Professional Documents
Culture Documents
Aoi 2010
Aoi 2010
Southwestern Japan lies at the boundary between the occurred within 30 min from the mainshock are widely distributed
subducting Philippine Sea plate and the overriding Eurasian along and around both faults. This indicates that the mainshock
plate. A magnitude 8 megathrust earthquake ruptured the started at its hypocentre on the southern fault and the rupture
Tonankai and Nankai segments in 1944 and 1946, respectively, transferred to the northern fault.
but the neighbouring Tokai segment of the plate boundary On the basis of this two-plane-fault model, we estimated the
remained locked1 . A large megathrust earthquake in the Tokai earthquake rupture process with a multi-time-window inversion
region has therefore been expected. In 2009, a magnitude 6.4 scheme6 , using near-source strong-motion data from NIED’s
earthquake took place in Suruga Bay, within the Philippine Sea K-NET and KiK-net5 stations (Methods). Large slip areas extend
subducting plate, close to the Tokai segment. Here, we use a around the connecting point of Faults I and II (Fig. 3; Supple-
fault-slip model to examine the impact of the stress changes2 mentary Note S2). Fault I, which ruptured first, was dominated
caused by the Suruga Bay event on the Tokai segment. We show by strike–slip, whereas Fault II underwent predominantly reverse
that the occurrence rate of plate-boundary seismicity increased faulting motion. So far there have been only a few reports of
following the earthquake. Most of the presumed strongly earthquakes rupturing two conjugate fault planes simultaneously7 .
locked patches of the Tokai segment are located within areas In the case of the Suruga Bay event, the high stress concentration
of increased stress. Rupturing of a locked patch—following a generated by the large slip on Fault I that occurred nearby the
threshold level of seismic stress—could trigger the rupture of connecting point probably gave rise to the rupture of the conjugate
the entire Tokai segment, leading to a megathrust earthquake. Fault II. The total seismic moment is 4.35 × 1018 Nm (moment
Megathrust earthquakes in southwest Japan occur at intervals magnitude 6.4) and the seismic moments released from the two
of 100–200 years1 on the top surface of the Philippine Sea plate faults are comparable.
subducting from the Nankai–Suruga trough. The magnitude-, M-, The slip model previously described was used to calculate the co-
8-class earthquake seismogenic zone is divided into three main seismic Coulomb failure stress changes2 (hereafter 1CFS; see Meth-
segments (Fig. 1). Two megathrust earthquakes struck along the ods) caused by the Suruga Bay earthquake on the presumed Tokai
Nankai trough in 1944 and 1946; however, the Tokai segment fault area (Fig. 4a; Supplementary Movie S1). Red and blue denote
did not rupture at that time and it is still in a locked state. In zones of stress increase and decrease, respectively. The ‘eggplant’-
contrast, all three zones ruptured in 1854 (when two megathrust shaped source region of the anticipated Tokai earthquake and the
events occurred within a time period of ∼30 h) and 1707. Therefore, depth contours of the upper boundary of the subducting Philippine
the occurrence of the anticipated Tokai earthquake is of great Sea plate (Fig. 4a) are based on the study of plate configuration
concern for earthquake hazard assessment in the region. The Large and historical earthquake records, and used in official seismic-risk
Scale Earthquake Countermeasures Act established in Japan in 1978 evaluation by the Japanese government8 (Supplementary Note S1).
recognized officially the importance of mitigating the effects of this The rake at any given location in the source area was determined
anticipated megathrust event. by consideration of the plate motion direction. The average rake
In the decades before the occurrence of a megathrust earthquake, direction is indicated by a thick arrow in Fig. 4a. Considering
precursory reverse fault (compression-type) events often take place the uncertainty of the slip model, we did not use in the 1CFS
within the oceanic plate, oceanward of the source region of the calculations the relatively small slip on the northern fault plane that
major event3 . The 2009 Suruga Bay earthquake (M6.4), which is shallower than the Tokai fault plane. Stress change estimates are
occurred within the Philippine Sea plate and had a reverse fault predominantly positive in inland regions northwest of the Suruga
component, could be the result of tectonic-stress accumulation Bay hypocentre, exceeding 0.1 MPa up to ∼8 km from the tip of the
that has reached some critical level in the neighbouring, locked faults and gradually tapering off to around 0.03 MPa at ∼15 km.
