Geology and Remote Sensing: Landsat, ASTER and Sentinel-2

Geology and remote sensing have been partners since the first Landsat images showed faults and rock units no field map had recorded. In remote sensing mineral exploration, satellites and aircraft measure reflected and emitted energy from the ground. Geologists then use the spectral fingerprints of minerals and the shape of the terrain to map rock types, structures and hydrothermal alteration over areas far larger than any field team could walk.

Try it: Pixel check, will this satellite resolve your target? →

A classic review by Floyd Sabins in Ore Geology Reviews summed up the two jobs remote sensing does for explorers. It maps the geology and the faults and fractures that localise ore deposits, and it recognises hydrothermally altered rocks by their spectral signatures. Both still hold. What has changed is the data. Free, analysis-ready archives, better sensors, cloud computing and machine learning have turned mineral exploration using remote sensing from a specialist exercise into a standard first step for any licence, ahead of the ground work described in our guide to the stages of a mineral exploration program.

“Australia’s national ASTER geoscience maps were the first continent-scale maps of Earth’s surface mineralogy.”

How geology and remote sensing work together

Remote sensing measures electromagnetic energy. In the visible and near-infrared (VNIR, roughly 0.4–1.0 µm) and shortwave infrared (SWIR, roughly 1.0–2.5 µm), sensors record sunlight reflected from the surface. In the thermal infrared (TIR, roughly 8–12 µm), they record energy emitted by the ground itself. Minerals absorb energy at characteristic wavelengths because of their chemistry and crystal structure, so a reflectance curve works like a fingerprint.

• 🟢 VNIR: sensitive to ferric iron in hematite, goethite and jarosite, the rusty gossans and iron-stained soils above weathered sulphides. Also used to mask vegetation.
• 🟣 SWIR: sensitive to hydroxyl-bearing minerals and carbonates. Al-OH minerals such as kaolinite, illite, muscovite (sericite) and alunite absorb near 2.2 µm; Mg-OH minerals such as chlorite and epidote, and carbonates, absorb near 2.3 µm.
• TIR: sensitive to silicate and carbonate structure, useful for mapping quartz-rich rocks, silicification and lithology.
• Elevation and radar: digital elevation models and radar images reveal lineaments, fault scarps, folds and drainage patterns, even where spectral data are masked.

The geology part of geology and remote sensing is interpretation. A clay anomaly can be hydrothermal alteration around a porphyry, or it can be weathered shale or a dry lake bed. A lineament can be a mineralised shear zone, a dyke, or a fence line. Remote-sensing products become useful only when a geologist reads them against the known rock types, deposit models and structural history of the area.

The satellites explorers use: Landsat, Sentinel-2, ASTER and hyperspectral

Four families of sensors do most of the work in remote sensing and mineral exploration. They differ in how many wavelengths they record, how sharp their pixels are, and how often they revisit.

Landsat 8 and 9

The US Geological Survey’s Landsat 8 and 9 carry the Operational Land Imager (OLI), with nine bands at 30 m (and a 15 m panchromatic band), including two SWIR bands at 1.57–1.65 µm and 2.11–2.29 µm. Their thermal sensor adds two bands at 100 m. Each satellite revisits every 16 days. Landsat 8 mineral exploration workflows remain the standard regional first pass, and the price history explains a lot of that.

Before the USGS opened the Landsat archive in 2008, a single scene cost between $400 and $4,000. After the free-data policy, the same scene cost nothing. A district-scale study that once needed a budget line for imagery now needs only processing time.

Sentinel-2

The European Space Agency’s Sentinel-2 mission carries a Multispectral Imager with 13 bands: four at 10 m, six at 20 m and three at 60 m, over a 290 km swath. The twin-satellite design gives a revisit of about five days at the Equator. Copernicus data are free, full and open. Its sharper pixels are valuable for mapping gossans and iron oxides on small licences.

ASTER

ASTER, a Japanese instrument on NASA’s Terra satellite (launched in December 1999), has 14 bands across VNIR, SWIR and TIR, with pixels from 15 to 90 m. Its six SWIR bands made it the favourite satellite for mapping clay, mica and carbonate alteration. The SWIR detectors failed in April 2008, so SWIR data since then are unusable. The pre-2008 archive remains one of the most valuable free mineral-mapping datasets available, and the VNIR and TIR bands continue to work.

