Why volcanoes scream before they blow up

Researchers have recorded an Alaskan volcano's 'scream' before it erupted, and are now closer to understanding what made it do so.

Redoubt Volcano looms over the Kenai River in Alaska in 2004. The volcano erupted in March 2009, after gurgling for months.

Al Grillo/AP

July 15, 2013

As the volcanic Mount Redoubt, some 100 miles southwest of Anchorage, Alaska, began to swell in winter 2009, several earthquakes per hour jiggled the surrounding towns. Spring came, and the volcano heaved steam and dust into the air. At the end of March, that heaving turned explosive, sending mudflows reeling down the volcano’s sides.

It was an unremarkable volcanic performance – except for one detail: screaming, at an unusually high pitch that offers new insight about the strange, still mysterious progression of mile-deep events that precede a volcanic eruption.

Harmonic tremor, a sustained release of energy, is common in volcanic eruptions – but never has the frequency been so high as it was at Redoubt, beginning at a frequency of about 1 hertz and then ascending to about 30 hertz. Most volcanoes’ harmonic tremors hover at about 1 hertz, far too low for a human to hear. In humans, the audible frequency range begins at about 20 hertz, so a person engaged in the highly inadvisable activity of vacationing inside an active volcano’s crater would be able hear the tremor at its high frequency, just before the volcano explodes.

“Screaming at such a high pitch has not been observed before,” said Alicia Hotovec-Ellis, a University of Washington doctoral student and lead author on the study, published in The Journal of Volcanology and Geothermal Research, in an email interview. “A somewhat similar signal was observed at Montserrat, but only up to 3 Hz as opposed to 30 at Redoubt.”

To measure the frequency, Ms. Hotovec-Ellis converted into sound seismic data recorded from the volcano’s pre-eruption, and then sped up the recording. The first tape, a ten-second clip of ten minutes of data sped up 16 times, is a harmonic tremor ascending to higher and higher frequencies before coming to an abrupt halt just before the six eruptions, like cannon fire in the 1812 Overture. The ascension in pitch sounds as though the piccolo player went rogue.

Another recording, an hour of some 1,600 small earthquakes compressed into a one-minute clip, is an ominous drumbeat.

This is also the first time that the drumbeat of repeating earthquakes has been so explicitly linked to the harmonic tremor, suggesting that the screams are emitted from those pre-eruption earthquakes.

“Redoubt is special in that the frequency/pitch is so high, and that the earthquakes are well explained by a frictional sliding mechanism,” said Hotovec-Ellis.

Can Syria heal? For many, Step 1 is learning the difficult truth.

Once establishing that link between high frequency sound and repeated small earthquakes, the University of Washington team then partnered with another team from Stanford University to understand what was causing that sudden crescendo of pitch and quake – and then what brought it to a dramatic stop.

In Stanford team leader Eric Dunham’s model of the quake, he showed how pressure in Redoubt’s faults must have rapidly risen to about 100 times atmospheric pressure to create 30 earthquakes per second. As the pressure rose, the earthquakes became more insistent and more frequent. That extreme, localized pressure could occur because Redoubt’s faults are small, relative to the huge ones that snake the Californian spine, where earthquakes are expected to occur once per century.

The model’s results are published in a separate paper in Natural Geoscience, on which Dunham and Hotovec-Ellis are co-authors.

But what causes that tremendous pressure is still unknown, and scientists have floated several ideas. One explanation is that the rising pressure forces some of the magma to crystallize. The high-pitched noise is then a result of friction as the rest of the magma squeezes upward through the narrowing conduit in rapid earthquakes.

“There’s not a lot of constraints on what’s happening a mile underground before an eruption,” said Dunham, in a phone interview.

It’s unlikely that the volcanic scream could be directly used to better predict volcanoes, said Dunham, noting that by the time the volcano sings, several other more obvious signs will already have signaled the volcano’s impending blow. 

Still, applying the model to volcanic eruptions around the world could begin to build a better picture of the pre-eruption climate in the volcano’s underbelly, he said, noting that a broader understanding of those strange depths could mean improved predictive power.