Tokai source region. The analysis9 of background seismic activities and focal
The earthquake hypocentres of the Suruga Bay sequence mechanisms within the subducting Philippine Sea slab revealed
(mainshock and aftershock activity) were precisely relocated by the the presence of a relatively large locked zone along the plate
double-difference method4 using NIED’s Hi-net5 data (Fig. 2a–c; boundary. In ref. 10 it was found that the slow-slip event that
Methods and Supplementary Note S1). We can see a two-plane- took place in the Tokai region11 from 2000 to 2005 produced
shaped hypocentral distribution of aftershocks, indicating that the distinct activation and quiescent seismicity patterns within the
sequence occurred along two faults, located in the southern (Fault I) Philippine Sea plate. The authors assumed that the seismicity
and northern (Fault II) parts of the aftershock area and dipping to increase was the result of the slow-slip event quasistatic slip on
southeast and northeast, respectively. The mainshock hypocentre the plate interface and reflected the stress concentrations in some
is located on the southern fault. The space-time evolution of the strongly locked patches at the plate boundary, which they defined as
Suruga Bay sequence (Fig. 2d) shows that the aftershocks that asperities (Supplementary Note S1). They used GPS measurements
National Research Institute for Earth Science and Disaster Prevention, 3-1 Tennodai, Tsukuba, Ibaraki 305-0006, Japan. † Present address: Earthquake
Research Institute, The University of Tokyo, 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-0032, Japan. *e-mail: aoi@bosai.go.jp.
Fig. 4
34
Tonankai
??
4
4 185
33
Nankai 194
h
ug
a tro 7
32 rug 170
i–Su 6
ka 194 Eurasian plate
Nan 4
185
Pacific
0 100 Philippine plate
(km) Sea plate
31
132 133 134 135 136 137 138 139 140 141
Longitude (° E)
Figure 1 | Map showing the seismogenic zone of interplate megathrust earthquakes along the Nankai–Suruga trough, Japan. This large region can be
divided into three segments, Tokai, Tonankai and Nankai (areas outlined by thick, grey lines); M-8-class megathrust earthquakes have repeatedly ruptured
these segments, with inter-event times of 100–200 years, for example ref. 1. The 2009 Suruga Bay earthquake, whose faults are shown by two red
rectangles, occurred at the eastern edge of the seismogenic zone. The small and large dark green rectangles indicate the regions shown in Figs 2 and 3, and
in Fig. 4, respectively.
to confirm the proposed asperity model (grey-shadowed rectangles to failure by an amount equivalent to the length of time the tectonic
in Fig. 4a). The asperities can be understood as strong, highly loading would require to bring about a similar stress change. Quan-
stressed patches of material along the plate interface that will titative evaluation of the time advance would require knowledge
eventually rupture when their stressing state reaches a critical of the tectonic stressing rates in the asperities, which are difficult
threshold value. The increased 1CFS in the inland regions discussed to obtain using current data. The backslip rate distribution at the
above takes place within this presumably locked zone. Five (of a plate boundary, inverted from geodetic measurements14 , is too
total of six) asperities are located in areas of increased stress. As smooth for estimating the stressing rates on the seismicity-derived
discussed in previous studies, for example ref. 12, stress increases asperities. The rate- and state-dependent friction law15 predicts
(>0.01 MPa) have been shown to bring faults to failure, with the presence of a transient effect (Methods), which is the largest
delays ranging from seconds to decades. The 1CFS on the asperity immediately after the stress change, and is considerably greater in
that is closest to the Suruga Bay fault region has values that are magnitude than the permanent effect, but diminishes exponentially
significantly larger than 0.01 MPa (Fig. 4a). The other asperities with time16 . This effect can have important consequences not only
experience smaller stress changes of up to 0.01 MPa. Note that only on the occurrence of smaller-size aftershocks, but also on the
a relative stress increase is not enough to trigger earthquakes; this probability of relatively large future earthquakes13,17,18 .
