The quantity that the database tracks, known to geoscientists as SHmax, is the compass direction in which the greatest horizontal compression pushes through the crust at any given location. That single number matters enormously. It governs which faults are likely to slip in an earthquake, how tunnels and boreholes deform, whether deep geothermal reservoirs can be fractured safely, and where carbon dioxide or hydrogen can be stored underground without compromising the rock’s integrity. As humanity prepares to use the subsurface on an unprecedented scale, from geo-energy storage to radioactive waste repositories planned at depths of 400 to 1,000 metres, knowing the direction of maximum compression has become a practical necessity rather than an academic curiosity.

The story of measuring this stress stretches back nearly a century. The first crustal stress data were collected in the 1930s using surface relief methods, followed in the 1950s by flat jack and borehole relief techniques. The 1970s brought hydraulic fracturing, which measures the magnitude of the least principal stress, and the late 1970s introduced borehole breakouts, the tell-tale elongations that a drill bit leaves in rock when the surrounding compression is uneven. Meanwhile, the expansion of global seismological networks in the 1960s allowed earthquake focal mechanisms to serve as stress indicators for the deeper crust. These techniques converged in 1986, when the World Stress Map project was launched as a task force of the International Lithosphere Program, with a bold initial goal: to test the mid-1960s hypothesis that plate tectonic forces alone control the orientation of maximum horizontal stress in the crust.

For a while, the hypothesis held up. The first database release in 1989, containing 3,574 quality-ranked records, broadly supported the idea that the push and pull at plate boundaries dictates the stress compass. But the 1992 release, with roughly 7,300 records, already hinted at complications: lateral density and strength contrasts within the plates were producing stress patterns with wavelengths of several hundred kilometres. The 2005 release, comprising 15,969 records, confirmed those second-order patterns and revealed the first examples of stress rotations on smaller scales. Now, with the 2025 release, the expansion has been dramatic. Much of the growth comes from integrating the global compilation of earthquake focal mechanisms from the International Seismological Centre, along with more than 4,000 new records derived from boreholes around the world.

Integrating such wildly different measurement techniques is the database’s central technical challenge, and it is solved through a rigorous quality-ranking scheme. The 2025 edition employs eight established stress indicators, from overcoring, which samples rock volumes of roughly a cubic metre, to single earthquake focal mechanisms, which sample volumes on the order of a billion cubic metres. Each record is assigned a quality class from A to E, with A-quality data accurate to within plus or minus 15 degrees, B-quality to 20 degrees, C-quality to 25 degrees and D-quality to 40 degrees. Only A, B and C records are considered reliable for interpreting geodynamic processes. The 2025 scheme refines the older rules with hard-edged criteria suitable for automated Python-based assignment in the project’s new database infrastructure, called MaRS. It introduces a new X-quality class for records with missing information, rarely used indicators or methods not yet established, and tightens requirements for borehole breakouts, drilling-induced tensile fractures, hydraulic fracturing tests and overcoring measurements.

Because stress data are heavily clustered around plate boundaries and in sedimentary basins, raw records alone cannot reveal global patterns. To overcome this, the team computed mean SHmax orientations on regular global grids of 2, 1, 0.5 and 0.2 degrees, using search radii between 50 and 500 kilometres and producing 13 distinct datasets that filter the stress pattern at different wavelengths. The averaging relies on circular statistics for axial data, weighting each record by its quality class and by its inverse distance from the grid point, with a minimum distance threshold to prevent very nearby records from being overrepresented. A minimum of five records is required within each search radius, and all must lie on the same tectonic plate as the grid point, according to the global plate boundary model PB2002. The result is a family of stress maps that can zoom from continental sweeps down to regional detail.

What those maps reveal is striking. In many intraplate regions, where data density has increased most dramatically, the old hypothesis that plate boundary forces and relative plate motion primarily control the stress orientation needs to be revised. The maximum horizontal stress rotates in some areas by more than 50 degrees over spatial scales of just 50 to 500 kilometres. In the foreland of the Alps, the stress direction swings by roughly 50 degrees, from north-south in the east to about N130°E in the western Alps, driven by the excess gravitational potential energy of the high topography, which pushes the crust outward perpendicular to the mountain belt’s strike. Comparable patterns appear in the greater Himalaya region and the Rocky Mountains of Canada. Even more dramatic are rotations exceeding 50 degrees over distances of less than 100 kilometres in eastern Australia.

