11/3/99

Kathleen O'Toole, News Service (650) 725-1939; e-mail: kathleen.otoole@stanford.edu

Scientists image volcano dike build-up below Earth's surface

Geophysicists have been able to construct a moving image of a molten-rock-filled dike buried off the coast of Japan ­ a technique that could prove better than current methods for predicting volcanic eruptions.

The experiment, reported in the Oct. 29 issue of the journal, Science, involved estimating the volume and flow of molten rock using data collected in 1997 by satellites and tilt meters installed to monitor a seismically active area off the Izu peninsula southwest of Tokyo. The researchers are Stanford geophysics Professor Paul Segall, his former student Yosuke Aoki, now of the Earthquake Research Institute of the University of Tokyo, Stanford student Peter Cervelli and Seiichi Shimada of the National Research Institute for Earth Science and Disaster Prevention in Japan.

The area studied has experienced seismic activity 12 times since 1978 with magma reaching the submarine surface of Earth the last time in 1989, Segall said. Global Positioning System satellites continuously record surface distortions in the region, and tilt meters, like very precise carpenters' levels, also continuously measure changes in topography.

Using data from February and March of 1997 when a swarm of small earthquakes occurred, the scientists were able to see that a crack in the Earth's crust ­ what vulcanologists call a magma dike ­ filled with molten rock and expanded over a period of days, causing the Earth's surface to wrinkle up in different places and creating a swarm of small quakes. No volcanic eruption reached the surface of the sea floor, but lava flowed into the dike at a peak rate of almost 2 million cubic meters per day, Segall said. In the past, he said, "we have been able to locate the earthquakes very rapidly and tell that something is going on, but the earthquakes can't help us estimate the volume of the dike."

The researchers fed the monitoring data into a computer model that produced images that show the crack beginning to expand on the 60th day of 1997, reaching a peak flow on days 62 and 63 and being over in about eight days. The images are blurry because of a limited number of monitoring locations, but they are sufficient also to show the dike rising closer to the surface, Segall said. The data pattern also fit well with estimates of fault motion based on the magnitude of the quakes, giving the researchers more confidence in the accuracy of their dike images.

"We're pretty excited because this offers a way to scientifically image the dike and see its shape. From the point of view of hazard reduction, it gives you some time," Segall said.

The moving picture was constructed after the fact, but the method could be used to analyze real-time GPS, tilt and other monitoring data, Segall said. "It should be possible in the future to create such pictures within an hour or two of the actual activity."

That would allow remote monitoring and an alarm that could warn experts of activity. Currently vulcanologists have to make judgments about when and where eruptions might occur based on their experience and on-site monitoring of conditions.

The initial theoretical work for the computer modeling was developed by former Stanford graduate student Mark Matthews, Segall said, for potential use on silent earthquakes or fault slips too slow to generate seismic waves. Volcanic dikes, however, proved to be the first practical use.

-30-

By Kathleen O'Toole