New method helps predict quakes
By Paunie Samreth
Jan. 12, 2004 9:00 p.m.
UCLA seismologists are shaking things up with a new method that
may allow scientists to predict earthquakes months in advance.
The team of scientists from Russia, United States, Japan and
France included experts in pattern recognition, geodynamics and
statistical physics. They have already successfully predicted two
earthquakes, including the recent earthquake of magnitude 6.5 that
struck San Simeon, Calif. in late December.
Short-term earthquake prediction is a field that has garnered
little success; many scientists believe prediction is generally
impossible.
“People have tried a wide range of methods for many
decades and so far none of them have worked on a time scale shorter
than a decade,” said John Vidale, interim director of the
Institute of Geophysics and Planetary Physics at UCLA.
Reliable earthquake predictions could save millions of lives and
billions of dollars in damages, as well as prevent ecological
disaster.
Scientists heavily base earthquake predictions on subtle, but
detectable signals that emerge before strong earthquakes.
Scientists have studied a host of potential warning signals known
as precursors, which include foreshocks and even weird animal
behavior.
The team approached the problem of earthquake prediction as one
of pattern recognition. The goal was to find precursory patterns of
medium and low magnitude earthquake sequences that could signal a
larger earthquake in the future, said Vladimir Keilis-Borok, the
team’s leader and professor in residence at IGPP.
The researchers looked for patterns reflecting a rise in
earthquake occurrences and earthquake clustering in space and
time.
The team also looked for simultaneous occurrences of earthquakes
at increasing distances, and a rise in the ratio of medium
magnitude earthquakes to smaller earthquakes.
The detection of these patterns is utilized by the team to make
their predictions.
Predictions made decades before the actual quake are known as
long-term, while intermediate-term predictions are made years in
advance. Those made months before are called short-term, and
immediate predictions come within days or less.
Precursors are also divided by the same time intervals. Previous
studies looked for intermediate-term precursors to determine the
possibility of a large earthquake.
“In the new approach we study short-term precursors in
conjunction with intermediate-term ones,” Keilis-Borok
said.
The team considered a chain of earthquakes as a possible
short-term precursor. In the area where a chain emerged,
researchers looked for intermediate-term precursors.
“If they were present, the chain is recognized as actual
precursor, and a nine-month alarm starts for that area,”
Keilis-Borok said.
If no intermediate-term precursors were present, then the chain
of earthquakes was disregarded.
In July 2003, the team used this method to predict an earthquake
in Japan of magnitude 7 or higher by the end of December. A
magnitude 8.1 earthquake struck Hokkaido in late September.
With two successes and no misses, the team plans to continue
testing and collecting more data to improve their prediction
method.
Most seismologists are taking a wait-and-see attitude.
“Ordinarily, I wouldn’t believe it for a minute, but
since his first two guesses are right, we have to take this
seriously,” Vidale said.
“This appears to have some preliminary successes, but no
one is sure yet and time will tell,” said professor of
geophysics Emily Brodski.
The group is reluctant to release earthquake predication dates
to the general public for fear of panic and economic damage, but
they are watching for a magnitude 6.4 or higher earthquake by
September south of the Mojave Desert.
