ON THE steamy, tropical slopes of Arenal volcano in Costa Rica, clouds of
moisture hang above the trees like a blanket. There’s danger in the air too: the
volcano’s dark cone looms over nearby towns and villages, and has been erupting
almost continuously for three decades. Not on the face of it a place that anyone
in their right mind would want to climb.
But in 1992, geophysicist Milton Garces did just that, scrambling up Arenal’s
rainforested slopes to cock an ear at the volcanic vent at the summit. “I sat
there in the fog and the rain and listened for about a week,” he says. “I heard
explosions, I heard hissing, I heard this remarkable chugging sound, like a
train. Through all that time I was thinking that I could use these sounds to
find out what is happening inside the volcano.”
Not many people would have lingered so long on an erupting volcano. For
Garces, however, Arenal’s angry growls sounded like a whole new way of studying
volcanoes. Do the sounds volcanoes make before they blow their top change in
some predictable way? Could those sounds help scientists monitor active
volcanoes and forecast their deadly eruptions?
Advertisement
Inspired, Garces teamed up with Michael Buckingham, an expert in underwater
acoustics at the Scripps Institution of Oceanography in San Diego, California,
and set to work developing a computer model that could help answer those
questions. The result is the equivalent of a stethoscope for volcanoes. Like a
doctor listening for the irregularities in a heartbeat that would betray a
faulty valve, they are homing in on the subtle variations in the boiling hearts
of volcanoes that show, for example, whether a river of fresh magma charged with
explosive gases is surging up from the depths.
Garces, now based at the University of Hawaii at Manoa, has listened to
volcanoes from Antarctica to Italy. And his latest results confirm that a
volcano’s deafening clamour has a promising future in the troubled area of
predicting eruptions.
Volcanoes first caught his attention while he was studying underwater
acoustics. In 1989, Garces went to the Scripps Institution to work with
Buckingham. Here, he became involved in research aimed at detecting volcanic
eruptions on the seabed. Underwater volcano hunters, he learnt, wanted to
distinguish these eruptions from the confusion of background noise in the ocean.
But what is it that makes a volcano sound like a volcano, and what could these
noises reveal?
Garces became intrigued by the question, but realised that studying the
acoustics of underwater volcanoes wasn’t simple: most eruptions occur kilometres
down on the seabed, and learning their secrets would be difficult and expensive.
There had to be another way.
Volcanoes on dry land produce a cacophony of noise, ranging from high-pitched
whooshing and whining to low rumbles of infrasound with frequencies of less than
20 hertz, outside the range of human hearing. “These are the waves that
physically shake you,” says Garces—much like the roar that you feel in
your chest when a jet passes close overhead.
Infrasound, Garces learnt, can reveal a volcano’s innermost secrets. Some
infrasound is generated near the surface by exploding bubbles of gas and
rock— volcanic “belches” that throw magma and other debris into the air.
But infrasound also comes from the very heart of a volcano, produced by the
turbulent fluid dynamics of magma flows deep beneath the surface. Here,
geophysicists believe that oscillating streams of magma or gas bubbles expanding
and contracting with changes in pressure may give out deep growls and bellows.
Since the magma has a different density to its surroundings, the molten rock
traps these sounds and acts like a waveguide, channelling them up through the
magma-filled passage called the conduit to the volcano’s vent.
These waves of infrasound force the surface of the magma to vibrate, says
Steve McNutt, a volcano seismologist from the Geophysical Institute at the
University of Alaska in Fairbanks. “It behaves like the head of a drum,
radiating sound into the air,” he says (see Diagram).
This infrasound is a complex mixture of low frequencies. And because some of
it comes from deep inside a volcano, Garces believes that it contains vital
information about the processes that trigger eruptions—maybe even telling
you how likely a volcano is to blow its top: “There’s just an ocean of
information in these low frequencies.”
Geophysicists already eavesdrop on low-frequency vibrations using
seismometers. Many of the processes that generate infrasound inside a volcano
also send out waves known as volcanic tremor that race through the ground.
Measuring these waves can tell geophysicists much about a volcano, but so far,
attempts to use changes in the amount of tremor as a warning of eruption have
given mixed results. “It’s been useful, but there are big variations in how much
tremor you see before an eruption,” explains Bruce Julian, a seismologist at the
US Geological Survey at Menlo Park in California. “Sometimes you see none at
˛ą±ô±ô.”
