In 2011 Japan reeled from the effects of a devastating magnitude 9.0 earthquake.

But unnoticed in the chaos resulting from the quake, its major aftershocks, and the tsunami it caused, something strange happened.

About 16 minutes after the earthquake, but before the aftershocks hit, GPS stations all over Japan registered an eastward lurch—unconnected to any specific quake or aftershock.

A new analysis of data from the quake led by UChicago geophysicist Sunyoung Park suggests an extraordinary explanation for this movement: The waves of energy from the earthquake traveled downward to Earth’s core and then back up. This further displaced the tectonic plates on which Japan sits, causing a jolt that—according to the analysis, which was published June 18 in Science—permanently moved the entire island arc of Japan eastward by up to six millimeters.

Seismologists knew that large waves from earthquakes can travel through Earth and even reverberate off the core. But this is the first time the phenomenon has been identified as the cause of tectonic plate slippage near Earth’s surface. Such reverberation events are “a previously unrecognized source of seismic hazard,” says Park.

The 2011 earthquake, localized off the coast of the Tohoku region of Japan, ranks among the strongest ever recorded; the combination of quake and tsunami killed 20,000 people.

It was also one of the most thoroughly documented. Because Japan has a long history of earthquakes, the country has thousands of monitoring stations. Scientists immediately began poring over the recorded data to try to understand what happened, and hundreds of papers were published.

But years later, the strange eastward wiggle in the data was still bothering Park, an assistant professor of geophysical sciences.

This wiggle didn’t correlate with any of the aftershocks registered at the surface. But, like a large quake or aftershock, it was registered at precisely the same time by stations across Japan. Typically, Park says, “we would see an offset like this when there’s an actual earthquake happening. But here there was no known aftershock at this time, so we were quite curious.”

With collaborators Hiroo Kanamori of Caltech and Luis Rivera of the University of Strasbourg, Park began to rule out possible causes. An undersea landslide didn’t fit the data—too localized. Same problem for a slow slide at one of the faults.

Instead, the team came to believe it was due to a wave of energy from the quake that had radiated downward through the planet, struck Earth’s outer core—which is a liquid metal alloy—and reverberated back up to the crust. There, it triggered another slip along two major plate boundaries around Japan. The journey down and back up, about 3,600 miles (5,800 kilometers) round trip, took about 15 minutes.

Even as scientists have gotten better at understanding and predicting hurricanes and tornadoes, earthquakes have remained difficult to study. They occur infrequently—especially large quakes—and take place over very large areas. And, of course, most of the action takes place deep below ground, or worse, under the ocean, where few measurements are possible.

Scientists, including Park, are coming up with new and innovative ways to study seismic activity. The recent finding only adds to the field’s complexity, complicating our understanding of large quakes and tectonic plate movement.

Taking place over an area stretching about 1,800 miles (3,000 kilometers), the eastward jolt is the broadest seismic event ever recorded. It released about the same amount of energy as a 7.5 magnitude earthquake.

It is also the first seismic event on record to involve multiple major tectonic plate boundaries; it took place at the intersections of both the Pacific and Okhotsk plates and the Philippine Sea and Eurasian plates.

This unusual seismic event wasn’t immediately noticed, Park and her collaborators write in their paper, because seismic sensors are designed to look for the shorter high-frequency signals that accompany more typical quakes felt on Earth’s surface.

What’s more, “there was a ton of noise going on in the aftermath of the 9.0 quake,” Park explains. But by comparing both GPS and seismic data from stations across the country, the researchers were able to tease out the signal.

Park thinks it’s likely that vigorous shaking from the original quake weakened the plate boundaries. This weakening made it easier for the later-arriving wave from the core to trigger new plate movement near the original quake’s epicenter and along plate boundaries farther away.

“This indicates that large earthquakes can influence the fault even after the main shaking is over,” Park says.

Overall, she adds, it’s clear we still have much more to understand about large quakes and tectonic plate behavior: “This is adding an entirely new angle of seismic hazard we didn’t know about before.”