Volcano's Deadly Warning homepage

Bernard Chouet is a good listener. A volcano seismologist with the U.S. Geological Survey's Volcano Hazards Team in Menlo, Park, California, Chouet spent years patiently lending an ear to strange seismic resonance coming from volcanoes. In time he learned how these sounds could signal a dangerous rise in pressure as magma welling up from deep within the Earth tried to find its way out; if it didn't, the volcano eventually blew.

Using a new theory about these so-called long-period events, Chouet has successfully predicted several eruptions, including that of Alaska's Redoubt volcano in 1989. In this interview conducted for "Volcano's Deadly Warning," Chouet describes how he came about developing his novel theory, and how well it's holding up to scrutiny both as a theory and as a useful tool in the field.

NOVA: What's so extraordinary about volcanoes?

Chouet: Volcanoes are quite spectacular, especially at night—the scenery, the mountains, the eruptions with all the magma coming out of the vent and the gases flowing. The feeling of a mountain being alive is an extraordinary discovery for someone who has always felt that the Earth was a solid piece of rock. Then you see these explosions and realize that the Earth is an active planet beyond anything you have ever dreamed about.

NOVA: Why was it so important to you to know how volcanoes work?

Chouet: It was a sense of exploration, a sense of discovery. My feeling after studying engineering for many years was that pretty much everything had been explored and discovered, and there was no room for great exploration around the Earth. I felt that one of the last interesting frontiers to study was natural phenomena. I realized that although volcanoes had been looked at for a long time and people had always been fascinated by them, they were relatively poorly understood, and this was a frontier that was worth exploring.

NOVA: Who were some of the early pioneers in trying to understand volcanoes?

Chouet: The Japanese seismologist Takeshi Minakami was one of the early people who got interested in making seismic measurements on volcanoes and finding out if he could interpret anything. The instrumentation he had was rather limited, so he was mostly interested in classifying events and establishing some kind of order in the richness of the observations. He ended up classifying seismic events based on the character of their signature as A-type and B-type events.

NOVA: What's the difference between the two?

Chouet: A-type events have a very characteristic signature that starts with an impulsive first arrival. These events occur when a volcano first comes alive again and magma is moving at depth. To make its way to the surface magma must create a plumbing system it can flow through. So the volcano is readjusting itself with lots of earthquakes. The A-type earthquake is the sound of rock breaking as the volcano readjusts itself to the magma movement.

B-type events include two different types of processes, one of which is just like the A-type earthquake—it's rock breaking. The other type of process that was buried somewhere in the definition of B-type events is the long-period event. Unlike the A-type event, which reflects the brittle failure of rock, the long-period event reflects the change in flow pattern of the fluid that is being pushed through cracks.

Once the plumbing system of the volcano is unobstructed, magma can flow freely through this plumbing, and A-type earthquakes cease to occur. In this situation, you'd see almost exclusively long-period events. What the long-period events are telling you then is how the magma is evolving as it comes closer and closer to the surface. The long-period event has a distinct signature marked by an emergent signal and then a slowly dying single dominant tone. This is the sound of fluid under pressure. This long-period event gives us the means to quantitatively measure that pressure and to track the pressurization in the volcano.

NOVA: Can you give an analogy?

Chouet: Well, long-period signals in volcanoes and organ-pipe tones are very similar, for example. They are both representative of resonance phenomena. In an organ pipe, you have a column of air trapped between the walls of that cylinder, which is made of metal. Air is blown across a sharp edge at the bottom end of the pipe and sets up a standing wave in the pipe. There is a pressure variation along the pipe associated with the resonance of the air column in the pipe. You feel the pressure disturbance radiated from the open top of the pipe through the atmosphere to your ear.

In a volcano, a change in the flow pattern of the fluid in a crack may in a similar way trigger the acoustic resonance of the fluid, which applies a pressure variation on the crack surface. The resulting vibration of the crack wall is radiated into the solid in the form of seismic waves that propagate through the ground to the surface, where seismometers can pick them up.

Imagine what would happen if you were to place a cork on an organ pipe and seal all its openings and keep pumping air into a small hole at its base. If your pumping is brisk enough, you will induce resonance of the air filling the pipe with each pumping action, and each pumping action also raises the overall pressure in the pipe. If you keep pumping vigorously, you keep inducing resonance, and the pressure in the pipe keeps rising until eventually the cork blows out.