Hyperspectral satellites: EnMAP and EMIT

Hyperspectral, or imaging spectroscopy, sensors record hundreds of narrow contiguous bands, producing a near-continuous spectrum for each pixel. Germany’s EnMAP, launched in April 2022, covers 420–2450 nm at 30 m with more than 200 bands. NASA’s EMIT, installed on the International Space Station in July 2022 to map mineral dust sources, records 285 channels from 380 to 2500 nm at about 60 m. These sensors can separate mineral species that broadband sensors blur together.

Landsat or Sentinel-2: which to start with?

For most licences, use both. Sentinel-2’s 10 m and 20 m bands resolve smaller gossans, outcrops and tracks, and its roughly five-day revisit makes cloud-free, dry-season scenes easier to find. Landsat’s archive is longer, so it is the tool for comparing today’s surface with decades-old imagery, for example to see how artisanal workings or vegetation have changed. Where both are available, harmonised HLS data let you treat them as one time series. The pre-2008 ASTER archive then adds the SWIR detail neither of them has.

Radar, elevation and commercial high-resolution data

Optical sensors need clear skies and daylight. Radar satellites such as ESA’s Sentinel-1 send their own microwave signal, so they image through cloud and at night. Their response to surface roughness and moisture makes them useful for structural mapping in the humid tropics, where optical scenes are often clouded out. Global digital elevation models add the third dimension: hillshades lit from several directions pull out lineaments that a single image hides.

Commercial satellites add sharper pixels and, on some platforms, extra SWIR bands. They cost money per square kilometre and are best used on a narrowed target area after the free archives have done the regional screening. In practice, most geology and remote sensing programs start free and buy only what the first pass justifies.

Pixel size versus target size

Band count is one limit; pixel size is the other. Sentinel-2 records its visible and near-infrared bands at 10 m and its SWIR bands at 20 m. Landsat OLI, ASTER SWIR and EnMAP work at 30 m, and EMIT at about 60 m. A 30 m pixel covers 900 m², and anything narrower than that is averaged with the ground around it. Before you commission or download anything, check how many pixels your licence and your target actually occupy.

Mineral mapping using remote sensing: the core methods

Raw satellite images need processing before they say anything about minerals. A typical workflow for mineral mapping using remote sensing runs through the same steps, whatever the sensor.

1. Pre-processing. Radiometric calibration, atmospheric correction to surface reflectance, and masking of cloud, water and vegetation. Analysis-ready products such as NASA’s Harmonized Landsat and Sentinel-2 (HLS), which combines the two missions at 30 m for observations every two to three days, save much of this work.
2. Band ratios. Dividing one band by another cancels out brightness and topographic shading and highlights absorption features. Sabins’ review describes Landsat ratio images identifying two alteration assemblages: iron minerals, and clays plus alunite. Similar ratios using Sentinel-2 and ASTER bands are widely used.
3. Principal component analysis. Statistical transforms, including the “Crosta technique”, isolate the component that carries a given mineral’s signal.
4. Spectral matching. For hyperspectral data, algorithms such as Spectral Angle Mapper and matched filtering compare each pixel’s spectrum with library spectra of target minerals.
5. Structural mapping. Lineaments, fold traces and circular features are mapped from elevation models, radar and enhanced imagery.
6. Integration and ranking. Alteration and structure layers are combined with geology, geochemistry and geophysics in a GIS, often with machine-learning models, to rank targets.

Geological mapping using remote sensing

Not every use of remote sensing in mineral exploration is about alteration. Geological mapping using remote sensing is often the first and most valuable product, especially in regions where published maps are old, coarse or missing. False-colour composites separate granites from greenstones, sediments from volcanics and regolith from bedrock. Elevation models show fold closures and fault offsets. Radar highlights texture through light vegetation and cloud. None of it replaces boots on the outcrop; our guide to field geological mapping and map scales covers what the ground check adds.

Australia shows what systematic geology and remote sensing can deliver. CSIRO, Geoscience Australia and the state and territory surveys built national ASTER geoscience maps from about 3,000 scenes, released in 2012 as the first continent-scale maps of surface mineralogy, covering iron oxides, clays, carbonates, quartz, muscovite and chlorite. CSIRO reports they were downloaded 40,000 times in the first three months and became a default base map for exploration in Australia.

