Earth and Planetary Science Letters 310 (2011) 462–467
Contents lists available at SciVerse ScienceDirect
Earth and Planetary Science Letters
journal homepage: www.elsevier.com/locate/epsl
Structural controls on the Mw 9.0 2011 Offshore-Tohoku earthquake
B.L.N. Kennett a,⁎, A. Gorbatov b, E. Kiser c
a
b
c
Research School of Earth Sciences, The Australian National University, Canberra ACT 0200, Australia
Geoscience Australia, GPO Box 378, Canberra ACT 2600, Australia
Department of Earth and Planetary Sciences, Harvard University, 20 Oxford Street, Cambridge MA 02138, USA
a r t i c l e
i n f o
Article history:
Received 12 May 2011
Received in revised form 25 August 2011
Accepted 26 August 2011
Available online 29 September 2011
Editor: T.M. Harrison
Keywords:
earthquakes
Japan
joint seismic tomography
earthquake slip
distribution
a b s t r a c t
Joint seismic tomography exploiting P and S wave arrivals conducted before the 2011 Offshore Tohoku earthquake reveals an area comparable to the faulting surface for the 2011 March 11 event with different properties from other areas along the shallow part of the subduction zone. The differences are revealed by using a
measure R of the relative variations in shear wavespeed and bulk-sound speed. Within the faulting area there
are patches on the subduction zone with slightly reduced S wavespeed, and thus negative R, that appear to
separate portions of the rupture with very different character. On the down-dip side there is strong shortperiod radiation, whilst the largest slip occurs up-dip with most energy release at longer periods. Segmentation of the slip process can be imaged by back projection of seismograms from the US Array; the areas of
greatest energy release at short periods lie down-dip from the negative R anomalies. The main seismic moment release from broad-band seismograms lies on the updip side of the same anomalies. The structural variations on the subduction zone thus separate two regions with fundamental differences in the rupture process,
stronger long-period radiation up-dip and stronger short-period radiation down-dip. These variations are likely
to reflect features brought into the subduction zone, which may have acted as asperities that allowed this event
to build up 30–40 m of strain in the near trench zone, making it much bigger than expected. Thus minor changes
in the character of the subducted plate can have a significant influence on the behaviour of a great earthquake.
© 2011 Elsevier B.V. All rights reserved.
1. Introduction
The offshore zone from north-eastern Japan (Tohoku) has long been
recognised for its earthquake hazard (see, e.g., Richter 1956) particularly through tsunami, as in the destructive 1896 Sanriku tsunami event
and the 1933 earthquake at the trench line. Although most events
over the last century in the hypocentral zone of the 2011 Tohoku
event have been of magnitude around 7.2–7.5 (Miyagi-oki events at
about 30 year intervals), Kanamori et al. (2006) recognised the potential for a combination of a megathrust and a tsunamigenic event, since
only about a quarter of the plate motion had been taken up by the slip
in this regular earthquake sequence.
There were some differences in the estimates of hypocentres for
the 2011 Tohoku sequence provided by different agencies, depending
on the data and velocities models employed. Zhao et al. (in press)
have used the tomographic model of Huang et al. (2011) derived
from observations of P and S phases in northern Japan to locate the
2011 main event and its aftershocks using a full allowance for 3-D
structure. Their results show that the events occur at the interface between the subducting and overriding plates.
Loveless and Meade (2010) have made a very detailed analysis of
the observations from the dense networks of Global Positioning
⁎ Corresponding author.
0012-821X/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.epsl.2011.08.039
System (GPS) stations in Japan to estimate plate motion and interplate coupling. Their results indicate strong seismic coupling for the
Pacific margin, with coupling concentrated in the depth range 20–
60 km. This is the depth range in which there was major generation
of high frequency energy in the 2011 March 11 event, but not
where the maximum slip occurred. The coupling results for northeastern Japan are in good agreement with earlier studies, such as
Nishimura et al. (2000, 2004).
