The hazard is far from trivial. Abandoned underground mines worldwide have produced some of the most dramatic geotechnical failures on record. In Pennsylvania, USA, a continuous discontinuity in a limestone mine triggered the failure of 35 supporting pillars, collapsing the overburden into a 3-hectare subsidence zone and generating an airblast strong enough to propel three miners roughly 15 metres out of the entrance portal. In Tennessee, the successive collapse of 40 pillars over 5.7 hectares produced a seismic event of magnitude 3.1, a destructive airblast, and a sinkhole affecting 3.6 hectares beneath 122 metres of overburden. A 2005 cascade pillar failure in a Chinese gypsum mine killed 37 people and injured 38, destroying 88 residential units; the deadly airblast underground, with a calculated discharge of nearly 187,000 cubic metres, lasted 67 seconds, and the pressure waves reflected repeatedly through the workings, striking miners multiple times.

The consequences can extend far beyond the immediate collapse. Infrasound waves from mine collapses in South Korea were detected more than 200 kilometres away. In China’s Datong coalfield, toxic carbon monoxide migrating through ground fissures after sudden collapses killed four people between 1993 and 2005, required medical treatment for 546 others, and forced the relocation of approximately 9,000 residents. Salt mines fail in their own distinctive fashion: at Solotvyno in Ukraine, uncontrolled water inflow dissolved the salt dome itself, creating suffusion sinkholes 150 to 230 metres in diameter, while a Romanian salt mine collapse after heavy rainfall in 2001 contaminated the Olt River with brine. Even seismic events can be misread—researchers initially attributed a magnitude 2.4 signal at Lo Tacón in Spain to an earthquake, before demonstrating that the rapid collapse of roughly 20,000 cubic metres of rock had generated the signal itself.

Czechia’s slate mining district offers a more intimate scale, but the underlying physics is the same. The Lower Jeseník Upland, part of the Central European Variscides, is built from Lower Carboniferous slates and greywackes deposited as deep-water flysch sediments. Slate mining there is documented from 1776, and expanded dramatically after Emperor Joseph II’s Fire Patent restricted wooden roofing, making fire-resistant slate a commodity. By the Industrial Revolution, more than a hundred quarries and mines supplied the entire Austro-Hungarian Empire. The expulsion of the German-speaking mining community after the Second World War stripped the region of experienced owners, skilled workers, and crucially its mine maps, and underground slate mining finally ceased after 1989. Today, 112 abandoned underground slate mines have been documented in the region, many with unknown extents and no reliable plans.

The first case study, the Kunz Mine, is a six-storey excavation established in 1850 on a steep valley slope, extending roughly 80 metres below the surface—about 50 metres beneath the valley floor and below the groundwater table. Its lower four storeys have been flooded since the Second World War and are accessible only to sport divers. The mine was cut into a complex flexural structure whose steeply dipping slate layers are affected by near-surface creep, and the researchers classify the resulting failure as flexural toppling, a mechanism typical of closely fissile rocks such as slate, in which layers bend toward excavated voids after detachment along bedding planes. Excavation released stress in the rock mass, producing buckling and separation cavities, while recurrent rockfalls released during freeze–thaw cycles along a cubic network of fissures now litter the accessible passages with debris.

What makes the Kunz Mine scientifically exceptional is the team’s pioneering use of dendrogeomorphology—tree-ring analysis—to date the surface expression of an underground failure. The researchers sampled 23 Norway spruce trees showing signs of disturbance, extracting paired increment cores and cross-sections from roots intersecting tension cracks. Cross-dating against a reference chronology from 20 undisturbed trees revealed 44 growth disturbances between 1926 and 2017, including compression wood formation, growth suppressions, root exposure, and an abrasion scar. The analysis identified at least four distinct periods of gravitational activity: 1942, 1973, 1997, and 2008–2009, with exposed roots dating a crack-opening episode to 1970–1973. Notably, the 1997 activity coincided with a year of exceptionally high precipitation in northeastern Czechia, but no comparable rainfall anomalies explained the other event years, suggesting that microclimatic factors rather than precipitation alone drive the instability.

The second case, the Woodboys Mine, captured a rockfall in progress. Sometime in 2022, the toppling of an overhanging wall in a terminal room measuring 28 by 11 by 9 metres released a slab roughly 0.4 metres thick with planar dimensions of 15 by 12 metres—an estimated 72 cubic metres of rock. The layers landed largely intact within 8 metres of the wall base before fragmenting on impact, scattering debris up to 10 metres away. The freshly exposed scarp told the story of the failure’s long preparation: ochre-stained surfaces marked bedding planes where iron oxyhydroxides had developed and the wall remained mechanically intact, while dark grey coatings and flower-like aggregates of gypsum crystals, roughly half a centimetre across, documented a pre-existing loosened cavity. Scanning electron microscopy with energy-dispersive spectroscopy confirmed the crystals as gypsum, most likely produced by the weathering of pyrite in the slate—a chemical clock ticking for years before the collapse.