depends on how close faults are to failure at the moment of the stress Simple, but realistic, calculations (Methods) show that in areas
change, which is generally unknown. that underwent stress changes of 0.1 MPa the seismicity rate can
Our 1CFS results are not sensitive to small changes of slip on the be ∼9.4 times higher, three months after the mainshock, relative
Suruga Bay fault, but rather depend on the amplitude and position to what it would be without the stress change. Even for stress
of the large slip concentrations, which were found to be stable in the changes of 0.03 MPa, the forecasted seismicity rate, three months
slip inversion analysis. We carried out more analysis to check the after, is ∼2.0 times higher. The rate estimations should, however be
stability of the stress changes. The slip was inverted by considering interpreted with care. As documented in previous studies, the rate
only one rupture plane, with the same strike angle as Fault II, changes are sensitive to the choice of Aσ (where A is a rate-and-state
but a larger area that accommodates the extent of the aftershock parameter and σ is the normal stress), as well as the aftershock
distribution; the same overall 1CFS pattern was found as for our duration ta . As we show in Supplementary Note S4, for a stress
preferred two-fault plane model. In addition, the 1CFS amplitudes change of 0.1 MPa, and taking uncertainties of the two parameters,
are similar, except for the immediate vicinity of the Suruga Bay fault Aσ and ta , into account, the rate after three months from the
planes, where the fault plane geometry has an important influence mainshock can be between 2.4 and 10.2 times higher than the rate
(Supplementary Note S3). before the mainshock.
The earthquake-induced 1CFS can have both permanent In the days immediately following the Suruga Bay earthquake, a
and transient effects on future seismicity rates and earthquake cluster of thrust earthquakes, M ∼ 2.0 (orange circles in Fig. 4a),
occurrence probabilities13 . A permanent increase (or decrease) in occurred within a highly concentrated zone, which was ∼15 km
the Coulomb stress on a fault plane shortens (or lengthens) the time away from the Suruga Bay faults and close to the plate boundary. For
a
Moment tensor P-wave polarity
00
B’
2 4 6
34.9 Magnitude
34.9
A
Latitude (° N)
tI
Faul
B A
Latitude (° N)
34.8 Fa lt I
ult A’ Fau
II
34.8 Fa
ult
10 km II
34.7
138.3 138.4 138.5 138.6
Longitude (° E)
5 5
0 0.25 0.50 0.75 1.00
10 10 Slip (m)
Depth (km)
15 15
with a 0.25 m interval (see also Supplementary Note S2). The star and
20 20 circle indicate the rupture initiation point (hypocentre) and the connecting
point of Faults I and II, respectively. Blue and red beach-balls show the focal
25 25 mechanisms estimated from the moment tensor inversion and P-wave
polarity analysis, respectively.
30 30
triggered on the plate boundary by the Suruga Bay earthquake.
35 35 We have investigated the possibility of interplate triggering in
0 5 0 5 10 more detail. We searched for events that occurred within location
Distance (km) Distance (km)
uncertainty bounds from the plate interface, one month before and
after the Suruga Bay earthquake, using the Hi-net catalogue data
d
(Methods). We eliminated the earthquakes within 2 km from the
00
lt I Bay event. Comparing the interplate seismicity (M > 0.0) rate before
Fau
and after the Suruga Bay earthquake we found a sudden increase
34.8 Fa immediately after the mainshock; the relative increase is most
ult
II pronounced spatially close to the Suruga Bay fault region, in an
area indicated by an ellipse in Fig. 4a and referred to as ‘Cluster 1’
(see also Supplementary Note S5). Figure 4b shows the cumulative
10 km number of earthquakes that occurred within the Cluster 1 area.
