This past summer Montserrat, the 40-square-mile British colony in the Leeward Islands of the eastern Caribbean, was devastated by the eruption of its Soufrière Hills volcano. It first began spitting steam and ash on July 18, 1995; over the next two years it swelled, trembled violently, and began to explode. It gasped out ash plumes that rose above 40,000 feet before blanketing the island, and it sent pyroclastic flows—burning streams of volcanic rock, ash, and gas—flying down its flanks. Finally, on June 25, Soufrière Hills unleashed a particularly massive pyroclastic flow that devastated nine villages and killed at least ten people, with nine more missing and presumed dead. In early August, more flows virtually destroyed the capital, Plymouth, which lies near the foot of the volcano in the southern part of the island.
By September local authorities had declared the southern two-thirds of Montserrat off-limits, and more than half of the 11,000 inhabitants had been evacuated from the island. It’s not likely they’ll have much of a home to return to—if they ever can. Volcanologists at the Montserrat Volcano Observatory (established after the July 1995 eruption) predict that a major explosion, one that would shower the entire island with ash and rock, is imminent. The scientific consensus, according to a report issued by the mvo, remains that this is very likely to be a long-lived eruption that could last for years.
Volcanologists still have a long way to go before they’ll be able to predict eruptions precisely and accurately. But last May a team of geophysicists from Stanford reported the first successful test of a promising new system—one that relies on satellite measurements—at Kilauea on Hawaii. Susan Owen and her colleagues deployed a network of 13 stations on either side of the rift that surrounds the Napau crater. The stations are tied into the Global Positioning System of satellites, which can be used to calculate their positions to within half an inch. The relative change in distance of the stations tells the researchers how the ground around the volcano is shifting—for instance, as it swells with magma before an eruption.
Eight hours before Kilauea’s January 30 eruption—which didn’t kill anyone—the gps data did indeed show that the rift was expanding. By the time of the actual eruption, the crack had opened by nearly eight inches. The Stanford team couldn’t predict the eruption, unfortunately, because their current system gives them data only after a 24-hour delay. If we had this set up in real time, and we saw that there was a significant amount of extension going on in a particular place, says Owen, we’d be able to say with fairly good certainty that there would be an eruption. She and her colleagues hope to have such a system in place on Kilauea within a year. Montserrat might still be able to use one, too.