Danger in the air: volcanoes have a long reach.

On Dec. 15,1989, KLM flight 867 from Amsterdam was approaching its destination in Anchorage, Alaska, when the plane flew into what appeared to be a thin layer of normal clouds. Suddenly, according to flight-crew reports, it got very dark outside and the air in the cockpit filled with a brownish dust and the unmistakable smell of sulfur. One minute after beginning a high-power climb to escape the cloud, all four of the Boeing 747'S jet engines died when the combustion within them was extinguished. When the engines spun to a stop, the generators ceased making electrical power, leaving only battery-powered instruments functional. Airspeed sensors began to give false readings and then ceased to provide data. A cockpit warning light erroneously suggested there was a fire in one of the forward cargo bays.

Only after losing more than 3 kilometers of altitude did the pilots on the crippled jet get all engines restarted. Because the aircraft's front windows looked as if they'd been sandblasted, the flight crew could see what lay ahead only by leaning near the cabin walls and peering forward through the cockpit's side windows. The pilots landed the plane and its 231 passengers safely in Anchorage, but it took $80 million--including four new engines and a paint job--to restore the aircraft.

What could wreak such havoc?

Volcanic ash.

Flight 867's encounter with a volcanic plume was one of aviations most dramatic, but it's by no means unique. More than 90 aircraft have flown through ash clouds in the last couple of decades. None of those incidents has resulted in fatalities, but experts say that damages to the aircraft probably total at least $250 million.

The system now in place to warn airlines about ongoing volcanic eruptions and the locations and altitudes of the resulting ash plumes is valuable but not perfect, and scientists are working to make it better. They're improving techniques of interpreting satellite imagery, one of the cornerstones of the current warning system. They're also developing sensors that could form the heart of ground-based networks to monitor remote volcanoes or be mounted on an aircraft to scan for ash in its flight path.

LOOK OUT BELOW Of the 1,500 or so active volcanoes on Earth, around 600 have erupted in historical times. About 60 eruptions occur each year, several of which spew ash clouds up to altitudes where they threaten aircraft, says Ed Miller, a retired airline pilot now at the Air Line Pilots Association in Herndon, Va. On about 25 days per year, he notes, there's ash somewhere around the world at the altitudes where jets cruise.

Increasing the ash plumes' peril is their typically unthreatening appearance once they're blown downwind from a volcano. "The pilots [of KLM 867] said they flew into what appeared to be a regular cloud, and then everything went to hell," says Miller. That flight's encounter occurred almost 250 km away from Alaska's Redoubt volcano, which had begun erupting the day before.

Although the largest bits of ash fall to the ground in the vicinity of an erupting volcano, what stays aloft to waft further downwind is by no means benign. Even dust-size ash particles can sandblast windshields, infiltrate sensitive instruments, and clog engines.

Besides the particles of ash--which actually are minuscule fragments of shattered rock--volcanic plumes can contain high concentrations of water vapor, sulfur dioxide, and other noxious gases. Sulfur dioxide and water vapor react to form droplets of sulfuric acid, which over the long term can fade an aircraft's paint, etch the outside surface of its acrylic windows, and create sulfate deposits in engines.

"There's no known minimum quantity of airborne ash that won't do damage," says Leonard J. Salinas, a safety expert at United Airlines' headquarters in Chicago. "We must fly through clean air," he adds. United pilots are required to fly either upwind of a volcanic plume or at least 800 km downwind of the ash cloud's known edge. Otherwise, flights are delayed or canceled. Pilots for other U.S. airlines follow similar rules, Salinas notes.

Airlines get up-to-date information about volcanic eruptions and ash plumes from a network of nine Volcanic Ash Advisory Centers (VAACs) in various countries, where government-funded scientists track the altitudes and positions of ash clouds just as air-traffic controllers keep tabs on all the jets in their areas. VAAC personnel get their information from multiple sources: satellite imagery, volcano observatories, sightings by pilots, and even media reports. Once an eruption has begun, VAAC scientists issue advisories at least once every 6 hours until the ash cloud can't be detected.

SKY EYES Instruments on Earth-gazing spacecraft spy volcanic plumes in various ways. For instance, a satellite-mounted instrument called the Total Ozone Mapping Spectrometer measures the amount of radiation scattered by the atmosphere within six narrow bands of ultraviolet light, data that reveal not only ozone but also sulfur dioxide aerosols typically spewed by volcanoes and smokestacks. This technique sometimes isn't sufficient to detect volcanic plumes, however, because ash particles can fall out of or drift away from the plume containing the telltale aerosol droplets, says Miller.

Although the instrument can monitor known volcanic clouds once an eruption has been detected, it's not effective for early warning of new eruptions because its orbit--which covers the entire surface of the Earth once a day--doesn't permit frequent observations of a particular spot.

Similarly, sensors on several other satellites scan Earth in various infrared-wavelength bands for evidence of sulfur dioxide plumes, sulfate aerosols, and volcanic ash. For early-warning purposes, scientists look to observations from geosynchronous weather satellites, each of which constantly monitors an entire hemisphere and beams data to Earth every 30 minutes or so. In the past, analysts relied on readings at two key wavelengths--11 and 12 micrometers--to distinguish ash-bearing clouds from those made up only of water droplets, says Gary P. Ellrod of the National Oceanic and Atmospheric Administration in Camp Springs, Md. Comparison of cloud brightness as measured in these two channels typically shows a clear difference between clouds containing volcanic ash and those containing only water droplets.