34.7
138.3 138.4 138.5 138.6 The seismicity has a clear relative increase after the Suruga Bay
Longitude (° E) earthquake, followed by a gradual decay with time (Methods). Most
of the interplate earthquakes that occurred after the Suruga Bay
0 0.5 1 2 24 earthquake within Cluster 1 are located in areas of positive 1CFS
Elapsed time (h) (Fig. 4a). These findings strongly suggest the triggering of interplate
events by the Suruga Bay earthquake. Estimations of seismicity rate
Figure 2 | Suruga Bay seismic sequence. a, Hypocentres of the mainshock changes at the plate interface validate the range of the rate-and-state
and aftershocks are shown by bold and thin circles, respectively. The forecasts previously discussed (Supplementary Note S5). The same
symbol size and colour scale indicate the magnitude and depth of relationship between a stress increase and a transient rise in seismic
earthquakes, respectively. The two rectangles are the faults coloured activities should apply to earthquakes of larger magnitudes2 , not
according to the depth variation along their dip. b,c, Vertical cross-sections necessarily the smaller shocks, which are more frequent.
show the hypocentre distribution across the strike directions of Fault I (b) The increase of stress in some of the locked portions on the
and Fault II (c). Dashed lines indicate the two faults. d, The green contour plate interface implies, in the short term, a higher seismicity rate
circle shows the mainshock. The other circles denote the aftershocks. The and an increased chance of larger interplate earthquakes. Most
symbol size and colour scale indicate the magnitude and occurrence time of the presumed asperities on the Tokai fault plane are located
relative to the mainshock, respectively. in areas of stress increase and one of them, situated close to the
fault region of the Suruga Bay earthquake, experienced significantly
all events with well determined focal mechanism, the 1CFS resolved large 1CFS values. The breakage of an asperity could eventually
on the nodal plane consistent with the dip of the subducting trigger the rupture over the entire fault plane, leading to a major
slab was found to be positive. This suggests that the cluster was interplate earthquake.
Occurrence of the
Mt. Fuji 50
Cumulative number
15
40
20
25
Suruga 30
35.0 Bay
30 Izu 20
peninsula
Latitude (° N)
10
25
0
12/7 19/7 26/7 02/8 09/8 16/8 23/8 30/8 06/9 13/9
Time (day/month)
34.5 20
trough
0.1
15
10
0.01
Suruga
ΔCFS (MPa)
Average
rake angle 0.001
Pacific 0
plate ¬0.001
0 10 20 Philippine
(km) Sea plate ¬0.01
34.0
137.5 138.0 138.5 139.0 ¬0.1
Longitude (° E)
Figure 4 | Stress changes in the Tokai source area and triggered seismicity. a, The ‘eggplant’-shaped source region is delimited by a bold solid line.
Positive (red) and negative (blue) stress changes are shown in a logarithmic scale. Contours indicate the depths of the plate boundary. The thick arrow
indicates the average slip direction on the presumed Tokai fault plane and grey shaded rectangles show presumed asperities9,10 . Several overlapping
orange circles indicate the epicentres of plate-boundary earthquakes with a thrust-type focal mechanism. Green circles indicate interplate earthquakes
occurring in the Tokai source area, in the month after the Suruga Bay earthquake. The area of significant seismicity increase after the mainshock is shown
as a dotted ellipse (Cluster 1). Dotted rectangles show the conjugate faults of the Suruga Bay earthquake. b, Cumulative number of earthquakes (Cluster 1
area) occurring one month before to one month after the Suruga Bay earthquake. The red line shows a modified Omori-law fit to the data after the Suruga
Bay earthquake (Methods).
4. Waldhauser, F. & Ellsworth, W. L. A double-difference earthquake location 18. Toda, S. & Stein, R. S. Toggling of seismicity by the 1997 Kagoshima earthquake
algorithm: Method and application to the Northern Hayward fault, California. couplet: A demonstration of time-dependent stress transfer. J. Geophys. Res.
Bull. Seismol. Soc. Am. 90, 1353–1368 (2000). 108, 2567 (2003).
5. Okada, Y. et al. Recent progress of seismic observation networks in 19. Ukawa, M., Ishida, M., Matsumura, S. & Kasahara, K. Hypocenter
Japan—Hi-net, F-net, K-NET and KiK-net. Earth Planet. Space 56, determination method of the Kanto–Tokai observational network for
xv–xxviii (2004). microearthquakes [in Japanese with English abstract]. Rep. Natl. Res. Cent.
6. Hartzell, S. H. & Heaton, T. H. Inversion of strong ground motion and Disaster Prev. 53, 1–88 (1984).
teleseismic waveform data for the fault rupture history of the 1979 Imperial 20. Bouchon, M. A simple method to calculate Green’s functions for elastic layered
Valley, California, earthquake. Bull. Seismol. Soc. Am. 73, 1553–1583 (1983). media. Bull. Seismol. Soc. Am. 71, 959–971 (1981).
