Satellite Analytics: Improving Monitoring of the Mining Industry

Every generation of mining technology promises the same thing: do more, spend less, cut risk. One problem resisted all of it: continuously understanding what’s happening across vast, remote mine footprints. That problem is now solved, and it’s solved from roughly 500 kilometers up.

Spectral analysis is offered by various satellite imagery platforms like EOSDA, where self-developed algorithms transform orbital data into clear and understood environmental and geological insights. And that is the main leap forward: even a student without technical education can have a look at the latest mining satellite imagery available online and participate in mine monitoring. So do companies. They don’t have to run quarterly site visits and biennial airborne surveys with a risk of missing something important at a high cost.

Why Ground-Based Monitoring Fell Behind And What is Satellite Analytics

Satellite analytics is the pipeline that turns a raw orbital image into information that a mine manager can act on. A satellite hands you a grid of reflectance values across dozens of spectral bands, and then algorithms correct the raw image for atmospheric haze and sun angle so a scene from March and a scene from October are actually comparable. Then corrected data is run through spectral indices and machine learning models trained to recognize a specific signature — chlorophyll stress or iron oxide staining. After that, the result is packaged as a number, a flag, or a map layer that can be used by a non-specialist without a remote sensing degree.

November 5, 2015 Samarco Mineração, a joint venture between Vale and BHP Billiton, operated the Fundão tailings dam near Bento Rodrigues in Minas Gerais, Brazil.

Satellite mining data and satellite imagery for mining are worthless as raw pixels — their value shows up only after they clear that analytical layer, coming out the other end as a vegetation stress score, a water-quality flag, or a millimeter-scale displacement reading.

Ground-based monitoring never had that layer, and it never had continuous coverage either. The gaps are specific:

• Site inspections depend on someone physically reaching the location, which access, weather, and safety rules can delay for weeks.
• Airborne geophysical surveys deliver sharp resolution but get mobilized on multi-year cycles — a plane and sensor package isn’t an easy thing.
• Ground sensor networks report continuously, but only from the exact spot where the sensor sits. If you work with tens of square kilometers, you will need a myriad of those sensors.
• Lab geochemistry needs a physical sample collected, shipped, and processed before a number comes back — days at best, longer if the lab’s backed up.

The Role of Satellite Data in Modern Resource Exploration

The role of satellite analytics is narrow and specific: data screens a continent’s worth of ground for signs worth drilling, before anyone spends a dollar mobilizing a field crew. It doesn’t find deposits. It tells you where not to bother looking.

Exploration has always run on a brutal ratio — assess a huge area to justify drilling a tiny fraction of it. For decades that ratio got worked through ground crews and airborne surveys, one region at a time, at a pace measured in field seasons. Today a geologist can rank hundreds of blocks across an entire continent from a laptop, in the time it used to take to charter a single helicopter survey.

What the imagery actually picks out:

• Hydrothermal alteration minerals — the clay signatures left behind where mineralizing fluids once moved through rock.
• Iron oxide caps — the rusty surface expression of a sulfide body that’s been weathering for millions of years.
• Structural lineaments — faults and fracture zones that controlled where those fluids could travel in the first place.
• Vegetation stress in near-infrared bands — plants struggling in metal-rich soil, visible before the stress shows up to the naked eye.

What the Spectrum Actually Reveals

Every surface on a mine site is quietly giving away information. Rock, water, live vegetation, a chemical crust forming on a tailings pond — each reflects sunlight back in a pattern specific to what it’s made of. Human eyes catch almost none of it. Sensors aren’t limited that way, and which one you need depends entirely on what you’re trying to see:

• Multispectral — 4 to 12 discrete bands, 3 to 30 meter resolution. These sensors cover a whole site cheaply, flag vegetation stress, broad mineralogy, and water quality shifts.
• Hyperspectral — 100 to 400+ continuous bands, 5 to 30 meter resolution. The specialist tool for detailed mineral mapping and alteration zone identification.
• RGB (standard optical) — 3 bands, sub-meter resolution. Built for visual change detection and infrastructure mapping — the sharpest imagery here, but the least chemically informative.
• Thermal infrared — variable bands, 30 to 100-meter resolution. The one type that catches seepage at a tailings toe running warmer or wetter than the soil around it, or subsurface heat anomalies none of the optical bands would ever pick up.

What Happens in the Gap Between Inspections 

Two dates show what failure costs:

November 5, 2015
Samarco Mineração, a joint venture between Vale and BHP Billiton, operated the Fundão tailings dam near Bento Rodrigues in Minas Gerais, Brazil. When it failed, it released roughly 40 million cubic meters of iron ore tailings in a mudflow that destroyed the village outright, killed 19 people, and traveled more than 600 kilometers down the Rio Doce to reach the Atlantic, contaminating the water supply for dozens of municipalities along the way. 

January 25, 2019
Vale’s Feijão mine near Brumadinho, also in Minas Gerais, held waste behind a structure known as Dam I, built with the same upstream method used at Fundão. It failed in seconds around lunchtime, sending a wave of mining waste straight through the mine’s own cafeteria and administrative offices before continuing downstream, and killed 270 people — one of the deadliest mining disasters on record. 

Wrapping Up

In both cases, InSAR radar could have flagged the millimeter-scale wall deformation building at Fundão and Feijão in the weeks before failure, and infrared imagery could have caught seepage at the dam toe before it reached the surface — both signs that would’ve shown up in satellite passes days to weeks ahead of the collapse, not just at the next scheduled inspection. 

Mines that treat this as a compliance checkbox will keep discovering their liabilities after the fact. Mines that treat satellite mine monitoring as infrastructure — as permanent as a haul road — will be the ones still financeable when the next tailings dam fails somewhere else.

Energies Media Staff