Many African countries have excellent exposure in their arid and semi-arid regions, and variable cover elsewhere. In the Nigerian basement, the East African Mozambique Belt and the greenstone belts of Southern Africa, geology and remote sensing are used to map pegmatite fields, shear-hosted gold corridors and granite contacts before field teams deploy. Vegetated and deeply weathered terrain needs more care, which the limitations section covers.

“Before the Landsat archive became free, a single scene could cost between $400 and $4,000, according to the USGS.”

Remote sensing application in mineral exploration, deposit by deposit

The same sensors are used for every commodity, but what you look for changes with the deposit model. These are common starting points for the application of remote sensing in mineral exploration.

• ✔ Porphyry copper-gold: concentric alteration zoning (potassic core, phyllic sericite shell, argillic and propylitic outer halos) mapped with SWIR bands, plus iron-oxide caps from VNIR.
• ✔ Epithermal gold-silver: alunite, kaolinite and silica caps along structures; SWIR and TIR together are most useful.
• Orogenic gold: subtle sericite, carbonate and iron-oxide alteration along shear zones; structural mapping often matters more than spectra (see how to read a satellite gold map).
• Lithium pegmatites: pegmatite bodies and their host granites and schists mapped from texture, colour and mica-related spectral response, then checked in the field.
• Iron ore and manganese: strong iron-oxide responses and distinctive landforms on plateaux and ridges.
• Uranium and REE: host granites, carbonatites and structures; often integrated with airborne radiometrics.

Engineering geology and remote sensing

The same toolkit supports engineering geology and remote sensing work: mapping landslide scars, unstable slopes, karst, flood-prone ground and construction materials, and monitoring mine sites, tailings and haul roads over time. For a project moving towards development, repeat Sentinel-2 and Landsat imagery gives an independent record of land disturbance and rehabilitation.

Remote sensing and geophysics together

Remote sensing reads the surface; geophysics reads the subsurface. A clay-alteration halo that sits over a magnetic low and an IP chargeability high is a far stronger target than any one of the three alone. Many national geological surveys publish regional airborne magnetic and radiometric data (our overview of what each geophysical survey method sees explains them), and overlaying these on satellite alteration and structural maps is one of the most cost-effective steps in remote sensing for mineral exploration. Radiometric potassium highs, in particular, often coincide with the sericite and potassic alteration that SWIR bands detect.

What a good remote-sensing target report should contain

Commissioning the work or reading someone else’s, you should be able to trace every target back to data. Check for these elements:

• Data sources: which sensors and scenes were used, and why they suit the terrain and mineral.
• Method: the band ratios, transforms or spectral matching applied, in enough detail to judge them.
• Masks: how vegetation, water, cloud and cultural features were removed.
• Maps in GIS format: georeferenced layers you can overlay on your own data, not only images in a PDF.
• Ranked targets with reasoning: why each zone ranks where it does, based on spectral, structural and geological evidence.
• Limitations and next steps: an honest statement of uncertainty and a field program to test the targets.

Limitations: where geology and remote sensing need fieldwork

Optical remote sensing reads the top millimetres to centimetres of the surface. Everything below is inference. That is a strength, because alteration and structure at surface often point to mineralisation at depth, but it defines the limits too.

• ⚠ Vegetation and cover: dense forest, crops, transported sand and deep laterite mask bedrock. Leaf-off or dry-season imagery, soil and regolith mapping, and structural analysis help, but some terrain needs geochemistry or geophysics instead.
• ⚠ Spectral resolution: Landsat and Sentinel-2 have only two SWIR bands, so they detect “clay-type” alteration but cannot separate kaolinite from illite or alunite. That is what ASTER and hyperspectral data add; our guide to hyperspectral mineral mapping covers those sensors in detail.
• ⚠ Spatial resolution: 30 m pixels average everything inside them. Narrow veins and small outcrops fall below detection. For metre-scale follow-up, see our guide to drone geophysics and LiDAR surveys.
• Look-alikes: many non-mineralised rocks share spectral features with alteration minerals.
• Depth: optical sensors do not see below the surface. Depth estimates in any remote-sensing product are model-based, not measurements. Ground-penetrating radar is often pitched as the fix; our explainer on GPR for gold sets out how shallow it really reads.