An alternative view of the subduction field is provided by seismic
tomography, which has proved to be very successful in imaging the
high wavespeeds of the subduction zone (e.g. van der Hilst et al.,
1991; Widiyantoro et al., 1999). The density of events and the quality
of observations in the Japan region are such that both P and S wave
arrivals can be exploited to obtain complementary images of subduction behaviour (Gorbatov and Kennett, 2003; Huang et al., 2011;
Widiyantoro et al., 1999). Joint inversion of P and S arrival times to
extract images of shear wavespeed and bulk-sound speed provides
insight into the nature of the variations in physical properties by isolating the influences of the shear and bulk modulus. Such variations
can be used to map variability within the subducting plate (Gorbatov
and Kennett, 2003), notably features associated with the arrival of
lithosphere of different ages at the trench.
We demonstrate here, that the variations in structure of the subducting plate can be used to map out segments of the plate with a difference in character, and hence a different state in the earthquake
B.L.N. Kennett et al. / Earth and Planetary Science Letters 310 (2011) 462–467
cycle. Further, as in the 2004 Sumatra-Andaman event (Kennett and
Cummins, 2005), variations in the details of structure in the subduction zone play an important role in determining the way in which energy release occurs during the great 2011 Offshore-Tohoku earthquake,
which has a number of major sub-events.
463
Shear/Bulk-sound:
Depth slice at 25 km
N
45
2. Structure and moment release
The ratio of the variations in shear wavespeed and bulk-sound
speed provides a convenient way of probing subtleties of structural
variation that are hardly evident in individual wavespeed results.
Thus, Kennett and Cummins (2005) were able to map out variations
along the subduction zone from Sumatra to Andaman that marked
segmentation of the energy release in the great 2004 SumatraAndaman earthquake. For this event, zones where the variability of
bulk-sound speed was stronger than for shear wavespeed acted as
temporary barriers to slip and were marked by dips in the radiation
of high frequency energy (cf. Ishii et al., 2005).
We exploit the same joint-tomography results of Gorbatov and
Kennett (2003) as were used by Kennett and Cummins (2005) to
examine the configuration of the subduction zones in the Japanese region before the Mw 9.0 Offshore-Tohoku event. The joint tomographic
inversion makes use of both regional and teleseismic information,
which allows good resolution out to the trench line. The data are selected
from just source-receiver pairs that record both P and S phases; this
means that the propagation paths for the two wavetypes are very
close. The seismic images were created by initial separate inversions
of P and S wave travel times using non-linear tomography with a regional model embedded in a coarser global model. 3-D ray tracing
was used at each step in the iterative process for the regional models.
The P and S ray paths were then used in a final linearised inversion for
bulk-sound speed, the wavespeed derived from just the bulk modulus,
and shear wavespeed. Because the sampling of P and S waves are
equivalent, the resolution of the two images will be virtually identical
and we can take ratios between wavetypes with confidence. For northeastern Japan our available horizontal resolution is 0.5 × 0.5° out to the
trench line, i.e., approximately 55 km in the north–south direction and
42 km in the east–west direction. We are unable to resolve the shallow
details near the trench which are likely to be of importance in controlling the detail of slip.
From the joint tomographic inversions we have constructed a
measure of the relative variations in bulk-sound speed (δϕ/ϕ) and
shear wavespeed (δβ/β) with respect to the ak135 reference model
(Kennett et al., 1995) as:
R¼
1 þ δβ=β
−1;
1 þ δϕ=ϕ
ð1Þ
which to first order reduces to the differences in the variations between the wavespeeds δβ/β − δϕ/ϕ. For subducted slabs where the
wavespeeds are elevated compared to the reference, R is positive
when the variation in shear wavespeed dominates that in bulksound speed, and negative when the variation in bulk-sound speed
is more prominent. The quantity R is a sensitive indicator of the wavespeed properties and readily reveals variations that are difficult to
discern in comparison of separate tomographic images.
