After several months of explosive eruptions, the Japanese volcano Mount Unzen
seemed finally to have calmed down by May 1991. Though lava was still on the
move, it stayed below the surface of the crater, sluggishly building up into a
mound. The locals breathed a sigh of relief, convinced that the worst was over.
They were wrong. On 3 June, the mound exploded. A searing hot avalanche of ash
and lava rushed from the crater at several hundred kilometres per hour,
destroying 180 houses and killing 43 people.
The fickleness of volcanoes can make them especially dangerous. Some gently
ooze lava like squeezing toothpaste from a tube, while others suddenly explode
with little or no warning. Often the same volcano can switch unexpectedly from
鈥渇low鈥� to 鈥渂low鈥� with catastrophic consequences for those living nearby.
Despite volcanologists鈥� diligent recording of earthquakes, ground deformation
and gas emissions, at best they can only warn of an explosive eruption several
hours to days in advance. They simply don鈥檛 know enough about how volcanoes
work.
So last month, 60 volcanologists and geologists met on the exploded volcano
that is the Greek island of Santorini to compare notes and thrash out new ways
to pin the problem down. Many were optimistic. If they succeed, it may even be
possible one day to avert explosions, or to trigger them early, when everyone
has been evacuated to a safe distance.
Advertisement
To understand explosive eruptions, geologists need to study the plumbing in
the volcano鈥檚 interior. Magma heads up to the surface from deep within the
Earth. In many volcanoes it pauses in a reservoir called a magma chamber some 7
to 10 kilometres below the surface. From there magma travels to the surface
along a central passageway through the volcano called the conduit, tens to
hundreds of metres in diameter.
As it travels, geologists believe that the magma changes in fundamental ways,
which ultimately determine whether it bursts violently out at the surface, or
seeps gently down the mountainside. It all hinges on the large amounts of water
vapour and other gases dissolved in the liquid magma. As the molten rock travels
towards the surface, the pressure drops and the vapour forms bubbles. If the gas
leaks out gradually, nothing remains to power the explosion and the lava simply
flows. But if gas stays trapped in the magma until it reaches the surface, it is
suddenly, dramatically released in an explosion. It鈥檚 like a bottle of
champagne鈥攊f the pressure inside is suddenly released, bubbles rapidly
nucleate and expand, spraying out a foam. If the bottle is opened gradually, the
suds just gently spill out.
This much has been understood for many years. But to work out the details,
and so be able to predict or even avert disaster has proved extraordinarily
difficult. After all, no one can slice open a volcano and track the gases in the
moving magma. Instead, geologists have had to resort to more indirect
approaches.
One approach鈥攗sing computers to model the processes leading to
explosive eruptions鈥攚as developed in the late 1970s. But these early
attempts were very crude and relied on many assumptions. Since then, more
advanced models have been used to test ideas about bubble growth, magma flow and
other processes inside the conduit. But in the past few years there have been
attempts to include more details about viscosity, diffusion and other factors,
in the hope of predicting how and when real volcanoes will erupt.
In 1993 geophysicists Alex Proussevitch and Dork Sahagian, now at the
University of New Hampshire, Durham, published the most sophisticated model to
date. Their equations demonstrated in detail what many researchers had
suspected: the viscosity of the magma is crucial. For gas to escape, it must
diffuse through the magma and collect in bubbles. But to form and grow, the
bubbles must battle to push their walls out into the thick, viscous magma. The
greater the viscosity, the harder it is for the bubbles to form and grow, and
the more likely it is that the gas will be trapped in the magma long enough to
provoke an explosion at the surface. If the magma is runny, as in Hawaiian
volcanoes, gas bubbles can grow and move easily through the liquid. As a result,
the gas dissipates quickly, and magma reaching the surface simply oozes down the
long sloping flanks of the volcano.
