In May 2025, a miner in New South Wales was injured by a 2–3 tonne slab of rock that detached from an unsupported underground development face. The incident — documented in Safety Alert SA25-02 issued by the NSW Resources Regulator — was not triggered by seismicity, nor was it unpredictable. Scaling had been done. The rock was still there. It still fell.
This wasn’t a statistical anomaly. It was a pattern we know well. Ground falls remain the leading cause of fatalities and serious injuries in underground mining worldwide. And yet, the level of industry action — in budget, in modeling, in monitoring — doesn’t reflect that reality.
What the Numbers Show – and What They Don’t
Public safety data from agencies like MSHA, WorkSafeBC, and SERNAGEOMIN consistently report that falls of ground represent a top category for lost-time injuries and fatalities underground. The NSW incident is just one of many in 2025 alone. The true number of near-misses — undocumented or unofficially logged — is almost certainly higher.
And that’s the point: most rock falls don’t kill. But they could. They just haven’t — yet. And that has created a culture where ground control is still seen as a compliance item, not a critical control.
The Failure Isn’t Technological. It’s Strategic.
Yes, we have mesh, bolts, yielding arches, and shotcrete. Yes, we’ve improved digital design tools and introduced LiDAR scans in many operations. But most underground mines still treat unsupported or legacy headings as stable unless proven otherwise.
Predictive modeling tools exist. A 2025 study in Scientific Reports by Khoshmagham et al. modeled time-dependent deterioration of tunnel roofs due to mining-induced stress and moisture. Their results show measurable roof deformation over time — a creeping failure that doesn’t trigger until long after excavation.
But modeling like this is rarely part of operational planning. It lives in research labs or high-profile Tier 1 mines. For most mid-tier or regional operators, the standard risk model is still visual inspection and bolt pattern conformity.
Seeing What Fell – Without Entering the Void
One area of promise is remote reconstruction. Research from Asteriou et al. (2025) demonstrates how photogrammetry and structure-from-motion techniques — using drone imagery or even crowdsourced photography — can reconstruct a 3D model of a collapse zone without requiring re-entry.
Though currently applied in surface mining and civil rock slopes, the potential exists for underground applications, particularly in block caves or large stopes where entry post-collapse is unsafe or impossible.
But again, these tools are not standard. Their usage is ad hoc — not part of most jurisdictions’ post-incident protocols.
Real Case: What LKAB Is Doing Differently
One verifiable example of proactive monitoring is Sweden’s LKAB, which uses real-time deformation radar in its Kiruna underground iron ore mine to detect movement in the hanging wall and footwall zones. Their published case studies in collaboration with GroundProbe show how radar velocity monitoring has been used to predict deformation events in advance — allowing for withdrawal of equipment and personnel.
This isn’t hypothetical. It’s operational. And it shows what’s possible.
Skillings Commentary
“Underground rock falls aren’t random. They’re what happens when stress, geology, and human planning intersect — and no one’s watching close enough. We know the risk. We know the science. The only thing we haven’t figured out is how to treat prediction as prevention — not just a paper exercise.”
— Skillings Mining Review Editorial Board
Recommendations: What the Industry Can Do Now
| Priority | Recommendation |
| Publish all rockfall data, not just fatalities | Transparency drives accountability. Hidden near-misses help no one. |
| Fund predictive modeling as part of CAPEX | Not just mesh and bolts — fund the software and expertise to run stress-time models. |
| Audit unsupported headings | Re-inspect all legacy crosscuts, backfilled stopes, and intersections — especially in aging mines. |
| Use remote sensing post-collapse | Drones and image-based reconstructions can aid incident investigations without endangering teams. |
| Tie executive incentives to ground performance | Not just “safe days” — but geotechnical audit results, predictive tool adoption, and near-miss reduction. |
Conclusion
Underground mine rock falls are not yesterday’s problem. They are today’s quietest risk. As the NSW incident showed, even in regulated, developed settings, it takes only one wedge, one decision, one oversight.
The tools exist. The science exists. The examples exist.
What’s missing is the mandate.
Skillings will continue to track verified incidents, policy changes, and geotechnical innovations — and report not just what happened, but what should have happened sooner.
