The Nine-Day Signal: What Shook the Earth in 2023?

Dickson Fjord, eastern Greenland, Arctic North Atlantic · September–October 2023
Complete and free to read

The signal was faint. Its persistence was the problem.

On 16 September 2023, seismic instruments began recording a slow, remarkably regular vibration. Roughly every minute and a half, the pattern repeated.

An earthquake usually leaves a complicated mixture of frequencies. This was much closer to a single note—a hum too low for human ears.

Hours passed. The signal remained. Stations separated by thousands of kilometres were recording the same unfamiliar rhythm.

Researchers gave it a provisional name: USO—Unidentified Seismic Object.

It appeared to come from eastern Greenland. Whatever was producing it had not finished. [1, 3]

A Mountain Missing

At Dickson Fjord, part of a mountain had fallen into the water.

About 25 million cubic metres of rock and ice descended from the slope. Near the impact, the tsunami reached roughly 200 metres above sea level. Farther through the fjord system, much smaller waves damaged an unoccupied research station on Ella Ø, around 70 kilometres away.

The photographs left little doubt about the violence of the collapse. A mountain peak and glacier had occupied the view. Now there was a vast scar.

That explained the first disturbance. A falling mountain could shake the ground, and water displaced by it could devastate a shoreline.

But the instruments were still registering their slow rhythm days later.

The collapse was over. Something else was continuing. [1, 3]

The Explanations That Did Not Settle It

The investigation brought together researchers working on earthquakes, glaciers, landslides and waves. The international team behind the eventual Science paper numbered 68.

In interviews about the search, researchers described possibilities ranging from vibrating ice to hidden volcanic activity. Perhaps water moving within a glacier was producing a resonance. None supplied a convincing account of the persistent signal.

There was room for humour, too. Speaking to Quanta Magazine, seismologist Stephen Hicks joked:

“If we found no other explanation, we would have gone for sea monster or baby dragons.”

The monsters belonged to the conversation’s lighter moments. The serious difficulty remained the same: finding something that could repeat a movement for nine days while slowly losing energy. [3, 6]

The Fjord That Seemed Quiet

Three days after the collapse, a Danish military vessel visited the fjord. The team could document the changed landscape, but did not observe the oscillation later identified as the source of the global signal.

At distant seismic stations, it was still there. [4]

There was another complication. An instrument already measuring water levels in the fjord transmitted readings at 15-minute intervals. The suspected movement completed a cycle in about 90 seconds. Its rhythm could slip between those widely spaced observations.

The instrument also stood away from the part of the fjord where modelling placed the strongest oscillation. Being close to an event did not necessarily mean having the right record of it. [1]

The Water Left Behind

One explanation survived the discussion. After the initial tsunami, perhaps the water had continued rocking across the fjord.

Such an oscillation is called a seiche. Water rises towards one side of a basin, returns, and rises towards the other. Here, the basin was a deep Arctic inlet bounded by steep rock walls.

With each movement, the shifting water exerted a force on the surrounding ground. A relatively small stretch of sea could therefore generate seismic waves detectable far beyond Greenland.

The difficulty was duration. Why had the movement not quickly died away?

The Geological Survey of Denmark and Greenland records that one researcher even tried reproducing the effect in a bathtub. The attempt failed. The fjord’s particular shape mattered.

Computer models incorporated its depth, narrow channel and bends. They produced an oscillation that lost energy unusually slowly, with a period close to the one in the seismic recordings.

The fit was not exact, but the different lines of evidence converged. In September 2024, the team published its explanation: the mountain had started the disturbance; trapped, oscillating water had sustained it. [1, 3]

A Satellite Had Passed Over

The explanation could be tested against another kind of record.

On 17 September 2023, the day after the collapse, the Surface Water and Ocean Topography satellite—SWOT—had crossed Dickson Fjord.

Its radar measured water height across a broad strip of the surface. Instead of tracing only a narrow line beneath the spacecraft, it could reveal differences between the two sides of the channel.

In the resulting visualisation, the water was tilted: higher towards one shore, lower towards the other.

The measurements had captured a shape that no photograph from a passing vessel was likely to reveal. [5]

The Signal Returns

On 11 October, a further landslide-generated tsunami produced another prolonged signal from the same fjord.

The recurrence offered a second event to examine. A study published in Nature Communications in June 2025 combined satellite observations associated with the two episodes and the seismic record.

Across separate observations, the slopes of the water surface faced opposite directions. The researchers examined winds and tides as alternative explanations. Their analysis supported water moving back and forth across the channel.

These were snapshots, separated by gaps, rather than a continuous film. Reconstructing the movement between them required the seismic measurements and further analysis.

Nevertheless, the case had changed. A mechanism first reconstructed from the landscape and the vibrations now had direct observations of the water at its centre. [2, 4]

Evidence Assessment

Archive Assessment

The nine-day signal was a genuine anomaly with a well-supported natural explanation. The uncertainty concerned its source, not whether the instruments had recorded something real.

“Shook the Earth” describes its reach rather than a worldwide experience of violent shaking. Sensitive instruments detected small ground movements far from the fjord. The destruction was concentrated around the landslide and the tsunami’s path.

Some details of the reconstruction remain less certain than the central explanation. Estimates of the seiche’s initial size differ, and the satellite observations are intermittent. Those limits do not erase the agreement between the water measurements, seismic evidence and modelling.

The researchers traced the slope’s increasing vulnerability to the thinning of glacial ice that had helped support it. The sequence joined a changing mountainside to a remote inlet, and that inlet to instruments on other continents. [1–5]

What makes the case remarkable is the difference between what could be seen and what could be measured. The mountainside was visibly broken. Days later, the water gave visiting observers no obvious sign of the motion still registering far away.

A mountain had fallen. For nine days, the water kept moving.

Continue Through the Archive

Some unusual events leave little more than testimony. Others leave measurements so unfamiliar that understanding them becomes an investigation in itself.

Explore more cases in The Anomalous Archive →