The third site, the Soví kámen Mine, is one of the largest abandoned slate mines in the region, a multi-storey system spanning about 120 metres of vertical extent. Here, a sinkhole roughly 4 metres deep has developed in a large room, funnelling spoil and collapsed blocks into an unknown void below, while an underground stream that once flowed through the entire mining level now sinks into the debris just metres from the sinkhole margin and actively propagates erosion upstream. The researchers propose several scenarios: a former interconnecting chimney used by miners for gravity drainage may have been progressively destabilized by fluvial erosion, or a major ceiling collapse in the weakened, fault-fractured rock may have breached the rock divide between levels, redirecting the stream into the new connection. Distinct upward airflow from the sinkhole base during the cold season confirms that air and water still circulate freely through the seemingly isolated workings.

This circulation, the authors argue, is the missing ingredient in most collapse analyses. Because mine entrances sit at different altitudes, the systems are microclimatically dynamic: cold air descends and becomes trapped, freezing dripping water into icicles and frost, while warmer air masses interact with humid surfaces to produce condensation, fog, and dripping. At the Devil’s Mouth Shaft, cold air pooling at the base drives frost weathering through microgelivation and ice segregation, and heavy forestry vehicles travel directly over its weathered roof only a few metres above. Slate itself is chemically vulnerable: oxidation of iron sulphides generates sulphuric acid that accelerates alteration, while gypsification produces crystals that exert crystallization pressure within microfractures, progressively delaminating the rock. Hydric expansion and contraction of clay minerals, repeated hydration cycles of gypsum, and post-mining stress release along cleavage planes all conspire to weaken the workings long before any visible failure.

Crucially, the study shows that collapses rarely arrive unannounced. Geotechnical precursors—minor rockfalls, crack opening, acoustic emissions, changes in water inflow, tension cracks, floor heave—may persist for years before major failure, and many are identifiable in the Czech mines today. Yet accurate prediction of delayed sudden collapse remains impossible, and the mine environment is never in equilibrium: like a system repeatedly seeking balance, each collapse redistributes stress and drives further instability in cascading cycles. The researchers conclude that hazard assessment must therefore consider the broader spatiotemporal relationships among landscape elements, overlapping mine levels, and geological structure, and they call for individualized rather than standardized management. With recreational cabins already perched above the collapsing Kunz shaft and construction on undermined ground a persistent temptation, the authors urge that abandoned mine entrances be mapped together with their underground extents—and that loading the fragile rock above them be avoided before the landscape settles the matter on its own terms.

Subject of Research: Gravity-driven collapse mechanisms in abandoned underground slate mines in Czechia

Article Title: Collapses of legacy mines in a spatial and evolutionary context

Article References: Lenart, J., Kašing, M., Tichavský, R., Schuchová, K., & Šilhán, K. (2026). Collapses of legacy mines in a spatial and evolutionary context. Environmental Earth Sciences, 85(15), Article 385. https://doi.org/10.1007/s12665-026-13110-0

Image Credits: AI Generated

DOI: 10.1007/s12665-026-13110-0

Keywords: abandoned mines, slate mining, flexural toppling, rockfall, sinkhole, dendrogeomorphology, weathering, gypsification, pyrite oxidation, geohazards, Czechia, Environmental Earth Sciences

The study, led by R. C. Bussard of Penn State’s Department of Geosciences and published in Environmental Earth Sciences, focused on the Blue and White Golf Course at Penn State’s State College campus. The course traces its origins to 1924, when a predecessor known as the Campus Course was built on land that today hosts two eighteen-hole layouts, the Blue Course and the White Course. The entire region sits atop dolomite and limestone bedrock in Pennsylvania’s Nittany Valley, part of the broader Ridge and Valley Province. These carbonate rocks dissolve slowly in weakly acidic water, carving out the caves, springs, and sinkholes that define karst terrain. In the mid-2000s, a large sinkhole opened east of the Blue Course’s Hole 15, a reminder that the hazard is not hypothetical.

To map the hidden threat, the researchers combined two very different satellite perspectives. They processed 250 scenes from the Sentinel-1 C-band radar satellite spanning 2017 to 2025 using interferometric synthetic aperture radar, or InSAR, a technique that detects millimeter-scale ground movement by comparing the phase of radar signals from repeat passes. After rigorous quality filtering that retained 1,002 interferograms across 232 dates, they produced an average vertical velocity map validated against a continuous GNSS station. The result was striking: broad subsidence across the golf course, with rates exceeding 1 centimeter per year in the south and southwest portion of the Blue Course, while nearby campus infrastructure to the east remained essentially stable.