Today, Proussevitch and Sahagian have improved their model so that they can
estimate how much ash and rock will erupt, as well as other variables such as
the speed and duration of the eruption. 鈥淲e are very close to prediction of
eruption time and style,鈥� says Proussevitch.
But to make reliable predictions, the models need accurate information.
Several labs are working on ways to measure magma viscosity, at least for past
eruptions. For instance, Don Dingwell and other scientists at the University of
Bayreuth in Germany melt rock samples in furnaces and then carefully measure
viscosity by prodding the melted rock with a rod. Other groups time the descent
of a metal sphere through magma. Moreover, seismologists can roughly sketch the
size and location of magma chambers, and geochemists can get a reasonable idea
of a magma鈥檚 chemistry from certain crystals in solidified volcanic rock.
Fragmented picture
But in spite of Proussevitch鈥檚 optimism, at least one large gap remains.
Although bubble formation is certainly important, there is also the question of
exactly how the subterranean magma changes from a bubbly liquid to an explosive
cloud of rock and dust鈥攁 process called fragmentation. Without a good
physical understanding of this process, the models will always struggle. 鈥淲e鈥檙e
still waiting,鈥� says Proussevitch.
Fragmentation guarantees a violent eruption. But geologists agree only on the
basic details of how it happens. As water vapour fills the bubbles, fewer and
fewer water molecules remain dissolved to lubricate the magma, making it even
stiffer. Even the mere presence of bubbles makes the magma more viscous, just as
whipping air into egg whites thickens them. Finally, the force of the growing
bubbles exceeds the strength of the surrounding magma. 鈥淭hen you have trouble,鈥�
says Dingwell. The magma foam, now too rigid to deform safely, splits into
chunks of rock in a hot, dusty cloud of gas, which erupts from the surface at
the speed of sound.
How exactly does the magma fragment? 鈥淭hat鈥檚 a real black hole,鈥� says Oded
Navon, a geologist at Hebrew University of Jerusalem who studies bubble growth.
So far, the models tend to assume that fragmentation happens when the foam
expands and the films of magma between bubbles stretch so thin that they burst.
In Proussevitch and Sahagian鈥檚 model, the magma simply fragments when the bubble
concentration reaches 70 to 80 per cent.
But the picture could be much more complicated. An alternative theory hinges
on the fact that liquids can sometimes act like solids. For instance, glass
flows like a liquid, albeit extremely slowly, but you can still smash it if you
hit it sharply. The same is true of molten rock. 鈥淗it the stuff quick enough,鈥�
says Dingwell, 鈥渁nd it will break.鈥�
What could hit the magma and cause it to fragment? One possibility would be a
sudden, rapid expansion of the bubbles, perhaps because of a drop in pressure.
However, Steve Sparks, a volcanologist at the University of Bristol, has
calculated that this would not be enough to shatter the magma. Dingwell says
that the additional force needed to shatter it could come from other sources
such as a collapsing lava dome.
If magma breaks apart as a brittle solid, then predicting the conditions
necessary for explosive eruptions is easy, says Dingwell. Brittle failure is a
function of viscosity, which can be calculated from the temperature and
chemistry of magma. This would reveal how much and how fast the magma would have
to be deformed before fragmenting.
The issue might be settled by looking at the microscopic remnants of bubbles,
says Dingwell. If most bubble walls appear thinned, then the brittle failure may
not apply. Kathy Cashman, a volcanologist at the University of Oregon in Eugene,
says she has seen evidence of such ductile thinning in pumice (a porous volcanic
rock made of solidified foam) with a scanning electron microscope. However, most
ash fragments appear to have broken in a brittle fashion. Either those bubbles
could have fragmented as Dingwell envisages, or they may simply have broken on
their way through the air. 鈥淚鈥檓 not convinced that enough people have looked
carefully enough at enough samples to know,鈥� says Cashman.
