Satellite Images of Gold Deposits: What Each Sensor Shows

Satellite images of gold deposits do not show gold. Gold is present in rock at grams per tonne, far too little to colour a pixel that is 10 to 30 metres wide. What the images show is the rock that hot, mineral-rich water altered around a gold system: rusty iron stains, clay and mica minerals, silica caps and the faults that fed the fluids. A geologist maps those clues, then samples on the ground to find out whether any of them carry gold.

This guide is for licence owners, consultants and investors who want to know which satellite to look at, what each image can and cannot show, and how sharp the pictures are against a claim-sized target. It uses published instrument specifications from Copernicus, USGS and NASA, plus peer-reviewed case studies. For a plain-language look at the finished product, see our satellite gold map guide, which explains how to read a prospectivity map.

What satellite images of gold deposits actually show

Remote sensing means measuring the ground from a distance. In gold work it is used at the earliest stage of exploration, to choose where to spend money on fieldwork. Optical satellites record sunlight reflected from the surface in many wavelength bands. Different minerals absorb light at different wavelengths, so a band combination can separate altered rock from fresh rock. The review by Pour and Hashim (Springerplus, 2014) notes that Landsat images have been used for detecting alteration mineral assemblages associated with epithermal gold and porphyry copper mineralization. Hydrothermal alteration is the chemical change that hot fluids cause in the rock around a mineral deposit.

Five clues matter most. The table below pairs each clue with the part of the spectrum that reveals it and the sensors that cover that part.

The review explains that iron oxides absorb light between 0.4 and 1.1 µm, while clay, sulfate and carbonate minerals show their features in the shortwave infrared, which it calls the best spectral region for sensing hydrothermal alteration. That is why the SWIR bands decide how useful a satellite is for gold exploration. A camera that only records visible light can show you rust but cannot separate clay from soil.

Which satellites give usable images for gold exploration

Four groups of sensors are used. They trade sharpness against the number of colour bands and against how large an area one image covers. The figures below come from the operators’ own pages.

Chart

Spectral bands per sensor

number of bands recorded, including thermal bands where present

Chart

Width of one scene or swath

kilometres; selected sensors; wider means more ground per image

• Landsat 8 and 9 (USGS and NASA). The Landsat 9 imager has nine optical bands (all at 30 metres except one panchromatic band at 15 metres), including two SWIR bands, plus two thermal bands at 100 metres. Each satellite repeats every 16 days and Landsat 9 is offset by 8 days from Landsat 8. Landsat band 7 (2.11 to 2.29 µm) sits on the clay absorption feature, and the Pour and Hashim review notes that the ratio of band 5 to band 7 on earlier Landsat sensors is sensitive to hydroxyl-bearing minerals.
• Sentinel-2 (Copernicus, European Space Agency). The mission samples 13 bands: four at 10 metres, six at 20 metres and three at 60 metres, across a 290 kilometre swath. Its two SWIR bands, near 1,610 and 2,190 nanometres, are 20 metres wide. Two satellites together revisit a place every 5 days at the equator, which helps in cloudy regions because more passes give more clear views.
• ASTER (NASA and Japan, on the Terra satellite). It has 14 bands: visible and near-infrared at 15 metres, shortwave infrared at 30 metres and thermal at 90 metres, with a 60 kilometre swath. It has been widely used for alteration mapping, but its SWIR data have a gap, covered in the limits section below.
• Hyperspectral satellites. NASA’s Hyperion, retired in 2017, collected 220 spectral channels at 30 metres over scenes 7.7 kilometres wide. Germany’s EnMAP, now operating, records from 420 to 2,450 nanometres at 30 by 30 metres over a 30 kilometre swath, with a 27-day nadir revisit. Hundreds of narrow bands let a geologist tell individual clay minerals apart.

Landsat and Sentinel-2 data are public and can be downloaded without a licence fee, but they do not give the same mineral detail as hyperspectral data. A comparison of ASTER and the 224-band airborne AVIRIS instrument at Cuprite, Nevada found generally similar results, with AVIRIS identifying more individual minerals. The practical lesson: a broad-band satellite finds the zones, and a narrow-band one names the minerals inside them.

The first chart below compares how much ground one scene covers. Wide coverage is what makes satellites cheap for screening a large licence; narrow scenes suit a single prospect. The second chart compares the number of spectral bands. More bands mean finer separation of minerals, which matters more for gold work than a sharper pixel.

