講師 中谷正生（東京大学地震研究所）
題目 直球勝負—地震予知。
最も単純な地震サイクルのシナリオを基本に、「こういうことがextraにあってくれたら短期予知ができるだろう」というシナリオと、そのシナリオにあう観測事実を４つほどお話しします。準備過程の大きさということにこだわりながら考えます。
1. Introduction -long-term prediction
Abundant evidence shows that earthquakes are
unstable slip repeating on existing weak planes in the earth’s topmost part,
which is too cold (<300C) to deform stably. An earthquake occurs when the
stress on the weak plane, which increases slowly due to plate motion, reaches
the frictional strength. The earthquake (i.e. slip) relaxes the stress over the
slipped area, hence we know that there will be some time before the next
earthquake involving the whole slipped area. Although numerous small
earthquakes can occur anytime on the weak plane using a stress accumulation of
small spatial scale, their summed effect on the average stress over the area of
the large earthquake is negligible. In addition, we know where the major weak
planes are located geographically, (plate boundaries and mature faults within a
plate). These understandings leads to the following long-term earthquake
prediction. Large earthquakes occur only where large earthquakes occurred
before. After a large earthquake happens in one place, the next large one would
not occur for, say, at least 50% of the average repetition time at that
locality.
2. Hopes for short-term prediction.
The average repetition interval is of order
100 years for plate boundaries and 1000-10000 years for intra-plate faults.
This is the time for accumulating ~10 MPa of stress (difference between static
friction and dynamic friction). If we want to predict a plate-boundary
earthquake with, say, 1-year accuracy, we need to know the strength and the current
stress level with an accuracy of 0.1MPa, which way we do not know. So, we hope that
Mother Nature has something more in favor of us, which we call ‘earthquake
preparation processes’. Hopes are classified as: A) Material of the weak plane
may come into a funny state (or even breaks and slips partially) when the stress
comes very close to the strength. B) Strength may be decreased (over a large spatial
scale) by an external factor pretty quickly. C) Stress accumulation mechanism
may behave in a more complex way than steadfast loading at a constant rate.
Maybe, it accelerates at one point of time, quickly raising the stress on the
faults to its strength.
In this talk, I would like to discuss how possible
preparation processes cause ‘precursory anomalies of large earthquakes’.
Statistical significance of each type of anomaly is in grey-zone. However, each
earthquake can take different paths to the final failure, so taking statistics
for ‘earthquakes in general’ is not the most productive thing to do. My
approach is rather to look for physically plausible preparation processes,
betting on that many earthquakes do not occur all the sudden at the critical
stress level.
3. Concrete models of earthquake preparation processes.
In the talk, observations attributable to the following scenarios will be presented.
A) Failure of inhomogeneous materials, including frictional slip, is invariably preceded by changes of the state of the material, which is often an inevitable part of the physical process toward the final failure. An explosive increase of micro cracks is a typical precursory change, which can induce many observable anomalies, amplified by resultant groundwater flow.
A’) The above precursory
processes start when the local stress approaches the local strength. However,
it is not realistic to expect this occurs simultaneously all over the extent of
the eventual rupture plane of the coming large earthquake (say, 20 km x 100
km). It would be more realistic to assume that precursory failure occurs in a
limited part (called earthquake nucleus) of the future rupture plane and the
rest of the rupture occurs by a sudden increase caused by dynamic rupture
expanding from the nucleus. Rupture mechanics guarantees that nuclei grow
spontaneously (i.e. without further increase of the applied stress, thus within
a time period much shorter than the repetition interval) but quasi-stationarily
to a certain size. The critical question for prediction is whether the size of
nucleus and that of the final rupture are correlated.
B) The above scenarios are
beautiful in that the preparation processes are inevitable part of the physical
process to final failure. However, Mother Nature may be less interesting but
more favorable for us. For example, if high-pressure fluid is injected into the
pores on the weak plane, its strength drops. A 1 MPa increase of fluid pressure
causes ~0.7MPa drop of strength, easily affecting the earthquake occurrence. An
advantage of this scenario is that strength can be decreased
quasi-simultaneously all over the extent of the future rupture plane. (More
like, the dimension of the weakened part decides the size of the coming
earthquake.) Note that many reported precursory anomalies require that
preparation process occurring over a wide area comparable to the final rupture.
C) Stress accumulation on the
weak plane seems to be caused by a stable slip of the deeper extension of the
plane. If it accelerates at some time, this can be the time for earthquake.
Rapid loading by a nearby occurrence of large earthquake can have the same
effect.
Conclusion
In my opinion, short-term
prediction is possible only if earthquake is preceded by special preparation
processes beyond just an increase of stress at a constant rate. It is
physically plausible, though otherwise is also plausible. Useful preparation
processes are those that scale with the size of final failure.
References
Scholz, 2002, The mechanics of
Earthquakes and Faults (2nd ed.), Cambridge Univ. Press.