Californians living around the Mojave Desert, Coachella Valley,
Imperial Valley and eastern San Diego County braced for a
significant earthquake, prompted by announcements from UCLA
professor Vladimir Keilis-Borok and his team of scientists.
The quake, which was set to occur by Sept. 5, didn’t
happen.
While many earthquake prediction methods have come and gone
without seeing recurring success, Keilis-Borok’s idea was
different. After all, the team successfully predicted large
earthquakes that occurred in Japan and central California in
2003.
And while the predicted earthquake for September did not occur,
an earthquake can still occur in the area unpredicted, since the
probability of a temblor occurring in the area is generally 10
percent, Keilis-Borok said.
The method Keilis-Borok employed, which relies on looking at
smaller earthquakes in the past and detecting patterns ““
along with geodynamics and chaos theory ““ led to the
determination that an earthquake of a magnitude over 6.4 had a
50/50 chance of occurring in the area between January and September
““ five times more probable than normal.
But the failure of the earthquake’s occurrence will not
halt the team’s research.
“The way we work, we accumulate our errors and use them
along with many other things to improve our physical models, to use
wider set of data, and to design more precise prediction
algorithms,” said Keilis-Borok, a professor with the
Institute of Geophysics and Planetary Physics.
“Our method is not so precise, but it is better than
methods before,” he said.
The team still plans to look at earthquake patterns and will
continue to do so for California, Japan, the eastern Mediterranean
basin and Italy.
While the failure of the earthquake to occur puts some doubt in
the effectiveness of the team’s techniques, the fact that a
significant earthquake did not occur is a relief for many.
“We will learn from our mistakes, but we will not
celebrate our successes,” Keilis-Borok said.
“Prediction of disasters is not a sport.”
Putting out false alarms does more harm than good, said John
Vidale, interim director for the institute. “We’re only
breaking even in terms of public confidence.”
“Scientifically if we can show when the danger is higher
or lower, that’s very good,” Vidale said, noting that
for the public good, prediction needs to be as accurate as 80 to 90
percent.
A sudden slip of a fault line, an earthquake results in ground
shaking and a release of seismic energy.
An earthquake prediction must be made days or hours in advance
and must predict specific cities or fault lines to be the most
accurate, he said Because the area in question was so large, it
would have been difficult to recommend emergency action, said
acting state geologist Michael Reichle, noting that if the
earthquake had occurred in the projected area, it would not have
been felt by everyone there.
“Nevertheless that area has a reasonable probability of an
earthquake happening anytime,” Reichle said.
The use of pattern recognition will not be drastically changed,
but Keilis-Borok will improve the method to produce better models
and better data, he said.
“We will learn on that mistake, but it is too early to
change methodology,” Keilis-Borok said.
While officials are still wary of adopting the Keilis-Borok
approach to predicting earthquakes, the team’s method has not
been completely dismissed.
“(Keilis-Borok’s technique) still appears to be a
legitimate approach in research,” Reichle said.
The three attempts that the team made at predicting earthquakes
so far are not sufficient enough to make an accurate judgment,
Reichle said.
Keilis-Borok still believes it is possible to make sense of
earthquakes, which, unlike storms, strike without warning.
“Chaotic systems are predictable but have
limits,” Keilis-Borok said.
“I believe we can do much better to find those
limits.”