For 15 years after the 2011 Tohoku-Oki earthquake — among the most thoroughly instrumented natural disasters in history — a small anomaly in the GPS data sat unresolved in the archives. Approximately 15 minutes after the magnitude-9 main shock struck off the coast of northeastern Honshu, GPS stations distributed across the entirety of Japan registered a tiny but unambiguous step-like displacement to the east, of approximately five to six millimetres, occurring nearly simultaneously across the country. The shift was far too small to be felt by anyone, far too small to cause damage, and far too late to be attributable to the main rupture itself. It did not correspond to any known aftershock. It was reported in subsequent compilations of Tohoku data but was, for over a decade, an unexplained signal sitting in the corner of a vast dataset that researchers had not yet figured out how to interpret. On 18 June 2026, a team led by Sunyoung Park, an assistant professor of geophysical sciences at the University of Chicago, published a paper in Science proposing what they argue is the correct explanation. The shift, they conclude, was caused by a seismic wave that had travelled from the earthquake’s epicentre straight down through the Earth’s mantle, bounced off the outer core, returned to the surface, and arrived at Japan’s tectonic plate boundaries with enough residual energy to push the entire country a few millimetres eastward.
According to the University of Chicago’s announcement of the finding, the specific wave responsible is what seismologists call an ScS wave — a shear wave that travels downward from the earthquake’s source, reflects off the core-mantle boundary at approximately 2,890 kilometres depth, and returns to the surface as a shear wave. The reflection occurs because shear waves cannot propagate through liquids, and the Earth’s outer core is liquid iron and nickel. The wave bounces off the boundary the way a billiard ball bounces off a cushion. The round-trip travel time, for a wave originating at the depth of the Tohoku rupture, is approximately 13 minutes. When the wave returned to Japan, it arrived essentially simultaneously across the entire country — and it was strong enough, in the Park team’s interpretation, to trigger small additional slip on the already-stressed megathrust interfaces along Japan’s tectonic plate boundaries. The slip event itself, though only a few millimetres in displacement, released energy equivalent to approximately a magnitude-7.5 earthquake.
Why this had never been seen before
Seismologists have long known that large earthquakes generate seismic waves that travel throughout the planet’s interior and reflect off various internal boundaries. ScS waves, in particular, have been catalogued and studied for decades as a routine feature of the seismic record. What had not previously been observed was an ScS wave arriving back at the surface with enough energy to cause measurable, permanent ground displacement. As reported by Scientific American’s coverage of the Park study, the central reason is that ScS waves typically dissipate substantially during their long round-trip through the mantle — the deeper they travel, the more energy they lose to attenuation, and by the time they return to the surface, they are usually too weak to produce any detectable mechanical effect. The Tohoku event was different. The magnitude-9 main shock generated an ScS wave with a peak-to-peak amplitude exceeding one centimetre at Japan’s surface stations — a far stronger reflected wave than any previously recorded. The combination of unusually high amplitude and the synchronised arrival across a country already sitting on heavily stressed plate boundaries produced, in Park’s interpretation, the conditions for the wave to actually trigger small slip events along those boundaries.
The Park team spent substantial time ruling out alternative explanations before settling on the ScS-triggering interpretation. The possibility that the main shock had simply continued releasing energy for longer than thought could not account for a uniform shift across the entire country; the energy from the main rupture should have produced displacement concentrated near the epicentre and falling off with distance. The possibility of a submarine landslide triggered by the main shock could not explain the synchronised timing or the country-wide extent. The possibility of an unrecognised aftershock did not match any seismic record. Each alternative explanation could account for some features of the signal but not all of them. The ScS-triggered slip interpretation, in contrast, predicts exactly the geographic pattern, the synchronised timing, and the relationship to the main shock that the GPS data actually show.
What this means for seismic hazard
Per Science News’s coverage of the implications, the Park finding has substantive implications for how seismologists model the propagation of effects from large earthquakes. The traditional picture has held that the danger from a major earthquake is concentrated in the immediate vicinity of the rupture and in the aftershock sequence that follows. The new finding suggests that the influence of the largest earthquakes may extend much further, and through much deeper geological pathways, than this picture accommodates. A wave that travels 5,800 kilometres round-trip through the Earth’s interior, bounces off the core, returns to the surface, and triggers additional fault slip on the far side of the planet is operating on a different scale and through a different mechanism than the surface-wave aftershock pattern that has dominated seismic hazard modelling for the past century.
Park’s own framing of the finding, quoted in the University of Chicago announcement, is direct: “It’s striking because this is both an unprecedented length and area for a seismic event, and it is a previously unrecognised source of seismic hazard.” The practical implication is that the Tohoku event may have been triggering small slip events not only across Japan but potentially across other plate boundaries around the Pacific Rim that were exposed to the same returning ScS wave. Whether this kind of triggering will turn out to be a feature of all sufficiently large earthquakes, or whether it was specific to the unusual configuration of the Tohoku event, will require examination of GPS data from other recent magnitude-9 earthquakes — the 2004 Sumatra-Andaman event, the 1960 Valdivia event in Chile, the 1964 Alaska event, and the 2010 Chile event — to see whether comparable signals are detectable in the available data.
The deeper picture
As described in the abstract of the Park, Kanamori, and Rivera paper in Science, the finding fits within a longer scientific tradition of using large earthquakes as natural experiments on the structure of the Earth’s interior. Seismic waves from major events are, in practical terms, the only tool humans have for directly probing the structure of the planet beneath the relatively shallow depths accessible to drilling. The deepest hole ever drilled — the Kola Superdeep Borehole on Russia’s Kola Peninsula — reached approximately 12 kilometres before the heat and pressure exceeded the capacity of the drilling equipment. The Earth’s mantle begins at approximately 35 kilometres depth and continues to 2,890 kilometres. The outer core extends from there to 5,150 kilometres. The inner core sits beneath that, all the way to the planet’s centre at 6,371 kilometres. None of these layers has ever been directly sampled. Everything humans know about the structure of the planet beneath the upper crust has been inferred from seismic waves — their travel times, their reflections, their refractions, their changes in velocity as they pass through different layers — generated by earthquakes and recorded at the surface.
The 2011 Tohoku event, in retrospect, generated the largest and most thoroughly instrumented seismic dataset of any earthquake in human history. The five to six millimetres of country-wide eastward shift that nobody initially knew how to explain has now, 15 years later, been identified as the surface fingerprint of a wave that travelled deep enough to touch the boundary of the molten outer core and return. It is, in a substantive sense, the first time the species has directly witnessed the seismic interior of the planet reaching back up to nudge the surface. The signal is small. The implications are not.