Fig. 1 displays a depth slice at 25 km depth across Japan and the
southern Kuriles for the measure R defined in Eq. (1). This depth
slice is chosen to pass through the hypocentre for the 2011 event. Towards the trench the slice cuts into the subducting oceanic lithosphere and shows strong positive R corresponding to presence of
high shear wavespeeds. Away from the subduction zone, we see concentrations of negative R and thus stronger bulk-sound speed variation that correspond directly to the main volcanic arc in Honshu,
where partial melt may occur. This suggests that fluids contribute significantly to the mapped variations in R. The effect of fluid content is
40
1915
H
S1
S2
S3
35
Deviation [%]
-2.25
135
140
145
0.
2.25
150
E
Fig. 1. Depth slice across Japan and the southern Kuriles at 25 km depth, showing the
variations in the relative variations of the shear wavespeed and bulk-sound speed from
the reference model ak135. Negative values of the ratio R (Eq. (1)) imply dominance of
bulk modulus variations, and positive values dominance of shear modulus variations.
The area of slip for the 2011 Offshore Tohoku event is outlined by a grey rectangle. The
hypocentre is marked by a red pentagon. The locations of the three points of initiation
of high frequency energy from Furumura et al. (2011) are marked by diamonds: S1 —
33 km, S2 — 19 km and S3 — 50 km. Seismic events from 2004 to 2010 are indicated by
white diamonds, scaled by magnitude. The outlines of the rupture areas of historical earthquakes are shown by dashed lines.
much larger on the shear modulus than bulk modulus, the consequent reduction in the shear modulus can produce a negative value
for R even when the bulk-sound speed is almost unchanged.
The area of slip associated with the 2011 event is indicated by a
grey rectangle in Fig. 1, and is confined to a single segment of the subduction zone delineated by the impact of a sea-mount chain on the
trench in the south and a fossil transform fault in the north. It is readily
seen that this rectangle outlines a region with a distinctive character in
the tomographic image compared with other segments close to the
trench lines. Patches with negative R (light orange) are enclosed in a
near neutral zone whose extent corresponds closely with the main
slip zone for the 2011 March 11 event. The light orange zone of negative
R arises from a slight reduction in the shear wavespeed. The broad patterns of variation in R in offshore-Tohoku are very different from the
isolated patches of negative R found for the 2004 Sumatra-Andaman
event.
Whilst the offshore-Tohoku area is the most consistently anomalous patch in Fig. 1, we can note a number of other features. The
very prominent patch of negative R near 135° behind the Nankai
trough lies close to the subducted seamount recognised by Kodaira
et al. (2000). This seamount appears to have had a significant effect
on the slip patterns in the great 1946 Nankai earthquake (Cummins
et al., 2002). There is also a significant zone of negative R in the link
between the Boso peninsula and the Pacific trench that has been
identified as a zone of strong coupling between the subduction zone
and the overlying plate by Loveless and Meade (2010). There is also
a modest spot of negative R south of Hokkaido, close to the location
of large recent events.
In Fig. 1 we also show the approximate slip areas for the major historical events offshore from Tohoku. As noted by Kanamori et al.
(2006) these events generally divide into two classes. Firstly large
events along the trench line north of 37.5°, and secondly slightly
smaller events closer to the coast. The region of negative R largely
separates these two zones. The inferred area of slip for the 869 CE
464
B.L.N. Kennett et al. / Earth and Planetary Science Letters 310 (2011) 462–467
Jogan earthquake that produced a tsunami similar to that in 2011 also
lies to the east of the orange, negative, zone. The only event that lies
in the light orange zone with negative R is the 1915 Miyagi-Oki M 7.5
event.
Nishimura et al. (2000) have identified a zone close to the Japanese
coast at 39.5°N that has not had large earthquakes yet is able to absorb
postseismic sliding. This patch corresponds directly to the zone of
slightly negative R (light orange) in Fig. 1.