Cracks and fissures
No one knows why this is. But Garces believes that part of the problem lies
with features in the ground beneath a volcano that scatter and distort the
seismic waves. “The waves are radiating through a very complicated path that has
layers, fissures, and cracks,” says Garces. On the other hand, he argues, if you
“listen” to a volcano’s rumblings by replacing seismometers with microphones,
the task becomes considerably easier. “Do your measurements in the atmosphere
and you get a much more accurate measure of the physics of the volcanic system,”
says Garces. “The air is a much simpler material.”
The first step in deciphering the complex language of a volcano was to model
its physical structure mathematically, using parameters such as the size of the
conduit, the rigidity of the bedrock and the gas content of the magma. The model
generated an infrasound spectrum that they could compare with the real thing:
tweak the parameters until the two spectra matched, and they should have a good
idea of what goes on inside the volcano.
Model number one was based on simple physics. To start with, they decided to
recreate the short, sharp explosive blasts that occur at the summit of an
erupting volcano. The first moment after a blast, a strong shock wave radiates
in all directions, like the sonic boom of a jet. This is followed by a low,
infrasonic rumble that tapers off gradually.
Garces and Buckingham represented the conduit of a volcano as a cylinder of
rock filled with magma. The sharp, percussive boom of the initial blast comes
from a sudden burp of gas that explodes near the surface. This creates acoustic
waves that ripple out into the magma, bounce around inside the conduit and then
die away—producing the long, low rumble that subsides slowly.
Mathematically, this model is almost identical to an organ pipe. Blowing air
into an organ pipe creates a broad spectrum of vibrations that bounce back and
forth inside. Some vibrations have wavelengths that match the length of the pipe
exactly, forcing it to vibrate in resonance. These become louder, while other
vibrations with wavelengths that don’t match the length of the pipe are lost.
Decode this spectrum of sounds and you can learn all about the organ pipe that
created them.
By 1994, the two were ready to test their model against real explosions. They
chose one of the world’s most studied volcanoes, on Stromboli, an island among
the Italy’s Lipari group. The volcano on Stromboli has been active more or less
continuously for about 2000 years. Because it is so well understood, there were
real numbers to feed into the model. For example, to represent Stromboli’s upper
conduit where the explosions occur, Garces used a cylinder 10 metres wide and
100 metres deep. He then added other basics, such as the physical properties of
the magma and the strength of the rock surrounding it.
This accounted for the organ pipe, but the model also needed acoustic waves
for resonance—and this required explosions. The source of explosive
eruptions at Stromboli is believed to be a constriction in the conduit, about 60
metres below the volcanic vent. The constriction acts like the nozzle on a spray
can. As magma flows upwards through it, the pressure drops suddenly and volcanic
gases come out of solution. This creates gas bubbles that expand quickly. As
they grow, changes in pressure force them to oscillate rapidly, blasting
acoustic waves into the conduit. “It’s like high explosives detonating
underwater,” says Buckingham. They based this aspect of the model on research on
underwater explosions.
Later that year, Garces travelled to Stromboli to record some real explosions
to compare with their simulation. With help from the Vesuvian Observatory in
Naples, he set up a listening station about 150 metres from an active vent on
the west side of the volcano. For eight days, Garces recorded thousands of
booming explosions and the long rumble of tremor with an infrasound
microphone.
Back in the lab, Garces and Buckingham tweaked the parameters of the model
slightly—adjusting the strength of the explosions, for instance, and the
amount of gas dissolved in the magma. They were able to re-create the infrasound
fingerprint of Stromboli’s explosions in striking detail. Buckingham admits that
he was surprised that they got such a close match, especially as they were using
such a simple model. “Actually, I was amazed,” he says.
Garces and Buckingham feel pretty confident that their organ pipe model is on
the right track. The four strongest peaks at low frequencies matched, as did the
peaks at higher frequencies. It also matched a conspicuous valley in the
spectrum that related to how far down in the magma column the explosion source
lies. “It’s not just one feature—it’s a combination of them,” Buckingham
says. “Sure, you can get one of them to work, but to get more than one is more
than coincidence.”
But even then, something was missing. While Garces was listening to Stromboli
with his microphones, a network of seismometers on the island were picking up
the steady background hum of volcanic tremor. The whole island was shaking.
When Garces compared the spectral fingerprint of these tremors to the
explosions he recorded, he discovered that the patterns were completely
different—as if one throat was singing with two voices. How could this be?
“This puzzled me for a very long time,” Garces admits.
