NOVA Online | Deadly Shadow of Vesuvius | Can We Predict Eruptions?

Eruption of Mt. Pinatubo, Philippines, June 1991.

Can We Predict Eruptions?
by Peter Tyson

In a word, yes. But that assertion, like saying we can predict the weather, bears significant caveats. Volcanologists can predict eruptions—if they have a thorough understanding of a volcano's eruptive history, if they can install the proper instrumentation on a volcano well in advance of an eruption, and if they can continuously monitor and adequately interpret data coming from that equipment. But even then, like their counterparts in meteorology, volcanologists can only offer probabilities that an event will occur; they can never be sure how severe a predicted eruption will be or, for that matter, whether it will even break the surface.

Still, under ideal conditions, volcanologists have recently met with a great deal of success in foretelling eruptions. While they were caught off guard by the exact timing and magnitude of the 1980 Mt. St. Helens eruption, for example, their timely warnings of an impending blow prompted the U.S. Forest Service to evacuate people from dangerous areas near the volcano. Though 57 people died in the eruption, perhaps 20,000 lives were saved, says Dr. William Rose, a volcanologist at Michigan Technological University. Similarly, a USGS SWAT team that rushed to the Philippines' Mt. Pinatubo in the spring of 1991 successfully augured the June eruption, leading to evacuations that saved thousands if not tens of thousands of lives and millions of dollars worth of military equipment at the nearby Clark Air Force Base.

A Kilauea lava fountain spews from Puu Oo vent on March 13, 1985.

Not surprisingly, volcanologists have had the most success at volcanoes that host their own observatories. In 1912, Thomas A. Jaggar, head of the Geology Department at the Massachusetts Institute of Technology, founded the first volcano observatory in the United States on Kilauea. (There are now three others—in Menlo Park, California; Anchorage, Alaska; and Vancouver, Washington, near Mt. St. Helens.) Over the succeeding decades, researchers at the Hawaiian Volcano Observatory developed many of the techniques used today and can now predict Kilauea's eruptions to a tee.

They know when and how Kilauea will erupt because it does so frequently and predictably, and because after decades of intensive study they know the volcano inside and out. Learning as much as possible about a volcano's previous behavior is the essential first step in anticipating future blows, just as knowing a career criminal's record can help indicate what he might do next. "There is no doubt that the eruptive history of a volcano is the main key for long-term prediction," says Dr. Yuri Doubik, a Russian volcanologist who has studied past eruptions on the Kamchatka Peninsula for 35 years. Such work entails laboriously picking through the physical remains of previous eruptions. And mapping such old lava flows, pyroclastic deposits, and other volcanic debris distributed around a crater can reveal much about the timing, type, direction, and magnitude of previous blows.

Close-up of a seismograph drum.

Satellite data can greatly aid such mapping, and volcanologists are looking forward to using images generated by the Earth Observing System after it is launched in 1999. The satellite's purpose is to study environmental ills such as global warming and depletion of the ozone layer, but it will also gather information of use to volcanologists, including gas concentrations in the atmosphere over volcanoes and images clear enough to reveal the fallout from former eruptions.

The Volcanologist's Toolkit

When a volcano's eruptive history is known, researchers can more confidently turn to modern techniques to help them call the next eruption. The most valuable among these, volcanologists agree, is monitoring a volcano's seismicity—the frequency and distribution of underlying earthquakes. Use of the seismologist's tool in volcanology has come a long way since Frank Perret, one-time assistant to Thomas Edison, gleaned the frequency of the small shocks that continually shake Vesuvius's flanks by biting down on the metal frame of his bed, which was set in cement. Today sophisticated seismographs can register the magnitude, escalation, and epicenters of earthquakes that occur as magma moves beneath volcanoes. The more seismographs technicians deploy on a volcano, the more complete the picture they get of the mountain's plumbing.

Installing tiltmeter, with Tavurvur in distance

Tavurvur volcano erupts in the distance as workers install a tiltmeter at Rabaul, Papua New Guinea, September 1994.

Seismic networks can transmit data by radio 24 hours a day to computer-equipped monitoring stations well out of harm's reach. This enables scientists to safely watch for changes in "nature's noise," as one volcanologist labeled the geophysical status quo within a volcano. Computer-based seismic data acquisition and analysis systems, which in essence constitute portable observatories, enabled the USGS Volcano Disaster Assistance Program's crisis-response team to successfully predict the 1991 eruption of Mt. Pinatubo. Such "mobile observatories" themselves now constitute a major weapon in the prediction arsenal.

While seismicity is the workhorse, monitoring ground deformation is another up-and-coming technique that allows three-dimensional mapping of what's occurring underground. Magma rising from the depths often pushes the skin of a volcano up and out, like a balloon filling with air. Sensitive tiltmeters and surveying instruments can measure and record the slightest changes, which help volcanologists determine, for example, roughly how deep a magma source is, how fast it is moving, and where on a volcano it might erupt. Such monitoring has helped scientists anticipate eruptions at Hawaii's Kilauea and Mauna Loa volcanoes, which deform in predictable ways and at predictable rates.

A USGS team prepares to fly a gas-measuring flight at Montserrat, August 1995.

