For nine days, the Earth hummed in harmony with a pulse of about 92 seconds a narrowband vibration that appeared almost metronomic on seismographs far removed from the Arctic. It was not a tectonic fault line, but a fjord a long, steep-sided channel of water in East Greenland, where a landslide wave had become trapped and begun sloshing.
In the instance of Dickson Fjord, the initiation shock was experienced when a rock ice avalanche occurred from Hvide Støvhorn, a mountain that stood at a height of about 1,200 meters relative to the level of the fjord. This led to the displacement of 25 million cubic meters of metamorphic rock and glacier ice, which in turn hastened the movement of 2.2 million cubic meters of glacier ice. Upon impact at a speed of 47 m/s, the force pushed the water in a near-field runup of 200 meters.
Fjords can be considered as natural waveguides, and the geometry of Dickson’s fjord seems to be particularly favorable for amplification with nearly parallel shores, a width of 2.7 kilometers, and depths of hundreds of meters. The first tsunami caused damage that was quite far from the strike area, with water levels remaining at 4 meters at a research outpost on Ella Island, 72 kilometers away.
The more interesting engineering event happened after the initial wave. Rather than spreading out into the open ocean, the energy reorganized itself into a standing wave, a seiche, which oscillated back and forth between the two sides. Simulations and data agreed on a cycle of 7.4 meters amplitude, with a period of 90-92 seconds. This was significant because it caused the moving water to become a periodic horizontal force on the bottom of the fjord, calculated at 5×10^11 newtons, effectively pushing the crust like a slow piston.
This mechanical push was recorded in the global seismograms as a monochromatic very long period wave with a central frequency of 10.88 mHz. The directionality of the wavefield was physically consistent with a fjord-scale oscillator, where the Rayleigh waves were strongest perpendicular to the channel and the Love waves were strongest in the direction of the channel axis. A second seiche event in the same gully later confirmed that this pattern was not a rare anomaly but a danger pattern.
To confirm the seiche, it was essential to monitor the water itself, which was no small feat in so remote a fjord that it could freeze over. A standard radar altimeter measures a thin strip of water beneath a passing spacecraft, excluding the cross-channel slope that would suggest a standing wave. The secret to the detection was Ka-band interferometric swath altimetry, which had been designed for the Surface Water and Ocean Topography mission, measuring a wide strip of the water surface rather than a thin track. Scientists calculated cross-channel slopes that suggested a difference in water level of up to 2 meters, reversing from one flight to the next as would be expected for back-and-forth motion.
The analysis of this signal was as much an analytics challenge as it was a sensing challenge. The SWOT “pixel cloud” observations needed human analysis to reject bad passes, geophysical corrections (tropospheric delays and a geoid model), and special tide processing to prevent aliasing in a region with limited repeat coverage and seasonal sea ice. This work has shown that fjord-scale extremes can now be described using a combination of remote sensing, regression with uncertainty, and physics-based modeling.
The broader implication is at work upstream of the fjord margins. Glacier retreat reduces the influence of glacier debuttressing on valley sides, making them more vulnerable to a transition from slow deformation to failure, particularly in regions where rock, water, and ice interact in deep channels. The nine day seismic hum in Dickson Fjord shows that a local phenomenon can be one system, from ocean to orbit, with a specific pattern for future study in other high-latitude fjords.
