Smoke filled the cabin as the Boeing 747 plunged towards snow-covered mountains in southern Alaska. All four engines had shut down, and it took the pilots eight long minutes to regain control of the aircraft. No one on board was hurt — but they had a very close call with an erupting volcano. The jet had flown through an ash cloud.
Incidents such as this near miss from 1989 show why geologists have long sought to forecast volcanic eruptions: to protect people on the ground and in the air. Now scientists are one step closer to this goal.
Maurizio Ripepe, a geophysicist at the University of Florence and his colleagues have created the world’s first automated volcano early-warning system, which alerts authorities near Mount Etna in Sicily about one hour before an eruption. The team described the system last month1 in the Journal of Geophysical Research: Solid Earth.
The approach relies on the fact that volcanoes are noisy. Their rumblings and explosions can sound like a jet engine or even a high-pitched whistle, but they also produce low-frequency infrasound waves that people cannot hear. Unlike seismic waves, infrasound waves can travel for thousands of miles, allowing scientists to spot volcanic eruptions from afar. When Krakatoa erupted in Indonesia in 1883, its infrasound signal travelled around the globe twice.
Bubble and squeak
With that in mind, Ripepe and his colleagues turned to Mount Etna, Europe’s largest active volcano. At first, they wanted to create a simple system that could detect an eruption using data from an existing array of infrasound sensors, and automatically alert authorities. But their ambitions grew when they discovered that the volcano often produces infrasound waves before it erupts, making prediction possible.
Although the finding was a surprise, the scientists say that it makes sense given that Mount Etna is an ‘open-vent’ volcano with exposed magma. As gas rises out of that magma before an eruption, it causes air in the volcano’s crater to slosh back and forth — creating sound waves like those in a woodwind instrument. And just as the sound of a musical instrument depends on its shape, the geometry of a volcano’s crater also affects the sounds it can produce.
The team created its early-warning system in early 2010 and analysed its performance during 59 eruptions over the next eight years. The system — an algorithm that analyses infrasound signals from the sensor array — successfully predicted 57 of those events and sent messages to the scientists about one hour before an eruption took place. The extended test run was so successful that in 2015, the scientists programmed the system to send automatic e-mail and text-message alerts to the Italian Civil Protection Department in Rome and the Sicilian city of Catania.
An automated alert system can broadcast warnings faster than predictions that require experts to vet information beforehand, says John Lyons, a geophysicist at the Alaska Volcano Observatory. And time is of the essence for communities near volcanoes, or passengers in a jetliner that can fly faster than 800 kilometres per hour. “You’re covering a lot of ground really fast, so if there is an ash cloud that has suddenly popped up, then the pilots need to know that information as soon as possible,” he says. “Every minute counts.”
Although Lyons worries about the potential for false alarms, he says that the system is a pivotal step forward — not only for Etna, but perhaps for similar volcanoes around the globe.
These could include Kilauea, an open-vent volcano on Hawaii’s Big Island whose months-long eruption this summer destroyed whole neighbourhoods, says David Fee, a geophysicist at the Alaska Volcano Observatory. But Fee, who has studied Kilauea, says that it differs from Etna in some key ways. Eruptions at Kilauea can originate from the volcano’s summit and an area on its flank called the East Rift Zone. Etna, whose internal plumbing is far simpler, only erupts from its summit.
Because of this, Lyons says that Mount Pavlof in Alaska, one of the United States’s most active volcanoes, could be a better test for an early-warning system. Pavlof has a similar structure to that of Etna and has shown a similar increase in infrasound activity before the most energetic phase of its eruptions. Its frequent activity could also give researchers a large set of observations with which to tune their algorithm for predicting eruptions.
Ripepe and his colleagues are beginning to test their early-warning approach in Iceland. Working with the Icelandic Meteorological Office in Reykjavik, the scientists have installed five sensor arrays across the island to monitor infrasound waves from multiple volcanoes. Among them is the infamous Eyjafjallajökull, whose last eruption, in 2010, shut down air traffic across northwestern Europe for weeks.
Nature 563, 456-457 (2018)