as USGS and others) or from field and laboratory measurements of samples from the study area.

5. Lithological mapping. Combining visible, SWIR and thermal bands helps distinguish rock types, such as mafic versus felsic rocks, carbonates and siliciclastic sediments. The thermal bands of ASTER are especially helpful for silica content.

6. Structural interpretation and lineament mapping. Faults, shear zones, folds and fracture systems control many mineral deposits. Satellite imagery and DEMs allow geologists to map linear features, offsets and drainage patterns over large areas. See our article on lineament extraction from satellite data.

7. Multi-temporal and change analysis. Comparing images over time can reveal active mining, vegetation changes and seasonal effects, and helps choose the best images with minimal cloud or vegetation cover.

A Practical Workflow for Remote Sensing in Exploration

Step 1: Define the exploration model. What deposit type are you targeting (for example orogenic gold, porphyry copper, iron oxide-copper-gold, lithium pegmatites or rare earths)? What alteration, structure and lithology are expected? This determines which spectral features to look for and which sensors are suitable.

Step 2: Select and acquire data. Choose imagery with minimal cloud cover, low vegetation cover and good illumination. The dry season is often best in vegetated regions. Download free data from the relevant archives, and consider commercial data for detail.

Step 3: Pre-process.

Radiometric and atmospheric correction to convert raw digital numbers to surface reflectance, using tools such as Sen2Cor for Sentinel-2, or standard surface reflectance products for Landsat and ASTER.
Geometric correction and co-registration so that layers align.
Masking of clouds, shadows, water, vegetation, urban areas and active agricultural fields, which can distort spectral analysis.
Topographic correction where steep terrain causes strong illumination effects.

Step 4: Enhance and map. Apply band ratios, PCA, indices and classification, and produce alteration and lithology maps. Compare multiple methods to see which areas are consistently highlighted.

Step 5: Integrate with other data. Overlay results with geological maps, geophysics (magnetics, radiometrics), geochemistry, known mineral occurrences and structural data. Alteration anomalies that coincide with favourable structures and geochemical signatures are higher priority.

Step 6: Field validation. Visit selected sites, collect samples and measure spectra with portable instruments if available. Check whether the satellite anomaly reflects real alteration or a false positive such as a recent burn scar, soil colour or non-geological surface cover. Use the results to calibrate the interpretation.

Step 7: Rank targets and plan follow-up. Prioritise areas for geochemical sampling, geophysics and drilling. Update the model with new data.

Applications in East Africa and Kenya

Remote sensing is widely used in the geologically diverse terrains of East Africa. Examples of applications include:

Orogenic gold in greenstone belts and shear zones. Structural interpretation of shear zones and detection of iron oxide and hydroxyl alteration along them. Western Kenya’s greenstone belts and associated shear zones are an example of a setting where structural mapping and alteration mapping are complementary. For related work, see our structural geology articles.
Mineral sands and heavy mineral deposits along the coast, mapped using spectral and geomorphological information.
Rare earth and carbonatite complexes, which have characteristic circular structures and spectral and radiometric signatures.
Industrial minerals, such as limestone, gypsum and soda ash, in the Rift Valley.
Geothermal exploration, using thermal anomalies, alteration minerals and structure.
Groundwater and hydrogeology, mapping lineaments, drainage and recharge zones.

Cloud and dense vegetation can limit optical remote sensing in high-rainfall or forested areas, whereas arid and semi-arid areas such as northern Kenya offer excellent exposure and are ideal for spectral mapping.

Limitations and Pitfalls

Honest practitioners stress that remote sensing is a guide, not proof:

Spatial resolution. At 10 to 30 m per pixel, small features may be missed or mixed with surroundings.
Spectral resolution. Multispectral sensors with few SWIR bands cannot separate many similar minerals. Hyperspectral data or ASTER perform better.
Surface cover. Vegetation, soil, regolith, dust and weathering rinds hide the underlying rock. In deeply weathered tropical terrains, surface signatures may not represent bedrock alteration.
False positives. Iron-stained soils, burnt areas, wetlands and anthropogenic features can mimic alteration anomalies.
Depth. Optical sensors see only the top few micrometres to centimetres of the surface. Buried deposits are not directly detected, though structural and indirect clues may help.
Illumination and topography. Shadows and slope effects can create artefacts.
Data quality and processing choices. Atmospheric correction and masking decisions affect results.
ASTER SWIR data gap. Newer data do not include the SWIR bands, so analysis depends on the archive.

Because of these limitations, remote sensing results should always be integrated with geology, geophysics and geochemistry and validated in the field.

Choosing the Right Sensor for Your Project

Need	Suitable data	Why
Detailed alteration mineral mapping	ASTER (SWIR archive), hyperspectral	More SWIR bands
Regional iron oxide and clay screening	Landsat 8/9, Sentinel-2	Free, global, easy ratios
Higher spatial detail and frequent updates	Sentinel-2, commercial imagery	10 m or finer resolution
Structure and geomorphology under cloud	Sentinel-1 radar, DEMs	All-weather, relief and roughness
Long-term change monitoring	Landsat archive	Decades of consistent data
Lithology and silica content	ASTER thermal bands	Thermal emissivity features

Costs and Accessibility

Landsat, Sentinel and ASTER archive data are available free of charge, although some access routes and processing tools require registration. Open-source software such as QGIS, SNAP and Python libraries, and cloud platforms such as Google Earth Engine, make advanced analysis accessible without costly licences. Commercial high-resolution imagery and hyperspectral data cost money, but can be worthwhile for detailed target generation. The main cost is professional expertise, which determines whether the outputs are geologically meaningful.

Frequently Asked Questions

Can satellites find gold directly? No. They detect surface properties such as alteration minerals, iron oxides, structures and lithology that may be associated with gold systems.

Which is better for mineral exploration, ASTER, Landsat or Sentinel-2? ASTER has the best SWIR mineral discrimination, Landsat provides a long archive and Sentinel-2 offers higher resolution and frequent coverage. Many projects use all three.

How much of the ground can I map? Anything from a single concession to entire countries, which is why remote sensing is so cost-effective in early-stage exploration.

Does vegetation prevent mapping? Dense vegetation limits optical mapping. Strategies include dry-season imagery, vegetation masking, radar and structural interpretation.

Do I still need field work? Yes. Field validation is essential to confirm interpretations and calibrate the results.

Conclusion

Remote sensing with ASTER, Landsat and Sentinel-2 gives mineral explorers a powerful, low-cost way to map lithology, alteration and structure across large areas and to focus costly field work on the best targets. Success depends on choosing the right sensor and methods, careful processing, geological interpretation and field validation. Used well, it shortens the path from concept to drill target.

For remote sensing, GIS, structural geology and mineral exploration services, contact Reg. Geol. Mbugua Githae (MGSK) through consult.innercoregeoconsultants.co.ke.