Australia’s Bowen Basin offers a particularly instructive natural experiment. An exceptionally dense dataset from 680 vertical boreholes spread over an area of 300 by 100 kilometres in the basin’s northern part shows no lateral or vertical variation in stress orientation at all, despite the presence of a fault network and lateral stiffness contrasts; the mean orientation is N18°E with a standard deviation of plus or minus 18 degrees. Yet further south, in the southern Bowen Basin and the overlying Surat Basin, the stress orientation rotates by up to 60 degrees within 100 kilometres. Similar rotations in the greater Texas-New Mexico region occur without significant topography and show no correlation with faults, and the rotations are gradual, supporting the interpretation that faults exert limited control on the stress field beyond distances of a few kilometres, if at all. The key, the authors suggest, lies in the magnitude of the horizontal differential stress: where several plate tectonic forces superimpose and the difference between the maximum and minimum horizontal stress magnitudes is small, regional topography and stiffness contrasts gain relatively greater leverage and can twist the stress compass substantially. In northeastern Australia, forces from the Solomon subduction and New Guinea collision zone dominate, but further south the added influence of the New Hebrides subduction zone may create a low-differential-stress state ripe for rotation.

Not everywhere is so chaotic. Large intraplate regions such as central western Europe and northeastern North America show remarkably uniform stress orientations over distances greater than 1,000 kilometres, parallel to the direction of absolute plate motion, confirming that plate tectonic forces remain the primary control in those settings. The new database also confirms that stress orientation generally does not rotate with depth: the vast majority of records from deep boreholes show no significant rotation, and no systematic difference emerges between shallow borehole data and earthquake-derived data from greater depths, with exceptions mainly where boreholes cross faults or where mechanical decoupling horizons exist. The stress regime, by contrast, does change with depth, which matters for practical applications, since maps dominated by earthquake data reflect conditions where earthquakes nucleate rather than the shallow conditions relevant to repositories and reservoirs.

The implications ripple outward from pure geodynamics to the energy transition. Stress rotations serve as proxies, together with geomechanical-numerical models, for quantifying the relative importance of plate boundary forces against regional and local controls, and they directly affect assessments of tectonic fault criticality and the safe design of geothermal systems and storage sites. The project is now expanding beyond orientations to compile stress magnitude data, beginning with Germany and neighbouring countries, and developing a corresponding quality-ranking scheme. With 77,365 A-to-C-quality records now plotted on a single global map, and user-defined stress maps available through the project’s online tool, the hidden architecture of pressure beneath our feet has never been so clearly visible, or so full of surprises.

Subject of Research: Contemporary crustal stress orientation patterns from the World Stress Map Database 2025

Article Title: Patterns of contemporary horizontal stress orientation in the Earth's crust derived from the World Stress Map Database 2025

Article References: Heidbach, O., & Rajabi, M. (2026). Patterns of contemporary horizontal stress orientation in the Earth's crust derived from the World Stress Map Database 2025. Solid Earth, 17(5), 735-745. https://doi.org/10.5194/se-17-735-2026

Image Credits: AI Generated

DOI: 10.5194/se-17-735-2026

Keywords: World Stress Map, crustal stress, maximum horizontal stress, plate tectonics, geomechanics, SHmax orientation, intraplate stress rotations, borehole breakouts, earthquake focal mechanisms, Alpine foreland, Bowen Basin, geothermal energy

Between 2006 and 2017, North Korea carried out six underground nuclear tests at the Punggye-ri facility, which lies beneath Mt. Mantap in the country’s northeast. The final and largest of these explosions, conducted in September 2017, had an estimated yield of 100 to 250 kilotons. The United States Geological Survey registered the event as a magnitude 6.3 earthquake, and satellite radar measurements revealed substantial deformation of Mantap’s summit, evidence that the blast had physically warped the rock above the test cavity. For years afterward, most scientific attention focused on the standard questions of nuclear test monitoring: where the explosion occurred, when it happened, and how large it was.