Eventually, he came up with an explanation. Perhaps the magma in the conduit
had separated into two layers, like a thick film of oil floating on top of
water. If the densities were sufficiently different, acoustic waves would be
reflected at the interface. This would turn the volcanic conduit into two
separate organ pipes, one on top of the other. Reverberations from explosions at
the surface would be trapped in the upper pipe, while infrasound from tremors
deep in the earth would be locked in the lower pipe.
Small gas bubbles, released far below the conduit, are the key to the
layering effect, says Garces. The most abundant gases in magma are water vapour
and carbon dioxide. As the magma rises, its pressure drops and, just like a
freshly-opened bottle of champagne, the gas comes out as bubbles. Carbon dioxide
leaves the magma first since it doesn’t combine readily with molten rock. But
the magma only relinquishes its grip on the water in the last few hundred metres
of the conduit. The result is a light, frothy layer of magma that is filled with
bubbles of steam, sitting on a layer of denser molten rock.
Loud and clear
But even this complex picture is an oversimplification. In 1997, Garces went
back to Arenal with two seismologists from the University of California in Santa
Cruz, Michael Hagerty and Susan Schwartz, to record more tremors and
explosions.
What they found was curious. Arenal’s frequent explosions of gas and ash
always came through loud and clear, but the low growl of the tremor was fickle.
Sometimes it seemed muffled or undetectable, as if a thick blanket had been
thrown over the volcano, and at other times the low rumblings were clearly
detectable.
Garces believes Arenal may operate in two modes. When fresh magma flows
slowly up into the conduit, the frothy, spongy layer has time to form on top as
the gases fizz out. As they’d already discovered, this traps infrasound beneath
it in the denser layer. But if the flow of magma is more vigorous, there’s no
time for the frothy layer to form. That means that the infrasound from tremor
can pass freely through the conduit to the vent, and the microphones can pick it
up clearly. It seems that a volcano can behave like one or two organ pipes,
depending on how fast magma flows through the volcano.
Complex behaviour like this seems to present Garces’s model with a problem.
There are thousands of volcanoes on the planet, and they’re all as individual as
people. They erupt different kinds of magma with different acoustic properties
and their plumbing systems are as varied as the sounds they make. “I think
volcanoes do all kinds of things to make noise,” says Julian. If no two
volcanoes are alike, what chance is there of using infrasound to forecast
eruptions?
Fortunately for the researchers, it seems that there’s a very good one. The
infrasound signature of a volcano is especially sensitive to the amount of gas
locked up inside its magma. And this gas is critical to the way a volcano
behaves. Gases dissolved in the magma provide the driving force behind an eruption
(“When volcanoes get violent”, żěè¶ĚĘÓƵ, 26 October 1996, p 28).
The danger is that this gas will come out of solution very fast, burst
through the vent in one gigantic blast, spewing lava, rocks and other debris
onto nearby towns or villages. “It’s the most important parameter. The more gas
trapped inside, the more explosive an eruption might be,” says Garces.
This is why Garces’s acoustic model is so promising. If infrasound tells you
that the volume of gas in the magma is increasing rapidly, it could warn you of
an impending explosive eruption. “It would be extremely important if you could
tell that sort of thing,” says Julian. New data seem to confirm that forecasting
eruptions using infrasound may soon be possible. In May last year, Garces spent
some time recording the boom of infrasound on Sakurajima volcano on Kyushu, one
of Japan’s southern islands. He found that he could pick out clear changes in
the character of the infrasound spectrum before an eruption. “You could
definitely see changes in the acoustic signal,” he says. “If you can correctly
interpret these changes, you could issue a warning.”
In the next few years, Garces predicts, volcano models will emerge that can
translate volcano sounds into information almost instantaneously. Such a
volcanic stethoscope could convert the spectral fingerprint of tremor and
explosions into the kind of measurement that would mean a lot for those living
close to a volcano—so if the concentration of gas in the magma was rising,
for instance, they would be put on alert. “It’ll be a couple of years before we
can do that,” he says. “But we’ll get there eventually.”
After years of struggling to understand what Arenal was telling him that day
in 1992 when he first climbed its slopes, Garces is busier than ever. In the
past year he’s churned out a flurry of new papers, and thanks to a tour of
active volcanoes in Japan, he has a pile of high-quality acoustic recordings to
test against the model. And even though no one knows yet whether his model
reproduces exactly what happens inside volcanoes, Garces feels he’s at least
taken the most important first step—listening to volcanoes. “Basically
it’s been a labour of love for the last five years, trying to get people to
understand that this can help us. It was quite frustrating,” he says. “But now
they’re asking me to work with them.”