By themselves, the variations in R along the subduction zone may
seem a curiosity, but as in the case of the 2004 Sumatra-Andaman
event we will see that they strongly influence the pattern of energy
release in the 2011 Offshore-Tohoku event. There is considerable evidence to suggest that this great earthquake delivered its impact through
a sequence of sub-events, which themselves will have been of considerable magnitude. The most direct information comes from the seismograms recorded across Tohoku (Furumura et al., 2011) from the
2011 Offshore-Tohoku event. Three distinct sets of high-frequency arrivals can be followed across the whole of northern Honshu. The timing
patterns of the P wave arrivals require the first two sub-events (S1, S2)
to lie at around 38.5°N with the second 45 s later and further away from
the coast. The third group of energy arrives a further 40 s later from a
more southerly position (S3 near 37°N) close to the coast. These three
points of initiation of high frequency energy are shown in Fig. 1 along
with the hypocentre for the entire event sequence H.
The map view in Fig. 1 demonstrates the presence of a region with
a very different ratio R in the slip zone for the 2011 event, but does
not represent the structure along the subduction zone interface
where the slip occurred. In Fig. 2 we display the ratio R once again,
but now with a section taken along an inclined plane approximately
coincident with the top of the subduction zone. We have used the
same colour scheme for the deviations of R from zero as in Fig. 1,
and once again included the approximate slip areas for the major historical events.
E
140
142
144
40
50 km
The reference point for the inclined plane is taken at 38°N, 142°E
and the plane has azimuth 195° and dip of 10°. Although there may
be some increase in dip towards the west, the major effects of the
geometry of subduction are accounted for by the inclined plane, particularly given the available resolution of our tomographic results. We use a
larger magnification in Fig. 2, and concentrate on the region of major
slip in the 2011 Offshore-Tohoku event. Once again we show the
three initiation points for high-frequency energy recorded in Japan
identified by Furumura et al. (2011). We note the clear separation
of the initiation points S1 (down-dip from hypocentre) and S2 (updip from hypocentre) by the zone of enhanced ratio R. The third source
of high-frequency energy S3 could be shallow, but more likely lies on
the subduction zone interface. As we see from Fig. 2, the deeper location
for S3 lies just at the edge of the zone where the contrasts in the ratio R
are most evident.
In Fig. 3 we show a vertical cross-section through the subducting
Pacific plate at 38°N for bulk-sound speed, shear wavespeed and the
ratio R introduced in Eq. (1). The zone associated with the 2011
Offshore-Tohoku event is marked by triangles. At greater depths the
dominance of shear wavespeed in older subducting lithosphere,
noted by Gorbatov and Kennett (2003), is very clear. However, structures are rather more complex as the trench is approached. Near the
trench the bulk sound-speed shows stronger deviations from the
reference model approaching or exceeding those for shear wavespeed,
which remain strong. The display for the ratio R clarifies these effects.
There is a clear zone of difference with negative R along the interface between the overlying plate and the subducting material.
Many different source inversions have already been undertaken,
and there are significant differences between models depending on
the sources of information employed, and the assumptions made
about rupture properties. Geodetic observations from land-based stations and the nature of the tsunami require a very substantial slip
component up-dip from the hypocentre near the trench of 40 m or
more (e.g. Ammon et al., in press; Fujii et al., in press, Lay et al., in
press, Simons et al., 2011; Yoshida et al., in press). These very large
slips are supported by direct observations of 50 m slip at the toe of
the fault by JAMSTEC reflection work.
The variations in different source models are most marked in the
vicinity of the trench where, as pointed out by Fujii et al. (in press),
all the available methods have difficulty resolving the up-dip extent
25 km
0
1915
H
S1
S2
0 km
38 N
Bulk-sound
200
38
Depth [km]
400
S3
140
145
Shear
200
400
140
36
145
Ratio
200
400
140
N
Fig. 2. Detail of the joint tomographic image for the ratio R on an inclined plane
through the upper part of the subduction zone; reference point: 142°E, 38°N, 25 km
depth, plane with dip azimuth 195°, dip angle 10°. The locations of the hypocentre
and the points of initiation of high frequency energy from Furumura et al. (2011)
are marked as in Fig. 1, and the outlines of the rupture areas of historical earthquakes
are shown by dashed lines. The depths on the inclined plane are indicated by dashed
lines at 0, 25 and 50 km depth.