▶ Remote sensing for mining explained in plain words, with the main instrument families · 2:01

How sharp the pictures are against a claim-sized target

Pixel size is the ground distance one image cell covers. A 20 metre pixel averages everything inside a 20 by 20 metre square into one number, so a vein 2 metres wide never appears by itself. It shows only as a slight tint on a pixel that also contains waste rock and soil. What satellites map well is a wide alteration zone, often tens to hundreds of metres across.

Chart

Pixel size: iron bands against clay bands

metres on the ground; smaller is sharper

The third chart shows the trade inside each sensor: the visible bands are sharper than the SWIR bands that identify clay. Sentinel-2 gives 10 metre iron-oxide detail but 20 metre clay detail; ASTER gives 15 and 30 metres. Use the calculator below to see how many pixels fall on your own licence at each size.

Gold deposit types and how they look from space

The pattern on an image depends on the deposit model. Epithermal gold, formed from hot water close to the surface, is the best-documented target. The Pour and Hashim review reports that principal component analysis of ASTER bands mapped alunite, illite, smectite and kaolinite zones at Los Menucos, Patagonia, Argentina, and that a matched-filtering method on ASTER SWIR data gave regional and local alteration maps at the Somun Cura Massif in the same region. Our guide to epithermal gold styles and how to explore them explains the alteration zones in more detail.

Other styles show different clues. Gold in shear zones and veins is often controlled by faults, so structure matters as much as colour. Gold carried in porphyry copper systems sits inside broad, concentric alteration, which the review says is mapped with the same bands. Placer gold in river gravels is a terrain problem: elevation models and drainage patterns show old channels, and alteration in the source area points upstream. Check which model fits your ground before you read any image; our page on types of gold deposits and their host rocks lists the main styles, and the global gold deposits map shows where the belts lie.

Nevada shows why ground conditions matter. It was the leading gold-producing US state in 2025, with about 64% of estimated domestic production, according to USGS (Mineral Commodity Summaries, 2026). Its dry, sparsely vegetated ground leaves rock exposed, which suits optical imagery.

Worked examples from published studies

Cuprite, Nevada. This district has two alteration centres in volcanic rock, and USGS used it to test ASTER’s mineral-mapping ability. In a 2003 study in Economic Geology, USGS scientists showed that ASTER’s bands separate opal in a silicified core, alunite and kaolinite in an opalized zone, and a limonitic argillized margin. That sequence of core, middle and rim is the template explorers look for in the field. Cuprite is an alteration-mapping benchmark; this guide does not present it as a mine.

Patagonia, Argentina. The two ASTER studies named above worked at epithermal gold areas. They did not find gold with the image. They produced mineral maps that told geologists where altered rock lay, so that sampling could follow.

Chocolate Mountains, California. The same review cites a study that applied matched filtering to Hyperion and ASTER data for gold-associated rock types, and found that Hyperion gave better accuracy than ASTER, although the rock information from ASTER was mostly similar and ASTER’s wider coverage suited regional mapping. Both sensors were useful.

In all three cases the image narrowed the search. None replaced sampling.

▶ How Landsat, Sentinel-2, ASTER and radar help find gold even though satellites cannot see it · 4:25

How to use satellite images on your own ground in six steps

1. Draw the boundary. Load your licence or claim polygon (KML, KMZ or shapefile) so every layer is cut to the same area.
2. Pick clear dates without cloud or snow. Dry-season scenes show bare rock. Check several years, because one scene may carry haze or shadow.
3. Make an iron-oxide layer and a clay layer. Use a visible-to-near-infrared ratio for rust and a SWIR ratio for clay, then compare them to a geology map.
4. Mask what is not rock. Remove water, vegetation, cloud shadow, roads and mine dumps, because they fake anomalies.
5. Add structure. Draw faults and ridges from an elevation model and mark where alteration crosses them.
6. Rank, then check on the ground. Take rock chips and soil samples at the best few zones and send them to a laboratory. Only assays can show gold.

Infographic

From polygon to sampling targets

1. 1DrawLoad the licence polygon.
2. 2Choose datesPick clear, dry-season scenes.
3. 3MapMake iron-oxide and clay layers.
4. 4MaskRemove water, plants and dumps.
5. 5CheckSample the top-ranked zones.