Whichever mechanism creates large explosive eruptions, the amount of gas in
the magma is also crucial: more gas in the magma means a greater chance of
fragmentation and explosion. What鈥檚 more, different concentrations of gas in the
magma chamber could explain why one volcano can switch from 鈥渂low鈥� to 鈥渇low鈥� and
back. For instance, Mount St Helens in Washington switched between lava dome
growth and violent eruptions several times between May and October 1980, and
geologists are naturally keen to know why.
Watery puzzle
One idea is that the dissolved water could be unevenly distributed in the
magma chamber. To check this, Fred Anderson of the University of Chicago and his
colleagues examined microscopic pieces of magma from a 700 000-year-old eruption
in Long Valley, California, that were trapped inside crystals and so protected
from losing their gas. In 1989 the team reported that ash and rock from early
explosions came from water-rich magma, while later and slightly less explosive
magma was drier.
But the puzzle is that water-rich magma, which should explode, can also erupt
calmly. Lava flows from smaller eruptions at Mount Unzen and Mount St Helens
contain dark, spindly needles of hornblende, a mineral thought to form only in
water-rich magmas. The magma must have lost its water, but how?
One possibility is that gas may leak through cracks in the conduit walls.
This could render an explosive magma harmless, depending on the speed of the
ascending magma. In 1991, Claude Jaupart and Claude All猫gre,
volcanologists at the Institut de Physique du Globe in Paris, proposed that if
the magma rises quickly, gas won鈥檛 have time to escape and may explode. A
sluggish pace would create a lava dome. That seems to have occurred at Mount St
Helens, according to work by Elliot Endo, a geophysicist with the US Geological
Survey in Vancouver, Washington. Endo looked at seismic records and found that
during lava dome formation, the magma moved considerably more slowly. At
present, magma speed can only be determined after the event. But if computers
could process the seismic signals faster, it might be possible to track the
speed of the magma in real time, says Endo. If so, this could help to predict
the likelihood of an explosion.
Gas escapes
However, direct evidence that gas really can be lost in the conduit is still
thin on the ground. Mark Stasiuk of the University of Lancaster and his
colleagues have recently been looking for clues to gas escape in an extinct
volcano at Mule Creek, northeast of Silver City, New Mexico. About 20 million
years ago, the volcano erupted violently and then began building a lava dome.
Mule Creek followed a common pattern of silica-rich eruptions. Many explode,
sending up ash and pumice for hours or days before waning. Then they often
switch and slowly extrude a lava dome. However, even then portions of the lava
dome may collapse into dangerous flows of hot ash, or the lava dome can explode
unexpectedly as at Mount Unzen.
Working with Steve Tait, a volcanologist at the Institut de Physique du Globe
in Paris, Stasiuk has examined the inside of the uppermost 300 metres of the
volcano at Mule Creek. Half of the volcano had eroded away. Fractures in the
rock provide some of the first physical evidence about how gas could escape from
lava domes, says Stasiuk.
Some of the fractures at Mule Creek pass from the solidified magma into the
rock wall of the conduit. Pieces of fragmented magma inside the fractures
suggest they might have violently expelled gas. And solidified bubbles called
vesicles seem to line up in the rock. These chains of bubbles may have allowed
gas to escape when the rock was molten.
If variations in the gas content of magma really are the key to why some
volcanoes explode and why others don鈥檛, that could be bad news for the chances
of predicting a volcano鈥檚 behaviour. Mapping the distribution of gas in the
magma chamber, at least for the moment, is almost impossible, and geologists
would have a similarly hard time estimating the number of cracks in a real
conduit.
Disaster or diversion
But there could be a silver lining. Some geologists believe that one day the
danger of some volcanoes could be controlled. Proussevitch envisages tunnelling
a few hundred meters into the ground to a branch of the magma conduit.
Explosives could then open the conduit, relieving the chamber of the overlying
weight of 2-3 kilometres of mountain. An instant decompression could release the
force of the volcano months or years before a natural eruption. The goal would
be to prevent people being evacuated unnecessarily and ensure that no one is
caught off guard by the eruption. 鈥淚t鈥檚 really very possible and cheap to do,鈥�
he says.