The spectral-resolution point is easiest to see on a wavelength axis. Landsat’s SWIR2 band spans 2.11–2.29 µm and Sentinel-2’s B12 roughly the same window, so the Al-OH feature near 2.2 µm and the edge of the Mg-OH and carbonate features near 2.3 µm fall into one band. ASTER’s published band table splits the same window into five narrow SWIR bands (bands 5 to 9, 2.145–2.430 µm), and the hyperspectral sensors cover it continuously.

Timing helps more than many people expect. An anomaly that appears in a single wet-season scene may be soil moisture, crop residue or a burn scar; one that persists across several dry-season scenes over different years is far more likely to reflect the rocks. Comparing multiple dates is a cheap way to strip out transient effects, and it is why seasonal validation is part of how we check anomalies before they reach a report. For geology and remote sensing in savanna and farmed terrain across much of Africa, choosing the right months can decide whether the bedrock signal is visible at all.

How we apply geology and remote sensing to your licence

Our satellite-based mineral detection service applies the methods above to your own area. We analyse multispectral and, where appropriate, hyperspectral data, choosing the source to suit the mineral and terrain, and process it through our own algorithms to flag likely mineralised target zones, alteration halos, faults and fractures and deposit-associated geological patterns.

• What you send: coordinates, a KML/KMZ file or a polygon, plus the country and target mineral(s).
• Turnaround: 5–20 business days depending on area size and mineral complexity.
• 📊 Premium mineral-intelligence report: high-potential zones, prospectivity heatmaps, estimated location and depth ranges, indicative quantity, geological interpretation (faults, alteration, host rock), seasonal anomaly validation, and PDF plus georeferenced GIS files.
• 📊 Premium+: adds TargetMax™ Drilling Intelligence: drilling-angle recommendations and interactive 3D subsurface models of vein structures and mineral distribution.
• 📊 Why screen first: exploration timelines from months to days, and up to 80–85% lower early-exploration cost, with no ground disturbance.
• ⚠ What it isn’t: a resource estimate. Our targets tell you where to sample and drill first.

We have scanned 100,000+ hectares for 20+ mineral types across 25+ countries in Africa, the Americas, Asia and Australia. See a sample of satellite-driven 3D mineral prospectivity mapping, or start by drawing your boundary on mining.farmonaut.com: Map Your Mining Site.

A do-it-yourself starting workflow

Want to explore the free data yourself before commissioning work? This sequence is a sensible start. It won’t replace an expert interpretation from one of the exploration consultants who design field programs, but it teaches you what your ground looks like from orbit.

1. Get the boundary right: your licence polygon in WGS84 as KML or shapefile.
2. Download cloud-free, dry-season scenes of Sentinel-2 and Landsat (surface reflectance or HLS) and any pre-2008 ASTER scenes.
3. Build false-colour composites to see rock units and structures, and a simple iron-oxide ratio from the red and blue bands.
4. Build a SWIR clay ratio from the two SWIR bands, and mask vegetation using a vegetation index.
5. Map lineaments from a digital elevation model with hillshades from several sun angles.
6. Overlay the public geology map and any known occurrences, and list the anomalies that coincide with favourable rocks and structures.
7. Field-check the best few, and record what each anomaly actually is on the ground.

Guides in this series

Two guides go deeper on parts of geology and remote sensing covered above.

• Drone geophysics and mine survey drones: magnetics, LiDAR and stockpile volumes flown from small aircraft.
• Hyperspectral remote sensing for mineral exploration: how imaging spectrometers separate individual alteration minerals.

Going further on gold: satellite maps and radar

Two narrower guides pick up where this one stops, both focused on gold.

• Satellite gold maps, layer by layer: what a prospectivity map actually plots, how its colours are ranked, and what to check before you trust a target.
• Ground penetrating radar and gold locators: what GPR can resolve in dry and wet ground, and why long-range locators are not a survey method.

Frequently asked questions

Reviewed September 2026 against the USGS Landsat 9 mission page and its EROS note on opening the Landsat archive, ESA’s SentiWiki Sentinel-2 specification and Copernicus free-data policy, NASA JPL’s ASTER band table and SWIR anomaly notice, the eoPortal EMIT entry, NASA Earthdata’s HLS catalogue, CSIRO’s national ASTER geoscience map pages, and Sabins (1999) in Ore Geology Reviews.

Satellite specifications and data policies change, so check the providers’ own documentation for the latest band tables and access terms. Remote-sensing anomalies are exploration targets, not mineral resources, and must be verified in the field.