145 E
-2.25
0.
2.25
-2.25
0.
2.25
Perturbation [%]
Fig. 3. Vertical cross sections at 38°N for the bulk-sound speed, shear wavespeed and
the ratio of relative variations in shear and bulk-sound speed R. The region of slip in
the 2011 Tohoku-oki event is marked with triangles. The use of the ratio measure R enhances features in the joint tomography that are not readily seen by direct comparison
of the images for the two wavespeeds.
B.L.N. Kennett et al. / Earth and Planetary Science Letters 310 (2011) 462–467
of rupture. Geodetic inversions based on onshore information have
both a limited azimuthal coverage, and limited sensitivity to slip far
offshore, so that there is strong dependence on a priori assumptions.
Seismic inversions involve dependence on assumptions made about
the nature of rupture expansion and the seismic velocity structure
in the source region. Inversions of seismic waveforms face increasing
difficulties as the segments of a fault approach the surface, and seismic moment estimation for the shallowest zones can become unstable. The assumed velocity structure has a strong influence on the
mapping of seismic moment to slip. The same amount of seismic moment release produces much larger slip in lower wavespeed materials. Tsunami inversions average over the total region of ocean
bottom uplift, but can provide good timing control on the pattern of seafloor uplift with the strongest control coming on dip at shallow depths.
The major slip in the 2011 Offshore-Tohoku event occurs up-dip from
the hypocentre, but seismic moment release occurs down-dip as well.
For most recent megathrust events there has been a strong link
between overall source properties and the seismic radiation at short
periods. The correlation is very good for the 2004 SumatraAndaman event. Though, there are some hints that the 2010 Maule,
Chile (Mw 8.8) event had stronger short-period radiation down-dip
with more longer-period energy radiated from shallower segments
(Kiser and Ishii, 2011; Lay et al., 2010).
This separation of frequency components is even more pronounced for the 2011 Offshore-Tohoku event. A number of investigations point to a concentration of short-period radiation down-dip
from the hypocentre (e.g. Kiser, 2011; Koper et al., in press; Simons
et al., 2011). Whereas, broadband seismic wave inversions indicate
the major seismic moment release occurs up-dip from the hypocentre
and may well extend to the trench (Ammon et al., in press; Lay et al.,
in press). In contrast the geodetic inversions require strong slip on the
465
up-dip side with a reduction of slip towards the trench (e.g. Simons et
al., 2011).
Where dense networks of seismograms are available it is possible
to combine the waveforms from the individual stations and back project to the source region to examine the patterns of coherent energy
release. This approach was pioneered by Ishii et al. (2005) using the
Hi-Net array in Japan to image the faulting pattern in the 2004
Sumatra-Andaman event. For the 2011 Offshore Tohoku event such
back projection has been carried out using both dense networks in
the Tokyo region (Honda et al., in press) and stations at teleseismic
distances from both Europe and North America (e.g., Koper et al., in
press, Simons et al., 2011). In particular the dense configurations of
stations deployed in the US Array allows detailed back projection to
be achieved for the Offshore-Tohoku event in which the time evolution of the radiation process can be tracked in space (Kiser, 2011;
Wang and Mori, in press; Yao et al., 2011; Zhang et al., in press) The
details of the inferred radiation pattern differ slightly as a consequence
of differences in the stacking procedures employed. All studies require
segmentation of the energy release with a shift to the south later in
the main sequence. Kiser (2011) has continued the process of backprojection for 25 min after the onset of the Tohoku event to examine
the pattern of energy release in both the main event and its immediate
aftershocks.
The back-projection process is likely to focus attention on energy
release in concentrated patches since this will lead to coherency in
the generated waveforms. Broadly distributed slip happening at the
same time will be less well recovered.