That's basically the process that goes on in volcanoes. You're seeing the pressure disturbances that are related to each pumping action, and then each long-period event is the result of one pumping action. The fluid filling the crack resonates and is trying to escape, but there's nowhere to go. The more pumping you do, the more pressurizing you have. Eventually you have a blow up.

NOVA: Apparently Minakami decided not to pursue the B-type signal. Why not?

Chouet: I think the answer is complexity. The B-type event looked very complex compared to the A-type. The A-type had a very short signature with a very sharp first arrival, so it looked like you could actually locate these things. The B-type, on the other hand, had this slowly emerging signal, which made it impossible to locate. In many cases it also displayed a very complex, long-lasting signature. In essence, Minakami concluded that this was something that we couldn't really address. At the time it was true—you couldn't really unravel all that complexity.

I also was trying to put things together in my head, and I thought trying to classify signatures was a little bit like classifying flowers. There are many types of flowers so you classify all these things, and then you have a feeling for the complexity and richness of nature. And that's fine, but if you do that on many different volcanoes—and there are quite a few volcanoes that are active at any one time—you're going to find that there's a very rich variety of seismic signatures on volcanoes. So you end up writing pages and pages and pages of signatures and classification, and people look at this and say, "Well, this is hopeless, because you've just got too much richness to deal with."

Chouet's solution

NOVA: So how did you approach this problem?

Chouet: What I wanted to know was, what specific events may occur in a volcano that are a telltale sign that you are actually proceeding with pressurization and toward a possible eruption. To do that you have to understand what is going on. You have to be able to interpret that process, that evolution of the volcano. You have to interpret the signature.

“Suddenly you realize the volcano is speaking to you, and you understand the language.”

NOVA: You began trying to interpret those signals on Mt. St. Helens, right?

Chouet: Yes. Mt. St. Helens erupted in 1980. It blew up and carved a huge crater and dispersed a lot of material all over the countryside, wiped out forests, killed people. During the summer of 1981, a colleague and I deployed a seismometer inside the crater, right next to the base of the lava dome that was growing in the crater. While we were there we recorded a lot of B-type events. Being so close to the lava dome, the events were easier to locate, and we saw that we had a collection of A-type events and B-type events, and among the B-type events we had these peculiar signatures.

NOVA: What was peculiar about them?

Chouet: It stared you in the face. Anyone seeing these wiggles on paper would say, "Wow, this is obviously different." I remembered from when I was still in engineering school what happens in hydroelectric plants in the pipelines that carry the water when you suddenly shut the valve controlling the jet of water hitting the turbine blades. If you stop the water flow very quickly it generates a very high pressure right there at the valve. That pressure pulse reverberates in the pipeline, going back and forth between both ends of the pipeline. Each time the pressure pulse arrives back at the valve, it hits the surface of the valve with a huge hammer blow. This is called the water hammer. The water hammer effect is well known in that field as a resonance effect in the pipeline.

NOVA: So these long-period events are like water hammers, telling you what's going on with the magma, the liquid rock?

Chouet: Yes. The long-period events are important because they reflect a process that involves the fluid. You care about the fluid. You want to know where the fluid is, what it's doing, how much pressure it's under, whether the pressure is going up or down. By looking at these long-period events, we have this direct window into the fluid.

NOVA: How so?

Chouet: The key principle is pressure, and how fast you're pressurizing the volcanic edifice. This is essentially a pressure-cooker situation. The evidence of this pressurization comes through the long-period events, which are a manifestation of pressure accumulating and magmatic or hydrothermal fluids—mostly in the form of gases—trying to move in response to this excess pressure and trying to shoot through the available fractures and cracks that permeate the edifice.

A somewhat analagous situation is what happens when you boil water in a teakettle. When the water starts to boil, you have this singing steam coming out of the teakettle. In a way the volcano is also singing its song. Individual long-period events are little chirping sounds the volcano makes while pressurizing. When the long-period events occur in rapid succession, a sustained signal results. The volcano then is literally singing its tune. This is a siren song because the volcano is telling you, "I'm under pressure here. I'm going to blow at the top."

NOVA: It must be quite exciting to see this line on a paper and be able to infer so much from it.

Chouet: It's a defining moment, because suddenly you realize the volcano is speaking to you, and you understand the language. It's a little bit like learning a foreign language. At first, you're sort of floating there, and you don't understand, you're lost. And suddenly the language is understood with clarity, and it's a completely new perspective. The idea that the volcano is sending information and that you are able to interpret this information and characterize what the volcano is doing, that's like learning a new language.