If you plan to process ore yourself, see our page on processing gold ore at home; it covers what happens after rock leaves the ground. Satellite screening takes days rather than a field season, but it only chooses where to look.

What the pictures cannot tell you

Be exact about the limits before you pay for anything.

• They see the surface only. Optical sensors read light reflected from the ground surface. Gold under transported cover, such as sand, gravel or volcanic ash, leaves no optical signal.
• Vegetation hides rock. Dense forest or crops cover alteration minerals. In vegetated country explorers fall back on radar, elevation data and ground chemistry.
• False positives are common. Rust forms on barren rocks too, and clay forms in weathered soils with no gold at all. An anomaly is a place to sample, not a deposit.
• ASTER’s SWIR data stop in 2008. The NASA data centre states that ASTER SWIR data collected since April 2008 show saturation and striping, and have not been included in the standard product since 1 April 2008. The archive before 2008 remains useful; for later dates use Sentinel-2 or Landsat SWIR bands and hyperspectral sensors. The same page records a gap in ASTER acquisitions after a power fault on 28 November 2024, with visible and near-infrared acquisitions resuming on 18 January 2025.
• Satellite results are exploration targets. They are not mineral resources or reserves, and they do not meet reporting codes such as JORC or NI 43-101 without drilling and a qualified person’s sign-off.

Used honestly, the method still saves money. It removes poor ground early, so drill metres go to the best rock.

▶ The honest answer on whether a satellite can detect gold · 2:00

Where imagery fits in a gold exploration programme

For a licence owner, the first use is screening: look at the whole licence before walking it, and rank areas by alteration and structure. For a consultant, it adds a regional layer to a geology report. For an investor, it is a way to ask whether a company’s target sits on altered rock that shows from space, not a verdict on the project. Farmonaut produces satellite mineral reports from licence polygons using multispectral or hyperspectral satellite data; the output is a target map, not a resource estimate.

Combine the image layer with geology maps, geochemistry and geophysics. Where two or three independent layers agree on the same spot, the target is stronger than any single layer.

Frequently asked questions

Can you see gold from a satellite?

No. Gold occurs at grams per tonne, well below what a 10 to 30 metre pixel can show. Satellites map the alteration minerals, rust and structure around gold systems, and those clues guide sampling.

What is the best satellite for finding gold?

There is no single best. Sentinel-2 gives frequent 10 and 20 metre coverage across a 290 kilometre swath. Landsat 8 and 9 add a long record at 30 metres. ASTER’s archive from before 2008 has six SWIR bands at 30 metres. Hyperspectral data such as EnMAP separate clay minerals best. Most teams combine several.

How do satellite imagery and gold exploration fit together?

Imagery is a reconnaissance tool. It maps alteration, structure and terrain over a large area quickly, so field crews and drill rigs go to a few high-ranked zones instead of the whole licence.

Can Google Earth show gold deposits?

Google Earth shows true-colour pictures, which are only the visible bands. You can spot mine workings, roads and bright rust-stained hills, but you cannot separate clay from soil or measure alteration without SWIR bands.

What resolution do I need to find a gold deposit?

Match the pixel to the target. Alteration zones tens to hundreds of metres across suit 10 to 30 metre pixels. Single veins need drone imagery or ground mapping. The calculator on this page shows pixel counts for your area.

Is Landsat or Sentinel-2 better for gold exploration?

Sentinel-2 has sharper pixels (10 and 20 metres against 30) and visits every 5 days. Landsat has a record that stretches back decades and thermal bands. Use Sentinel-2 for current detail and Landsat when you need history.

Can satellites find gold under the ground?

Not directly. Optical sensors read the surface. Under cover, geophysics and drilling are needed. Satellite data can still show the structures and alteration trends that continue under the cover.

Can I use satellite images to find gold on my own property?

You can screen the property for alteration, but ownership and permits matter more. Get the surface owner’s permission, check mineral rights and the licence rules, and never use mercury. Sample before you spend on equipment.

Does ASTER still provide shortwave infrared data?

The archive before April 2008 does. Data after that date lost usable SWIR bands to detector problems, according to NASA’s data centre. Many teams now use Sentinel-2 and Landsat SWIR bands instead.

How much does a satellite gold survey cost?

Public data from Landsat and Sentinel-2 can be downloaded without a licence fee, so the main cost is analysis time. Hyperspectral or high-resolution scenes and expert interpretation are priced per area, so request a quote for your polygon.