Ellrod and his colleagues have developed an analysis technique that incorporates data from a third, 3.9-gm, infrared channel monitored by these satellites. The minerals in volcanic ash reflect this wavelength particularly well. Results of tests at his agency, which serves as part of the Washington, D.C., VAAC, suggest the new method works well for any ash plume in daylight and for those plumes moving over large bodies of water at night. The scheme also does a better job of detecting older or more diffuse ash plumes than the method that uses only 11- and 12-[micro]m readings, says Ellrod. He and his colleagues describe the new image-analysis technique in the June 27 Journal of Geophysical Research (Atmospheres).

Another ash-detection scheme using satellite data blends information from four different wavelengths. That technique, developed by Frederick R. Mosher of NOAP/s Aviation Weather Center in Kansas City, Mo., takes the same three channels of data used by Ellrod's team and adds observations at 0.6 [micro]m, a wavelength of orange-red light. By using data from all four channels, analysts can generate an ash-cloud image that doesn't vary much between day and night. Tests suggest that this method also picks up thin cirrus clouds, however.

Soon, scientists will need to develop ways to detect ash clouds that don't depend on observations in the 12-[micro]m wavelength. That's because the scientists and engineers who designed the newest generation of geosynchronous satellites phased out that channel, which was originally used to detect moisture at low altitudes in Earth's atmosphere. Instead, the new satellites monitor another infrared wavelength, 13.3 [tm, says Ellrod. Although volcanic ash doesn't show up as well at that wavelength, he notes that he and his colleagues can still spot moderately dense ash plumes, such as those spewed from the Soufriere Hills volcano on Montserrat this summer.

Mosher's four-wavelength technique doesn't rely as heavily on the 12-[micro]m data as the three-wavelength technique does. The loss of observations at 12 [micro]m won't leave analysts unable to spot major volcanic eruptions, he says. However, they'll probably have trouble detecting thin ash clouds and low-altitude plumes at night.

A geosynchronous satellite that hovers over the Pacific still provides observations at 12 [micro]m, but the one that now watches over most of North America uses the new set of wavelengths, says Ellrod. When the current Pacific satellite eventually fails, the 12-[micro]m channel won't be available again until early next decade.

DOWN TO EARTH Even though geosynchronous satellites make round-the-clock observations and download data every 30 minutes or so, that still may not be enough to avert every ash-airplane collision. When Mount St. Helens explosively erupted in May 1980, its ash plume took just 5 minutes to reach the altitudes where aircraft typically cruise, notes United's Salinas. In that amount of time, a jet can travel as far as 65 km.

Therefore, says Salinas, an airline-industry goal is to have a warning system that can alert air-traffic controllers and airline dispatchers within 5 minutes of a volcanic eruption. Current satellites don't meet that target. In fact, he says, scientists have told him that a 5-minute warning is nearly impossible for peaks that aren't continuously watched by volcanologists at ground stations. However, a solution may be close at hand because the same sort of sensors used to survey volcanoes from space can be incorporated into ground-based equipment, as well.

For instance, scientists from Los Alamos (N.M.) National Laboratory have deployed ground-based, infrared-sensing equipment to monitor variations in pre-eruption emissions of ash and trace gases at several volcanoes. At Popocatepetl, about 70 km southeast of Mexico City, a team including the lab's Stephen P. Love was able to detect a steady rise in the concentration of silicon tetrafluoride gas emitted by the volcano just prior to eruptions in 1997, even from sites u]) to 17 km away. The gas results from a chemical reaction between hydrogen fluoride vapor and the silicate minerals in volcanic ash. Looking for silicon tetrafluoride increases with such sensors might provide a way to detect impending eruptions at Popocatepetl, says Love.

Researchers in Australia have taken the infrared-sensor technology even further. They're now testing sensors that could be incorporated into early-warning devices that monitor volcanic activity at remote sites and phone updates to scientists via satellite. A warning could go out immediately if an eruption occurs. In field trials last June, the scientists went to Anatahan, a small volcanic island about 300 km north of Guam. That peak, which hadn't shown activity in previous historical times, had erupted May 10, says Fred Prata of the Commonwealth Scientific and Industrial Research Organization in Aspendale, Australia.

During the June tests, he and his colleagues scanned the still-erupting Anatahan from a ship, a helicopter, and the island itself. With the sort of analytical techniques used to process satellite images, the researchers could distinguish the volcano's ash plume from regular clouds. The same sensors can also indicate sulfur dioxide emissions, and the devices would work day or night, says Prata. This month, the team is conducting additional field tests at Italy's Mt. Etna.

Possibly more exciting, the Australian researchers are also building the infrared sensors into ash-spotting equipment they can install on aircraft. At a jet's cruising altitude, the device probably could spot an ash plume more than 80 km away, giving pilots more than enough time to divert their plane around the danger and warn other aircraft passing through the area. The prototype system is scheduled to be tested on a NASA aircraft later this year.

The potential market for such equipment is huge. As many as 300 flights per day between North America and Europe pass near Iceland, which has at least 70 volcanoes. A similar number of passenger or cargo jets fly over or near 100 Alaskan, Russian and Japanese volcanoes on routes between North America and eastern Asia.

Onboard ash-plume sensors could provide aircraft and their passengers with extra protection. Says Prata: "With pilot reports [of volcanic plumes], ground-based sensors, and onboard ash-sensing equipment, we can pretty much solve the problem."