7. Robinson, D. P., Henry, C., Das, S. & Woodhouse, J. H. Simultaneous rupture 21. Kennett, B. L. N. & Kerry, N. J. Seismic waves in a stratified half space.
along two conjugate planes of the Wharton Basin earthquake. Science 292, Geophys. J. R. Astron. Soc. 57, 557–583 (1979).
1145–1148 (2001). 22. Toda, S., Stein, R. S., Richards-Dinger, K. & Bozkurt, S. B. Forecasting the
8. Central Disaster Prevention Council of Japan. Special committee evolution of seismicity in southern California: Animations built on earthquake
report of the Tokai earthquake (in Japanese). 17pp.; available at stress transfer. J. Geophys. Res. 110, B05S16 (2005).
http://www.bousai.go.jp/jishin/chubou/20011218/siryou2-2.pdf. 23. Hainzl, S. et al. Aftershock modeling based on uncertain stress calculations.
9. Matsumura, S. Focal zone of a future Tokai earthquake inferred from J. Geophys. Res. 114, B05309 (2009).
the seismicity pattern around the plate interface. Tectonophysics 273, 24. Utsu, T. A statistical study on the occurrence of aftershocks. Geophysics 30,
271–291 (1997). 521–605 (1961).
10. Matsumura, S., Satomura, M. & Uchiumi, S. Presumption of asperities for the 25. Ogata, Y. Estimation of the parameters in the modified Omori formula for
anticipated Tokai earthquake (seismic activity change and crustal deformation aftershock frequencies by the maximum likelihood procedure. J. Phys. Earth
in the Tokai region: Part 5) [in Japanese with English abstract]. Zisin 60, 31, 115–124 (1983).
267–277 (2008).
11. Ozawa, S. et al. Detection and monitoring of ongoing aseismic slip in the Tokai
region, central Japan. Science 298, 1009–1012 (2002). Acknowledgements
12. Parsons, T., Ji, C. & Kirby, E. Stress changes from the 2008 Wenchuan We thank Y. Okada, S. Hainzl, S. Matsumura, H. Hirose, Y. Ben-Zion, M. Imoto,
earthquake and increased hazard in the Sichuan basin. Nature 454, S. Noguchi, Z. Peng and T. Miyoshi for discussions and reviews of the manuscript.
509–510 (2008). M. Matsubara provided information on the focal mechanism of the triggered
13. Toda, S., Stein, R. S., Reasenberg, P. A., Dieterich, J. H. & Yoshida, A. Stress seismicity. The suggestions of T. Tada were helpful to improve the clarity and
transferred by the 1995 Mw = 6.9 Kobe, Japan, shock: Effect on aftershocks conciseness of the text.
and future earthquake probabilities. J. Geophys. Res. 103, 24543–24565 (1998).
14. Ohta, Y., Kimata, F. & Sagiya, T. Reexamination of the interplate coupling Author contributions
in the Tokai region, central Japan, based on the GPS data in 1997–2002. S.A. and B.E. conceived the analysis and wrote the paper. S.A., B.E., W.S. and Y.A. carried
Geophys. Res. Lett. 31, L24604 (2004). out the analysis. All the authors contributed to the observations.
15. Dieterich, J. A constitutive law for rate of earthquake production and its
application to earthquake clustering. J. Geophys. Res. 99, 2601–2618 (1994).
16. King, G. C. P., Stein, R. S. & Lin, J. Static stress changes and the triggering of Additional information
earthquakes. Bull. Seismol. Soc. Am. 84, 935–953 (1994). The authors declare no competing financial interests. Supplementary information
17. Stein, R. S., Barka, A. A. & Dieterich, J. H. Progressive failure on the North accompanies this paper on www.nature.com/naturegeoscience. Reprints and permissions
Anatolian fault since 1939 by earthquake stress triggering. Geophys. J. Int. 128, information is available online at http://npg.nature.com/reprintsandpermissions.
594–604 (1997). Correspondence and requests for materials should be addressed to S.A.