But many geologists urge caution. 鈥淲e鈥檙e a long way off,鈥� says Sparks. 鈥淚t鈥檚
conceivable, but our level of understanding is still not very sophisticated.鈥�
One of the major hindrances is mapping the conduit geometry, which would be
critical for placing a tunnel and predicting its effects. Existing seismic
techniques do not have the resolution needed to locate the conduit accurately,
though they can find magma chambers.
Questions of feasibility aside, politics might pose an even bigger problem.
鈥淛ust imagine the lawsuits if it didn鈥檛 work out,鈥� says Stasiuk. 鈥淣obody would
want to set one off. The goal is to be able to say when and how it will happen,
then evacuate in plenty of time and watch the fireworks from a safe seat.鈥� Tait
agrees. 鈥淭he bottom line is we鈥檝e got to have a reasonable way of really
predicting the evolution of magma,鈥� he says. 鈥淭hat will determine whether a
volcano is a tourist attraction or a major catastrophe.鈥�
* * *
How to build your own volcano
IN AN explosive eruption, bubbly magma tears itself apart deep underground.
快猫短视频s have been recreating this fragmentation in the laboratory. Searching
for new ideas, they explode materials such as pine sap, freon, and carbonated
water. Some of the results have overturned long-standing theories.
A team led by Brad Sturtevant at the California Institute of Technology and
Steve Sparks at the University of Bristol has developed a sophisticated version
of an exploding champagne bottle. The team studies explosive bubble growth in
shock tubes, cylinders which are used in aeronautical laboratories to create
rapid changes in pressure. Pressure inside the tube is increased and large
amounts of CO2 are forced into the water. When a computer-controlled
knife punctures the aluminium lid, pressure suddenly drops and CO2
escapes from the water into bubbles. The foam expands and shoots out of the tube
into a large tank in fractions of a second. 鈥淚t goes whoosh,鈥� says Heidy Mader,
a physicist at the University of Bristol, 鈥渁nd that鈥檚 all you see.鈥�
High-speed cameras reveal more detail. Pictures published in 1994 clearly
showed that the foam rapidly accelerates and then fragments鈥攅xactly the
opposite of the accepted model. The experiments can also illuminate the origin
of volcanic deposits. By creating an explosion from acetone and pine sap, Jeremy
Phillips at Bristol produced stretched bubbles similar to those found in pumice,
and previously thought to have deformed while flying through the air. If the
bubbles deform while travelling up the conduit, as the experiment suggests, the
increase in surface area as a result of stretching should increase the diffusion
of water vapour into the bubble, possibly leading to a runaway effect.
Recent experiments with the Caltech equipment have created longer-lasting
eruptions. For the first time in the laboratory, Mader and colleagues have
observed bubble-rich liquid flowing up the tube in pulses. These variations in
eruption rate resemble those of real volcanoes, she says. In the past,
geologists have speculated that sudden erosion of the conduit or an elastic
magma chamber could be responsible for such pulses. But the simulations suggest
that changing patterns of bubble growth could alter the eruption strength, says
Mader.
The main challenge for researchers is applying insights from laboratory
simulations to real volcanoes. 鈥淲e can鈥檛 guarantee what we see in the lab
reflects what we see in nature,鈥� says Mader. 鈥淭hat鈥檚 an immense headache.鈥� For
example, the simulations create foams with 100 000 times fewer bubbles per cubic
metre than those observed in pumice or volcanic glass, suggesting that rates of
gas escape and foam acceleration could be much greater in real eruptions. And
the tube is far shorter than a real volcanic conduit.
Laboratory simulations have not yet led to mathematical equations describing
fragmentation, but they have provided new ideas about what happens when bubbles
rush out of solution and propel the magma to the surface. 鈥淭he most useful thing
these experiments do,鈥� says Mader, 鈥渋s make people think.鈥