One drawback is that studying ground deformation has required scientists to climb volcanoes to take measurements—a perilous undertaking. But USGS volcanologists are now testing a prototype of a fully automated ground-deformation system. Flown aboard satellites or aircraft, the so-called "synthetic aperture radar" can automatically and continuously transmit information on a volcano's ground movements to remote observatories. Though it will not penetrate dense vegetation and is sensitive to moisture, the radar provides a resolution of less than an inch under ideal conditions. "It's a tremendous tool because it gives a complete map of ground movements, and we don't have to go into the field to get it," says Dr. Dan Dzurisin, a geologist with the USGS Volcano Hazards Program (VHP) who is helping to perfect the new device. His colleague at the VHP, the volcanologist Dr. Robert Tilling, is equally optimistic: "We're confident that by the turn of the century, we'll have such a system and at low enough cost that it can be applied easily everywhere in the world."

Measuring Vapors

Such is the long-term hope as well for techniques to monitor volcanic gases. Magma deep underground lies under enormous pressure, which keeps vapors dissolved. But as magma rises toward the surface, the pressure eases and gases such as carbon dioxide and sulfur dioxide begin to bubble out of the liquid rock and into the air. Theoretically, changes in concentrations of CO₂ and SO₂ emitted by a volcano can be used to predict eruptions, as can the escalating output of gases in general. The USGS team that was sent to Pinatubo in the spring of 1991 successfully predicted the June eruption in part after watching SO₂ levels shoot up to unprecedented levels of 16,500 tons per day.

Alaska's Spurr volcano blows its top on August 19, 1992.

Monitoring of volcanic gases got its start in the 1950s when enterprising Japanese researchers put beakers of potassium hydroxide, a strong, basic solution, on Honshu's Asama volcano, which was beginning to show signs of erupting. As the highly acidic gases released by the crater seeped through holes in a crate covering the beakers, they increasingly altered the solution's composition in the months before a large eruption. Today, volcanologists use so-called "Japanese boxes" routinely, though again they must check the beakers manually. To surmount this problem, Dr. Stanley Williams, an Arizona State University volcanologist who was nearly killed during a small but deadly eruption of Colombia's Galeras volcano in 1993, is designing an electronic Japanese box that will automatically and continuously transmit data to a remote observatory. About the size of a briefcase, the battery-powered unit has tiny electrochemical sensors that create currents proportional to the amounts of various volcanic gases in the air.

Concurrently, Williams and others are working on infrared telescopes to monitor concentrations of gases escaping from volcanic vents. Williams's version is modeled after the correlation spectrometer, a device originally developed in the 1970s to monitor SO₂ and other toxic gases from factory smokestacks. His prototype unit measures the amount of infrared light absorbed by CO₂ molecules, from which an estimate of CO₂ concentrations in the air can be made. Dr. Kenneth McGee, a volcanologist at the Cascades Volcano Observatory in Vancouver, Washington, is perfecting an infrared spectrometer that he says will detect still other volcanic gases that absorb infrared light, including hydrochloric acid gas, carbon monoxide, methane, and water vapor.

Mt. St. Helens erupting, May 18, 1980.

Calling the Next Big One

While volcanologists feel confident that these ever-improving technologies will enable them to predict when an eruption is about to occur, they still cannot reliably estimate an impending eruption's size or exact nature. How large will the eruption be? Will it be explosive like Mt. St. Helens or effusive like Kilauea? Indeed, will it even open a vent in the surface? To be able to answer such questions, Tilling and USGS colleague Dr. Peter Lipman argued in a 1993 article in Nature for the need to develop "rugged, reliable real-time systems" to measure changes not only in seismicity, ground deformation, and gases, but also in gravitational and electromagnetic fields—in short, equipment to read the gamut of signals given out by a restless volcano. "There's no magic bullet in predicting volcanic eruptions," says Dr. Charles Connor, a volcanologist at the Southwest Research Institute in San Antonia, Texas. "The key thing is to cross-correlate as many different observations as possible."

Tilling says volcanologists also need to get a better handle on the basic mechanisms behind precursory signals, such as the long-period earthquakes that often precede eruptions. Dr. Bernard Chouet, a VHP volcano seismologist, says these quakes provide a "direct window" into the magmatic fluid moving about beneath a restless volcano. "These earthquakes are like stress gauges that light up and reflect the pressurization going on below," he says. Careful monitoring of such natural gauges can help forecast eruptive activity. The USGS team that successfully predicted Pinatubo's burst did so in part by watching the build-up of long-period quakes.

USGS Volcanologist Ken Yamashita surveys on the dome, Mt. St. Helens, March 1986.

The urgent need to improve methods to call the next Big One holds especially true for large caldera-forming eruptions. These true earth-shakers explode with such Herculean force that they leave behind vast, basin-like depressions—calderas—that can stretch many miles across. The largest caldera-forming eruptions, which fortunately have not occurred in human history, make the explosive eruption of Mt. St. Helens in 1980 seem like a firecracker. In the mid-1980s, three volcanic fields believed to hold the potential for one of these monumental cataclysm—California's Long Valley, Papua New Guinea's Rabaul, and Italy's Campi Flegrei—turned on almost simultaneously, throwing the volcanological community into a bit of a frenzy. All three centers calmed down without further ado, though Rabaul erupted a decade later (see Planning for Disaster).

Tilling, for one, is confident that such an apocalyptic blast will not come unheralded. "No volcano is going to suddenly produce one of these humongous eruptions without giving a lot of signals," he says. "But what will those signals be?"

Peter Tyson is Online Producer of NOVA. This piece was excerpted and updated from a feature article by Mr. Tyson that originally appeared in Technology Review (January 1996).

Photos: (1) USGS; (2) Jim D. Griggs; (3) C. Dan Miller; (4) Andy Lockhart; (5) C. Gardner; (6) Robert McGimsey; (7) Austin Post; (8) Steve Brantley.