How Arctic Permafrost Thaw Increases Seismic Risks
Permafrost, the natural cement of Arctic soil, is thawing due to warming. An international study shows that this melting makes earthquakes more destructive, with sudden landslides and floods. Researchers used satellite data and machine learning to illuminate these mechanisms.
How Climate Change Can Amplify the Impact of Earthquakes in the Arctic
Permafrost, the giant ice block that holds Arctic soil in place, is increasingly detaching, potentially making earthquakes more devastating. When the ice acting as cement melts, the soil becomes fragile, and each shock can trigger a cascade of landslides, liquefaction, and floods. This dynamic, observed during a recent seismic event, prompts scientists to reassess risks in polar regions.
What the study reveals: weakened permafrost amplifies seismic shaking
The international team analyzed a seismic event in a thawing permafrost zone. Observations show that, unlike traditional soils, frozen soil loses its cohesion as soon as the temperature exceeds the freezing point of interstitial ice. This loss of cohesion creates cracks that rapidly propagate under seismic waves, increasing the affected area.
Researchers found that areas where permafrost was farthest along in its thaw process exhibited ground displacements two to three times greater than those observed in regions where permafrost remained stable. This amplification is not due to an increase in earthquake magnitude, but to the geological support's fragility. The study highlights that prediction uncertainty increases when the soil model does not account for the degree of permafrost thaw.
Scientific explanation of the phenomenon
Permafrost is a key element in understanding the interactions between climate and seismicity. Ice that holds the soil in place plays a crucial role in soil stability. When ice melts, the soil becomes more unstable and susceptible to landslides and liquefaction. Seismic waves generated by an earthquake can then trigger a cascade of natural phenomena such as landslides, liquefaction, and floods.
How researchers combined satellite data and machine learning models
To quantify permafrost conditions, scientists exploited Copernicus satellite data, including soil temperature measurements and radar images that detect structural changes. These data were integrated into a predictive model based on a neural network, trained using machine learning. The model, similar to those used in weather forecasting like GraphCast or Pangu-Weather, allowed mapping permafrost depth and estimating soil failure probability during an earthquake.
The training process used archives of historical earthquakes combined with permafrost records, allowing the model to learn and improve its predictions over time.