Calling Redoubt

NOVA: Alaska's Redoubt volcano "spoke" to you just before it blew in December 1989. Can you tell me that story?

Chouet: One of my colleagues asked if I would pass by his lab. He and his colleagues had some interesting activity from a volcano in Alaska that they wanted me to look at. They said, "We're seeing an increasingly rapid occurrence of these types of events. We don't know what they are. They seem to be very similar from event to event."

I looked at their records, and I could see that it had started just a short while ago. By the time we were looking, this activity was building up to one event per minute. And it was obvious that all these events were long-period events. I said, "I think you have an eruption on your hands."

This came out the blue for them, so they were a little taken aback, thinking he's a little bit cocky, maybe he's joking or something. They went back to their business, and I went back to my office. The next morning I poked my head into the lab and asked them, "Well, has this volcano erupted yet?" And they said, "For all we know, no, but if you're so sure, why don't you call the scientist in charge?"

I went to my office and called Tom Miller, who was the scientist in charge at the Alaska Volcano Observatory in Anchorage. And Tom said, "I can't speak now because Redoubt is erupting!" I put the phone down and went to tell the guys in the lab, "Yeah, it's erupting right now." Suddenly I became part of the team. I had something to say that was of interest. They put me in the loop, and we started looking at the activity from there on.

Then on January 2, 1990, seismic activity changed from a linear increasing trend to very rapid acceleration. I told Tom Miller, "I think we're in a similar situation to where we were on December 14th, and we're going to have a major eruption on our hands within 24 hours or maybe two days from now." We went back and forth, because calling an eruption meant we would have to evacuate people. There was an oil terminal roughly 40 kilometers from Redoubt, and closing down the plant and evacuating people meant shutting down the operation, with the risk that the oil might freeze in the pipeline, so we needed to be sure of what we were saying.

Tom called these people from Anchorage, and they said, "Look, just a few hours ago we took a helicopter trip over that dome, and it's very quiet. Just a little wisp of steam coming out, it looks totally dead." We had to convince them that it wasn't as dead as they thought; underneath it was pressurizing.

I think the clincher was Tom faxing them a sheet of paper that showed the very rapid increase in long-period events. They realized something was shooting up to the sky and that maybe these people know what they're doing. So they evacuated the plant around 5 p.m. and at 7 p.m. the volcano blew up. They called Tom back and told him that they thought he was walking on water.

Making it universal

NOVA: How did that make you feel after so many years of working, monk-like, on developing this model?

Chouet: It's a wonderful feeling, because you feel that you understand the language now—that the volcano is talking to you and you understand what the volcano is saying. You can actually track the whole thing. Of course, we don't know everything about volcanoes. It's going to take a long time to get to the point where we're in a position to make very accurate forecasts of the state of a volcano. But this was a very nice first step and a great feeling.

NOVA: Why just a first step? Having made a successful prediction at Redoubt, couldn't you just apply that to all volcanoes?

Chouet: You've seen it for one volcano and that's it. That's one volcano. You have to show that there's some universality and that this process is applicable to other volcanoes as well. So you move to a different type of volcano and try to understand this different type of volcano using the same kind of model and see if it works. And you discover that you understand its language. But then you realize by looking at other volcanoes that there are still aspects of the language that you don't get, so they don't quite fit within the model.

NOVA: Why not?

Chouet: Because there's infinitely more richness in nature than one can imagine. You always try to break it down to the components and simplify. Then you realize that maybe you've simplified too much, so you add a bit of complexity to the model. You don't want to add too much because if your model has too many parameters and too much complexity then it becomes as complicated to understand as it is to understand nature to start with.

So you try to keep your model as simple as possible and see how much you can explain. But you have to keep modifying it to see if you can explain these other volcanoes that fall outside the range. It's a process of continuous feedback between observation and theory and modeling. I find it quite exciting when volcanoes cannot be explained by the model, because it means there's additional information that is buried in there from which one can learn more. If all the volcanoes follow the behavior predicted by the model, then I'm out of a job. But fortunately, nature's rich enough so we can always keep adding to this whole thing.

“Within 20 years we should be able to make forecasts of volcanic activity that are at least as accurate as weather forecasts.”

NOVA: Yet you have had success at other volcanoes, right?

