The Scarce Input in Mine Safety Is Judgment

At about 8:41 pm on 17 November 2016, a slope above an undisclosed copper open-pit mine gave way and killed 16 workers. The mine's own monitoring system never saw it coming. The failure mostly involved natural slope above the crest, outside the system's field of view, according to Carlà et al. in a 2019 Scientific Reports paper. A satellite had been looking at that ground the whole time.

That is the case for a thesis that mine operators and their investors have yet to price properly. Ground-movement data is now free, global and repeated every few days. The warning signs, in hindsight, were there. What is scarce is interpretation: knowing which of the many moving patches in a radar scene is the one that matters. Operators who skip that step are not saving money. They are carrying a risk they could have measured.

The data is no longer the constraint

Carlà et al. note that Sentinel-1's interferometric products give worldwide coverage and are freely distributed. For the 2016 pit, they report up to 30.2 mm of line-of-sight displacement in the period 9 to 15 November 2016, with the last image acquired on 15 November 2016. Their conclusion is that the event could have been located several days or weeks in advance.

The supply is about to widen. NASA says the initial public release of calibrated NISAR L-band data began July 20, 2026, covering observations acquired on or after June 17, 2026, with all products free through ASF DAAC and NASA Earthdata. NISAR images the same locations repeatedly on a 12-day cycle, even through cloud, according to NASA. Its L-band radar, built by NASA JPL, uses a 10-inch (25-centimeter) wavelength and penetrates vegetation canopies. Rain, cloud and scrub are the usual excuses for gaps in coverage of remote mines, and this instrument is built to ignore them. ASF also exposes the data through APIs and the asf_search Python package, for large-scale automated processing.

The evidence on past failures is stark. A April 2026 study by Wang et al. in the International Journal of Applied Earth Observation and Geoinformation analysed ten cases across surface mining complexes, covering open pits, tailings dams, heap leaches and waste dumps. Six involved fatalities. All of the failure sites showed precursory movement months to years before the event, the authors report, using open-access Sentinel-1 scenes from the Alaska Satellite Facility.

The stakes are not evenly spread. Wang et al. counted 722 slope failures in the Scopus-indexed literature, with 2,111 reported fatalities. Tailings dams numbered only 86 in the same literature but accounted for 1,730 reported fatalities. Ground-based systems, the authors add, are often limited by spatial coverage, operating cost and the need to install and maintain hardware.

Creep is everywhere

The obvious objection is that hindsight makes every failure look obvious. Dave Petley, writing on Eos about the Wang paper, makes it forcefully. Only a small proportion of the deformation visible in InSAR imagery of mines will lead to a slope failure with high loss or damage, he says, and the paper's own limits are plain: "Reliably detecting the precursory deformation associated with these particular events remains a major challenge."

The Cadia gold mine in New South Wales shows how misleading raw motion can be. Petley notes that a large cluster of post-failure deformation northeast of the site was planned, because the operation had chosen block caving with break-through to the surface. Wang et al. also report post-failure hotspots in several mining areas, some moving faster than they did before the failures. Speed alone, in other words, does not separate danger from design.

Cadia's tailings failure on 9 March 2018 also exposes the technology's limits. The last satellite acquisition before the event was on 26 February 2018, on a 12-day revisit. A July 2026 Scientific Reports paper by Bayaraa, Sheil and Rossi, published 20 July 2026, says ground instrumentation had been removed in the period before the failure, leaving InSAR as the only source of insight. It adds that geotechnical model predictions diverged from InSAR measurements after buttress construction, likely because InSAR captures rapid deformation poorly.

None of this weakens the thesis; it sharpens it. If most movement is harmless and some is lethal, a free feed does nothing for an operator who cannot sort the two. The Cadia case is an argument for fusing radar with geomechanical models and, in the Bayaraa team's approach, deep learning. It is also an argument against removing the ground instruments.

Where the money goes

New Space Economy makes the commercial point about satellites generally. Value shifts toward processing and analytics because customers need ranked targets, confidence estimates and geological interpretation, not an unprocessed scene. An EUSPA-supported project involving Eramet is already using Sentinel-1 to pick out millimeter-scale ground movement at mining areas, which the publication describes as extending the service chain from exploration into operational safety.

Investors do not need a radar to know what a safety failure costs. MINING.COM's Top 50 ranking says BHP suffered the largest dollar loss in the ranking, $26.4 billion or 11%, after a worker was killed at Escondida on September 23 and the world's largest copper mine suspended operations. The ranking's 50 most valuable mining stocks closed September at $2.26 trillion, the second-largest monthly decline in its history. A single stoppage at a flagship asset can move a valuation by more than most mines are worth. An operator that cannot show it reads its slopes and dams will find that discount arriving first in the cost of capital and in the insurance conversation.

What follows

The free data will also reach plaintiffs, regulators and short-sellers, none of whom need a mine's permission to look at it. Once every pit's deformation history is public, an unexplained acceleration becomes something anyone can point to, and "no apparent warning signs" stops working as a defence.

The best-run operators will treat radar as a trigger for work: put engineers on the anomaly, check it against the ground sensors, and document the decision either way. Others will treat it as someone else's dashboard. New Space Economy's line on exploration, that "A satellite does not replace the drill rig," applies in a pit as well. The satellite points. Someone still has to go and look.