During the summer of 2000, a burst of 7,000 M>3 shocks--including
five damaging M>6 shocks--occurred 75 miles south of
Tokyo, beneath the Pacific Ocean. We believe the total energy
release during this two-month-long 2000 Izu Islands swarm
is the highest ever recorded. The energy release is almost
10 times larger than the 1980-82 Long Valley, California,
swarm (near Mammoth Lakes, California). The rate of energy
release at Izu is about 100 times rate in Long Valley.
We infer that the swarm was caused by a blade-like injection
of molten rock into the earths crust over an event of
10 miles long and 10 miles deep. Fortunately, the magma did
not reach the earth's surface, but the blade (or 'dike') was
forced open by the magma pressure a total of about 65 feet.
The magma got close enough to the surface under Miyake volcanic
island to trigger several steam and debris eruptions, which
caused the island to be evacuated.
During a volcanic crisis, if we have continuous GPS data and
a good understanding of the geometry of the intruding magma,
we can forecast the rate and likely distribution of damaging
(M>6) earthquakes. Such a continuous network of GPS
receivers now exists in several volcanic sites. (The Plate
Boundary Observatory proposal by the NSF, NASA, and the USGS
to Congress would place them across the entire western U.S.)
Where Do Our Results Apply?
Volcanic areas, such as Hawaii, the Pacific Northwest, Alaska,
Yellowstone, and parts of northern and southern California.
Most earthquakes are followed by aftershocks that are smaller
than the mainshock and become less frequent with time. But
some earthquakes occur in swarms--a sustained high rate of
seismicity that eventually stops or peters out. Swarms are
common in volcanic areas, but sometimes occur on tectonic
faults such as the San Andreas. Why? We offer a new explanation
for the occurrence of swarms, attributing them to a sustained
increase in the rate at which the crust is stressed. We attribute
mainshock-aftershock sequences to a sudden but permanent increase
in stress, rather than a change ins stressing rate.
How Do We Use the Izu Swarm To Make
During the swarm, the calculated rate at which the surrounding
crust was stressed increased by a factor of a thousand, triggering
earthquakes more than 25 miles from the dike. The swarm was
well recorded by Japanese instruments. We use this extraordinary
occurrence to test a key prediction of the theory of 'rate/state
friction,' (by Jim Dieterich of the USGS), which if validated
can be used to forecast seismicity and earthquake hazards
in all settings and countries. This theory predicts that the
rate of earthquakes increases in direct proportion to the
rate at which the crust is stressed, just as we observed.
The theory also predicts that the duration of aftershocks
of the M=6 mainshocks during the swarm will be shortened as
the stressing rate increases. We found that aftershocks M=6
earthquakes at Izu normally last for a year; during the swarm,
they lasted a day, in accord with the theory.