Chouet: It's a very happy circumstance when you can see that you have another volcano coming on line, so to speak, that produces the kind of behavior that you'd expect based on the model you have developed. Popocatepetl in Mexico is very good in that respect. It's working just like Redoubt and other volcanoes such as Galeras and Pinatubo were working. So it's not one volcano, one particular case you're talking about. It means you're talking about a kind of universal mechanism at play. The more volcanoes that produce this kind of behavior, the more you feel reinforced in your use of such a model.

NOVA: How close do you think you are to a universal mechanism?

Chouet: How close we are away depends on how many people work in the field. So far this is a relatively small field, so the work is done by a few individuals. I would imagine that if one received adequate financial support for the carefully designed, large-scale experiments that are required, within 20 years one should be able to resolve a lot of questions and perhaps be in a situation where one could start making forecasts of volcanic activity that could be at least as accurate as weather forecasts.

Convincing colleagues

NOVA: You spent years on your own working on this. What was that like?

Chouet: I was working pretty much alone but I was also standing on the shoulders of giants. I was borrowing from different fields and putting this into the context of my own idea of what was going on. So I was benefiting from the work of all these other people, and that's usually the case in science. Even though it took overall perhaps five or six years developing the model, more and more during that time I could actually recreate signatures out of the model that looked familiar and similar to what was observed in nature. That's when I thought, This ought to be right. It's so similar, and I can explain the richness, I can explain the duration, I can explain all these different frequencies.

You still have the work of convincing your peers. And that's hard, because people are set in their ways. Scientists are very conservative by nature. They have to be, because it takes a long time to develop theories. Once a theory has been accepted, it's been tested and tested over and over again. Then someone comes along with a new observation that doesn't fit the theory. This rocks the boat so people have to decide whether to throw away the theory or to modify it so that they don't have to re-invent the wheel.

NOVA: Why are scientists so reluctant to accept a new theory?

Chouet: For the same reason that people with different religions fight each other. Each one believes that they have a corner on the truth, and actually we don't have a corner on the truth. You have to take all these bits of information coming from many different disciplines and reconstruct something that makes sense.

Sometimes people are too narrowly focused on their discipline. You talk to some of the people making gas measurements, and they say, "Well, the only way you can find out about the volcano and where it's going is by measuring gases." In some cases the volcano is just sealed enough to allow this gas to accumulate at depth. So if you try to interpret that volcano on the basis of quantity of gas, you'd say, "Well, the volcano is muy tranquilo—very quiet, and this is a good day to go in the crater." But it would just be the opposite.

We're not working in a vacuum where we suddenly get plopped on this planet and say, "Nobody has thought about this before." You can be sure that almost any idea you have, people have thought about it before. Maybe they didn't write about it, maybe they didn't pursue it. It's very humbling, because in a sense there's nothing really to invent. There are only things to be perceived and interpreted. It's a question of awareness and saying, "Am I getting all the messages there? Am I putting all these pieces together in the proper way?" If you're not, you're not making progress.

NOVA: You've been quite modest in this interview. According to many people that we've spoken to, you have taken a bigger step forward than most, if not all. What do you feel about that?

Chouet: I feel that everyone comes around in this life blessed with some kind of gift: a gift of imagination, say, or a gift of artistic design. And you look at these people and say, "Wow, this is incredible what they did." Mozart made fantastic music, Rodin did fantastic sculptures, Einstein developed fantastic theories. What was great about these people is that they used this gift, and they left it for people to enjoy. I don't think they had to pretend that they were better than others. They were just excited by what they were doing, and they shared it with others.

I view what I do in a similar vein. I think that I have a gift which makes me a natural person to look at volcanoes and wonder about their inner workings. I'm blessed in that I've met people along the way who were very helpful in terms of providing the tools. I'm blessed with patience and persistence, because this is a long-term endeavor. I consider these to be gifts, and I would be remiss not to use them.

Deadly Shadow of Vesuvius Rate the world's deadliest volcanoes, analyze a successful evacuation, and more.

Volcanoes Talking Bernard Chouet on a seismic signal that foretells eruptions. Emergency Response Team How a crack unit of volcanologists tackles volcanic unrest abroad. Anatomy of a Volcano Distinguish magma from tephra, lava from lahar, etc. Seismic Signals Discover the hidden volcanic signatures that volcanologists seek.

Interview conducted by David Belton, BBC, and edited by Lauren Aguirre and Peter Tyson, executive editor and editor in chief of NOVA online.