 |
| Cover of the Nature, 5 September 2002 issue (courtesy
of the Nature Japan) |
Seismic activities can be classified into either a major quake followed
by aftershocks or an earthquake swarm. In general, within major earthquakes
a maximum-scale quake is followed by a number of minor aftershocks.
Meanwhile, the swarm earthquake is a series of the quakes of a similar
scale to the mainshock. The pace of subsidence in such earthquake occurring
is gradual relative to aftershocks in normal earthquakes, and the seismic
activities are protracted.
The AD 2000 Izu islands earthquake that struck the northern Izu islands
showed one of the most energetic swarms ever recorded. We analysed the
seismicity data as well as the land survey of this swarm to demonstrate
that the sustained crustal deformation and increase in stressing rate
largely contribute to the occurrence of earthquake swarms.
Earthquake Swarms Produced by the Change in Stressing
Rate
Earthquake swarm has been considered to be "an exceptional phenomena"
which differs from the normal earthquakes. Dominant hypotheses that
explain the occurrence of the swarm include the immediate influence
of magma and ground water (ex. intrusion to the fault), peculiar inhomogeneity
of crustal structure and so on. As shown below, in the AD 2000 Izu islands
earthquake swarm, we found that the seismic activities were produced
by stress transfers due to crustal deformation, that was incited by
magma intrusions and extrusions. Although the mechanism of each earthquake
is the same as that of normal earthquake, an extraordinarily high stress
generated for a relatively short period rapidly elevates the rate of
earthquake occurrence, resulting in distinctive seismological behaviour.
The research result was published in Nature, the issue of September
5th, 2002 (co-authored with Dr. Takeshi Sagiya, Geographical Survey,
Japan and Dr. Ross Stein, US Geographical Survey).
Analysis of the AD 2000 Izu Islands Earthquake Swarm
The swarm that struck the northern Izu islands from June to August 2000
was one of the largest earthquakes ever recorded in the land territory
and surrounding areas, producing 7,000 shocks with magnitude ≥3 including
five magnitude ≥6 shocks. In addition to its enormous scale, a great
deal of attention was attracted to the detailed data of the seismic
activities obtained through thorough observations conducted by the researchers
of Japan Meteorological Agency and Tokyo University, as well as the
continuous GPS (Global Positioning System) observation by Geographical
Survey Institute. The displacement of the ground surface in relation
to the seismicity was monitored in almost real time (see Fig.1).
 |
|
Fig.1 GPS line-length changes
and cumulative earthquakes during the swarm.
|
Swarm Moved the Islands
- Laboratory-based Observation Using the Growth Model of Magma -
As a result of GPS observation, it was calculated that the distance
between Niijima and Kozujima islands was extended by approximately
80cm whilst that between Nijima and Toshima islands was drawn closer
by 20cm during the 2 months from June 26 to August 23, when the seismicity
was active. From longitudinal data analysis of displacement, although
abrupt changes of about several centimeters were produced by shocks
over magnitude 6, it was made clear that the distance between the
two islands were gradually extended for the two months. In order to
give an explanation to such crustal deformation, we assumed that a
vertical dyke spread 15km long by 5km wide in the waters off Miyakejima
continuously propagated for 2 months (dyke model hereinafter). The
dyke eventually opened by 20m. Based on this dyke model, the shear
stressing rate of the surrounding earth's crust was calculated as
a strike-slip fault that is common in the most earthquakes. Consequently,
the area was divided into two in accordance with the resulting variation
of stressing rate (Fig.2 bottom The warm and cold colors indicate
the increase and decrease of the stressing rate, respectively. ).
We estimated that the stressing rate in the offshore of Miyakejima,
close to the dyke was 10Mpa/year or over and it reached several Mpa
a year even on Nijima island, away from the dyke. Based on the GPS
observational data for the several years before the seismic activities,
the normal stressing rate is estimated at 0.01Mpa/year. That means,
more than a 1000-fold increase in stressing rate occurred in the area
of north west offshore of Miyakejima whilst a several hundred fold
increase occurred in the surrounding area of Nijima. The sustained
stressing rate was higher than normal. Meanwhile, most of the observed
shocks occurred in the area of increased stressing rate (Fig. 2 bottom).
Comparing the occurrence rate to that of the normal earthquakes, it
was observed that the change of the seismicity rate is more proportional
to the stressing rate variation. This finding conforms to a theory
obtained from the fault friction experiment. This theory, although
being quite straightforward, was verified for the first time by our
study, suggesting the possibility to apply it to prediction of an
earthquake. Through the observation of the swarm of this time, it
was revealed that:
1. The area of seismic activities expanded over
time;
2. The duration of aftershocks of earthquakes magnitude ≥6 was
extremely short.
These results also confirmed our laboratory-based fault friction
theory.
 |
 |
|
Fig.2 Two types of seismicity
produced by a stress step or an increase of stress rate.
|
Increase in Stressing-rate and Seismicity
As shown in the study of the Izu islands earthquake swarm, it has
become easier to calculate the shift of crustal stress by means of
the computer assisted numerical analysis, in spite of the difficulty
in calculation/estimation of the absolute stress value. Another major
clustering element of seismic activities, that is "majorshock
and aftershocks" can also been explained by the similar analysis.
Conventionally, the academic attention has been drawn to the aftershocks
provoked on the source fault. However, recent findings indicate the
influence of majorshock to the seismic activity away from the earthquake
source. The shift of seismic activities in the areas away from the
epicenter that was developed before-and-after the mainshock is not
necessarily consistent. We observed both increase and decrease in
seismic rate, depending on the area. In case of the Izu islands swarm,
where the deformation was produced by dyke intrusion, the remote triggering
of the seismic center can be explained by the calculation of static
stressing rate changes. In the example of the Kobe earthquake in 1995,
the stressing rate changes extended to the latent shear faults in
the surrounding crust that were triggered by the mainshock were calculated
as shown in Fig. 2, top. The earthquakes that occurred during the
following 18 months (aftershocks in a broad sense) were also plotted.
Although there are exceptions, an earthquake is likely to occur frequently
in the area where the stressing rate increased. In contrast, the area
where the stressing rate decreased, reduced earthquake activity is
observed. It should be noted that even the slightest variation in
stressing rate (less than atmospheric pressure) can cause a significant
change in seismic activities. This suggests the sensitivity of constitutive
balance of the crust that may be hovering around the critical level.
The achievements of the similar researches upon the other major earthquakes
around the world have manifested the correlation between the abrupt
change of stressing rate produced by the mainshock and seismic activities.
Possibility of Earthquake Prediction
As indicated above, the chain earthquake can be categorised into two
groups depending on the type of stressing rate change (Fig.2) .
1. Mainshock-aftershocks type: The sudden increase
in stressing rate triggers temporal bursts of seismic activities,
i.e. aftershocks which decay over time.
2. Earthquake swarm type: The seismic activities increase in proportion
to the gradual and sustained increase in stressing rate.
These properties of stressing rate and seismic behaviors can be applied
to the probabilistic prediction of sequencing seismic activities such
as aftershocks and swarms. The experimental attempt of aftershock
prediction has already started. Earthquake prediction/forecast has
still a long way to go for contemporary geoscience. However, we are
approaching the perception any earthquake occurrences are not individual
phenomenon acting in isolation but rather being interrelated to one
another within the crust. In that sense, there is plenty of scope
for anticipating the realization of earthquake prediction in the future.
Faulting Behavior Modeling
Team, Active Fault Research Center, AIST
Shinji TODA
|
|
No.7
Winter 2003
back
> 