The new research, led by Xingli Fan and colleagues, takes a broader view. Rather than characterizing individual explosions, the team set out to understand how the tests changed the seismic behavior of the region as a whole. To do so, the researchers analyzed seismic data recorded in China and South Korea since 2008, drawing on stations located between 80 and 200 kilometers from the test site. That distance matters: instruments far from the epicenter record the region’s background seismicity rather than just the violent, short-lived signals of the detonations themselves, allowing the team to track subtle changes in earthquake activity over nearly two decades.

The scale of what they found surprised even the authors. By combing through the continuous data streams, Fan and colleagues identified 1,399 local earthquakes around Mt. Mantap between 2008 and 2025, far more than previous earthquake catalogs had recorded. Many of these events were small enough to escape earlier detections, but together they paint a picture of a crust that has been fundamentally unsettled by the testing program. The sheer number of events suggests that the mountain and its surroundings experienced a level of seismic disturbance that conventional monitoring approaches had substantially underestimated.

What makes the sequence remarkable is its timing. After most underground nuclear explosions, seismologists observe a familiar pattern: a burst of aftershocks in the immediate vicinity of the blast cavity that decays rapidly over days or weeks, much like the aftershock sequences that follow natural earthquakes. Mt. Mantap refused to follow the script. Instead of decaying, seismic activity after the September 2017 test began roughly three weeks after the detonation and then continued to grow in both frequency and magnitude through 2025, intensifying years after the last explosion. This is not the signature of a crust settling back into equilibrium; it is the signature of an ongoing process.

High-precision relocation of the earthquakes revealed an equally striking organizational pattern. The events were not scattered randomly through the rock but concentrated along two roughly north-northwest–trending fault structures, some of which had existed before the testing began while others had gone unrecognized until now. That alignment indicates persistent and organized fault reactivation rather than the chaotic shattering one might expect from simple blast damage. The faults, in other words, appear to have been switched on as coherent structures, slipping repeatedly along their length in the years following the detonations.

The physical explanation proposed by the researchers is a gradual one. According to the study, the repeated nuclear explosions progressively damaged the shallow crust around Mt. Mantap and altered its internal stress field. In a region where many faults already sat close to failure, even modest perturbations to the stress balance could tip them into sliding. Rather than failing all at once, the faults became active gradually over several years as stresses redistributed through the damaged rock, producing the slow escalation in seismicity that the team documented. The 2017 explosion, with its enormous yield and the pronounced ground deformation it caused, likely delivered the decisive push.

The implications extend well beyond a single mountain on the Korean Peninsula. The study demonstrates that underground nuclear explosions have the potential to reactivate faults that were previously seismically quiet, awakening geological structures that no monitoring program would have flagged as hazardous. Earthquakes triggered in this way may be difficult to distinguish from naturally occurring tectonic activity, since they occur on real faults and can continue for years with no obvious temporal link to the explosion that caused them. For organizations tasked with verifying compliance with nuclear test bans, that ambiguity is a serious concern: a cluster of earthquakes near a former test site could represent natural tectonics, lingering explosion effects, or something else entirely.

Monitoring former nuclear test sites, the authors suggest, will require longer horizons and finer tools than have typically been applied. Dense regional seismic networks capable of detecting and precisely locating small events, combined with sustained observation over many years, appear essential to capturing delayed sequences like the one at Mt. Mantap. The study’s catalog of 1,399 earthquakes, assembled from stations hundreds of kilometers away, shows what such sustained analysis can reveal even at considerable distance from the source. As more nations conduct and then abandon underground nuclear testing, the geological legacies they leave behind may keep generating earthquakes, and scientific attention, long after the political headlines have moved on. For now, Mt. Mantap stands as the clearest demonstration yet that the consequences of nuclear testing are written not only in treaties and diplomacy but in the slow, patient mechanics of the Earth’s crust itself.

Subject of Research: Delayed earthquake activity caused by nuclear explosion–induced fault reactivation at North Korea's Mt. Mantap test site

Article Title: Mt. Mantap nuclear testing triggered delayed seismicity via fault reactivation

Article References: Mt. Mantap nuclear testing triggered delayed seismicity via fault reactivation. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: Mt. Mantap, Punggye-ri, nuclear test monitoring, fault reactivation, delayed seismicity, induced earthquakes, seismology, North Korea, intraplate faults, crustal stress, Science journal, nuclear explosions