Singing lava helps predict volcanic eruptions

Down below the ground, the sound of volcanoes might give us a clue as to if they are about to erupt. New research this month has opened our ears to the bubbling lava beneath the surface, and used this is models to make predictions about volcanic activity. Harry Lewis went to find out more about these studies from volcanologist Marie Edmunds at the University of Cambridge, and they caught up in a location with quite the opposite acoustics to an active volcano…

Marie - We are sitting in the library in the department of earth sciences, surrounded by centuries-old books.

Harry - There's been a couple of really interesting pieces of research in the field. Are there any physical attributes that we can measure to predict when a volcano might erupt?

Marie - There are two papers in particular that have been published in the last month that show our ability to forecast eruptions, I think really nicely. So the first one was published in Science Advances this month, and it's all about the volcano Kīlauea in Hawaii. What they've done is they've managed to develop a really complex model to understand how magma de-gasses, as it comes up to the surface. So magma vesiculates - just like when you pop the cork from a bottle of champagne - all the bubbles start to form in the magma. As the magma comes up towards the surface, all of these big bubbles form and make the magma less dense. Scientists can measure seismic waves coming through the magma and it turns out that this gas really affects the seismic velocity through the magma. And it actually causes these really strange seismic signals called very-long-period events. It turns out that these VLP events, very-long-period events, actually show us when there's a big influx of gas-rich magma at depth, which might mean that an eruption is going to happen, or it might mean that an ongoing eruption is going to slightly intensify for a short time.

Harry - And we've got the clip of this sound - it's been sped up so that we can actually hear it. Take a listen to this. [SOUND] How on earth do you go about getting that recording and accumulating that data?

Marie - Usually networks of seismometer are placed around a volcano and to get really good data, they need to be close to the source and to locate the seismic event, you need a network of them so that you can actually work out how deep the source is and how big it is.

Harry - And is it that we might be able to tell exactly when the volcano might erupt just from this one piece of data?

Marie - Well, that is a really great question because in fact, what it turns out is most useful is if you collect lots of different kinds of data at the same time. So there's the seismic data, but there's also really precise GNSS data, which is global navigation system data, or you might know it as GPS data. That allows scientists to measure very precisely how the ground is moving. So sometimes it inflates like a balloon and other times it deflates. And it turns out that these VLP events often coincide with periods of short-term inflation at the volcano, and essentially is telling us that really gassy magmas coming in at depth and recharging the chamber. There's a second paper published this month and this one's focused on one of the volcanoes in the Galapagos islands. And what they've done is they've created, again, a really sophisticated model to explain how the stress state of the crust, that is the country rocks around the magma chamber, changes in response to magma being pumped into a chamber in the crust. So again, you can use the analogy of a balloon being pumped up and that's causing pressure to build inside the balloon, but it's also stressing the rocks in which the balloon is sitting in. And combining that with very detailed measurements of inflation and deflation on the Earth's surface, they've looked at a particular eruption that happened at Sierra Negra and they were able to forecast it really quite precisely within a few months of the eruption happening. And when they looked all the data in hindsight, of course, they were able to predict it really precisely.

Harry - Is this a particular outlier in the field of volcanoes in general? Or could we do this with every volcano that there is?

Marie - What these two volcanoes have in common, Kīlauea and the Galapagos, they're both magmas that don't have a huge amount of water in or gas. So that means that when magmas sitting in the crust, all of the inflation signal that you see at the surface is caused by magma coming in or out. Now it's a much more complex picture at subduction zone volcanoes; those volcanoes that tend to erupt really explosively with big ash-rich gas columns. To model and forecast those kinds of eruptions, it needs a rather more sophisticated model. But certainly eruptions like Hawaii, Galapagos, Iceland, and even the Canary islands, we saw eruption there recently just before Christmas, these sorts of approaches and models are really powerful.

Harry - One question that sticks in my mind is that throughout history we've been aware of these eruptions, is it really important to be able to predict them still? Because surely nobody's in the close vicinity anymore. Haven't we learned from those previous eruptions?

Marie - Unfortunately not. I mean, many of these really hazardous volcanoes are located in regions of the world where people feel the need to live close to the volcano. And that could be for lots of different reasons. It could be because the ground is more fertile around the volcanoes. It could be that the volcano itself is affecting the water table, which makes it a good thing to live close to them. So unfortunately, eruption forecasting remains really important.

Julia - Thanks to Marie Edmunds for guiding us through that brilliant research. The sound of those very-long-period seismic events were provided by Leif Karlstrom, one of the researchers behind that first publication Marie mentioned, in Science Advances.