Complementing the radar data, the team analyzed 63 cloud-free Landsat 8 and 9 scenes from the same eight-year window to calculate the Normalized Differenced Moisture Index, a measure of vegetation and soil moisture derived from near-infrared and shortwave-infrared bands. The golf course showed consistently higher NDMI values than the surrounding landscape, frequently exceeding 0.15 along the fairways, indicating persistently wetter near-surface conditions than the area around it. The researchers also checked whether soil moisture fluctuations could be contaminating the InSAR signal itself, a known artifact in radar interferometry, and found no strong correlation between moisture variability and phase noise, giving them confidence that the deformation was real.

The field component of the study brought the investigation underground. The team installed soil pore water samplers at four locations within the golf course and two control sites outside its boundaries, each placed roughly 30 centimeters below the surface to avoid the root zone. Deployed for 72 hours during a late-July storm that delivered nearly 4 centimeters of rain, the samplers yielded water samples whose pH and conductivity were measured within an hour of collection. The findings were telling: pore water beneath the golf course averaged a pH of 6.50, noticeably more acidic than the 7.15 measured at control sites, while conductivity values were broadly similar across all samples.

That modest difference in acidity carries significant implications. Weakly acidic water is precisely what dissolves carbonate rock. Rainfall percolating through soil picks up carbon dioxide from the atmosphere and from soil respiration, forming carbonic acid that eats away at limestone and dolomite. The researchers suggest that fertilizer application and elevated soil carbon dioxide at the course may be enhancing this natural acidification, creating thermodynamic conditions more favorable for calcite dissolution beneath the fairways than beneath the surrounding suburbs. They are careful, however, to note the limitations: the sample size was small, and carbonate saturation cannot be determined from pH alone without alkalinity and dissolved inorganic carbon measurements.

What makes the situation particularly interesting is what the data ruled out. Many documented sinkhole crises worldwide stem from large-scale groundwater extraction, such as the water table declines exceeding 50 meters in Saudi Arabia’s Jouf region between 2009 and 2011, or the acceleration of sinkhole formation in Turkey’s Konya Basin over two decades. Extreme recharge events can also trigger collapse, as happened when Tropical Storm Debby’s rains spawned more than 200 sinkholes across Florida. Yet at the Blue and White Golf Course, groundwater records from a nearby USGS observation well show repeated seasonal and interannual fluctuations rather than sustained regional decline, and subsidence timing does not correlate with rainfall patterns either. The ground is sinking without any obvious regional driver.

The answer, the researchers argue, lies in local factors. When they compared ground deformation at pixels containing mapped karst features with pixels lacking them, the mean subsidence rate was substantially higher where features existed: 0.75 centimeters per year versus 0.41 centimeters per year. Pre-existing cavities and preferential recharge pathways appear to concentrate deformation along specific corridors, particularly where a cluster of karst features sits southwest of the course near a drainage ditch. Meanwhile, the estimated 75,000 cubic meters of water a typical golf course applies annually through irrigation may be creating localized recharge hotspots that feed water into those vulnerabilities even in the absence of regional decline or extreme storms.

The time series data add further nuance. Some sample sites showed step-wise deformation, with bursts of rapid subsidence interspersed with periods of stability, while others moved linearly or barely at all over the eight-year record. Site one, which was among the most acidic and showed the most variable NDMI, displayed the most erratic displacement history. Control sites outside the course showed minimal deformation despite sitting on similar geology, underscoring that the golf course’s management practices, rather than its underlying rock alone, distinguish it from its surroundings.

The researchers caution that shallow processes such as mechanical compaction from equipment traffic and organic matter decomposition cannot be fully excluded as contributors. But the convergence of evidence, wetter soils, more acidic pore water, faster subsidence near known karst features, and stability everywhere else, paints a coherent picture in which modest but repeated land-use modifications interact with pre-existing karst susceptibility to produce localized ground loss. Given that the region near Hole 15 already produced a sinkhole once, the southwest Blue Course could see more in the future, particularly if climate change brings heavier rainfall. For a golf industry worth $102 billion in the United States alone, the message is clear: understanding which drivers of sinkhole formation can be mitigated through smarter irrigation, fertilization, and monitoring may determine whether courses stay open for another six decades or become the next cautionary headline.

Subject of Research: Sinkhole hazard assessment at a karst golf course using remote sensing and field geochemistry

Article Title: Assessing sinkhole hazard at the blue and white golf course, state college, using a combined remote-sensing and field methods approach

Article References: Bussard, R. C., Housego, R., Wauthier, C., Marqeuz, M., & Miller, J. (2026). Assessing sinkhole hazard at the blue and white golf course, state college, using a combined remote-sensing and field methods approach. Environmental Earth Sciences, 85(15), Article 403. https://doi.org/10.1007/s12665-026-13134-6

Image Credits: AI Generated

DOI: 10.1007/s12665-026-13134-6

Keywords: sinkhole, karst, InSAR, remote sensing, groundwater, carbonate dissolution, golf course, Landsat, soil moisture, pore water, subsidence, Pennsylvania