After North Korea detonated its largest nuclear device beneath Mount Mantap in 2017, scientists weren’t surprised to detect a few earthquakes in the immediate aftermath of the underground test. A massive explosion fractures and deforms surrounding rock, typically triggering small aftershocks that fade away quickly as the crust adjusts.
But years later, the crust around the Punggye-ri test site was still restless, according to a study published today in Science. Rather than fading away, earthquake activity increased over the following years, the researchers report. Even in 2025, small earthquakes were occurring along two suspected faults near Mount Mantap, some as far as 20 to 30 kilometers from the test site.
The finding adds a striking example to a growing catalog of human-induced earthquakes. Injecting wastewater from oil and gas operations, mining, filling reservoirs, geothermal energy projects, and subsurface nuclear detonations have all been shown capable of triggering earthquakes. What is unusual at Mount Mantap is the longevity: The nuclear tests appear to have triggered ruptures on already stressed crust over years.
“It’s completely different from our expectation or textbook cases,” says study co-author Kwang-Hee Kim, a seismologist at Pusan National University. Normally, he says, earthquake activity triggered by a large explosion “will decay with time.”
North Korea conducted six underground nuclear tests at Punggye-ri between 2006 and ’17. The final explosion, estimated to be the equivalent of 100 to 250 kilotons of TNT, registered as a magnitude 6.3 seismic event by the U.S. Geological Survey (USGS). About 8 minutes later, a magnitude 4.1 event was interpreted as the collapse of its underground blast cavity. Apart from an isolated small earthquake soon afterward, there were no immediate aftershocks.
Sustained activity in the area began about 3 weeks later. In 2018, Columbia University seismologist Won-Young Kim and his colleagues located 13 posttest earthquakes along a roughly 700-meter fault near the test site. They concluded the blast had changed stress levels in the surrounding rock and predicted it could continue to adjust. But the seismic activity would prove to be far more persistent.
The new study, led by Kwang-Hee Kim and geophysicist Xingli Fan of the Chengdu University of Technology, found a yearslong intensification of seismic activity, organized along broader fault zones. The researchers searched through seismic recordings collected in China and South Korea between 2008 and ’25, using the wave forms of known earthquakes as templates to find much fainter events buried in background noise. They identified 1399 earthquakes near the Punggye-ri site and determined precise locations for 955. Most were tiny earthquakes below magnitude 2. But instead of fading after 2017, their frequency increased through 2025, as did the total seismic moment released by the sequence—the combined size of all the earthquakes.
The earthquakes also revealed a geographic pattern. They occurred along two roughly parallel lines running north-northwest. One follows the projected continuation of a fault mapped south of Mount Mantap decades ago; the other has no known surface trace. Unable to visit North Korea, Kwang-Hee Kim and his colleagues recognized the connection only after combing decades-old geological maps.
Explaining the lingering activity is a challenge. The researchers’ leading idea is that the nuclear tests repeatedly damaged the shallow rock around Mount Mantap, shifting stresses onto preexisting faults that were already close to rupturing. The mountain’s weight and rugged shape may also have determined which faults were most vulnerable: Modeling by the team found unusually large stress variations in places where many of the earthquakes cluster.
Faults can be poised to slip even when they appear dormant, says Zhigang Peng, a seismologist at the Georgia Institute of Technology who studies earthquake triggering. Human activities such as fluid injection can similarly push faults that are “subcritical, but not far from failure” toward rupture, he says.
But that does not fully explain the long delay. The brief stress pulse carried outward by seismic waves should have triggered ruptures almost immediately. The longer lasting stress change left behind by the nuclear blast, meanwhile, should decline rapidly with distance, making it difficult to account for earthquakes tens of kilometers away.
Another possibility is that slower moving fluids played a role, says USGS geophysicist Walter Mooney. Groundwater percolating through blast-damaged fractures may have weakened the faults, he says. The authors say they considered the idea but lack evidence for it.
For now, the mechanism remains uncertain. “The jury’s still out,” Mooney says. Even the study’s authors are cautious. “There are still many, many things I cannot explain,” Kwang-Hee Kim says.
Other researchers have also found evidence for the prolonged seismic activity around the test site. In June, seismologist Mengyi Ren of the China Earthquake Administration and colleagues reported in Seismological Research Letters that they had uncovered 647 previously undetected earthquakelike events from 2016 to ’24, including months in 2021 and ’22 with more than 30 events. But the studies differ: Ren’s team saw the earthquakes migrate toward the test site, whereas the new study found no clear migration.
For Kwang-Hee Kim, the result echoes an earlier investigation. In 2018, he and colleagues linked South Korea’s magnitude 5.5 Pohang earthquake to fluid injection at a geothermal project that activated a preexisting fault. Unsettled Mount Mantap, he says, suggests a very different kind of human disturbance can have a long seismic afterlife.