In Fig. 4, we show the energy release pattern inferred by Kiser
(2011) compared to the tomographic results introduced in Fig. 2,
with the same slice through the tomographic model on a plane approximately coincident with the top of the subduction zone. The
E
14 0
14 2
144
40
S1
38
S3
36
N
0
15
30
Moment
Fig. 4. Comparison of the seismic moment release model of Ammon et al. (in press), back-projection image of the 2011 Offshore-Tohoku event the using the US array (Kiser, 2011),
and the joint tomographic image on the same inclined plane through the subduction zone as used in Fig. 2. The patches of strongest energy release shown in red tones in the left
hand panel lie outside the regions with negative anomalies in the joint tomography. The two major sub-events in the main event are indicated as S1, S3 in order of occurrence to
correspond with the high-frequency initiation identified by Furumura et al. (2011). The open contour in the left hand panel corresponds to the M 7.1 foreshock on 2011 March 9.
The moment release in units of 1019 Nm is indicated by the colour code shown at the base of the right hand panel.
466
B.L.N. Kennett et al. / Earth and Planetary Science Letters 310 (2011) 462–467
contours of the inferred main radiation patches are superimposed on
the seismic moment release pattern inferred by Ammon et al. (in
press) and the image of the measure R of the relative variations in
shear and bulk-sound speed.
The Ammon et al. (in press) moment release model is plotted with
regular sized symbols on a grid across the fault plane colour coded to
the moment release, so that the underlying tomographic results can
still be seen. The fault plane chosen by Ammon et al. (in press) has
strike 202° and dip 12° and so is very close to our cut through the tomographic model. This inversion combines seismic and geodetic constraints and has a similar character to that produced by Simons et al.
(2011) using comparable data.
The radiation patches derived from our back-projection results are
shown in blue outlines, with lighter colours for later times. These
patches lie largely outside the orange zones with negative R. Indeed
the areas with largest energy release, shown by orange to red tones
in the left hand panel, have limited overlap with the negative R anomalies. Secondary areas of energy release also miss these zones. The negative R values in Fig. 4 come from a slight decrease in the shear
wavespeeds relative to the rest of the subduction zone.
There is a good correspondence with the back-projections undertaken by Koper et al. (in press) and Simons et al. (2011), using US and
European data, both in the locations of major short-period energy release and the time progression. Most short-period energy is released
down-dip from the hypocentre, except for the more northerly patch.
The sets of results we compare in Fig. 4 are based on very different
classes of information, direct use of seismograms for the backprojection, joint inversion of seismic and geodetic data for the moment release model and indirect inference from the patterns of passage times of seismic waves for the tomography. All have limitations
in spatial resolution, yet the correspondence in the behaviour is very
strong.
3. Discussion and conclusions
Our results indicate that the details of the slip distribution in the
2011 Offshore Tohoku event were strongly influenced by preexisting differences in physical properties associated with the subducting slab. The variations in physical properties are quite subtle,
but could be expressed in the results from the joint tomography constructed well before the event. The region with negative R largely separates the zone of strong short-period radiation from that with larger
long-period radiation (cf. Fig. 2 of Koper et al., in press).
Lay et al. (in press) have shown that the teleseismic P wave data is
compatible with a seismic moment distribution that has components
both up-dip and down-dip from the hypocentre, for which the largest
slip occurs close to the trench as a consequence of the lowered seismic wavespeeds near the surface. The short-period radiation may appear to be associated with a less important component of slip, but still
reflects the details of the rupture process.
Slip on the off-shore Tohoku area started with an M 7.1 event
2 days before the main event, and has continued through a sequence
of major aftershocks. Nearly all the inferred energy release patches
from back-projection avoid the negative R anomalies from the joint
tomographic results, though the hypocentre for the foreshock lies between two patches at 39°N (Fig. 2). The slip release continues for 2 h
after the main shock and except for some aftershocks near the trench
line, where the tomographic resolution diminishes, continue to avoid
the negative R tomographic anomalies. The patterns of slip inferred
from seismic inversions (Lay et al., in press) or combined seismic/
geodetic work (Ammon et al., 2011; Simons et al., 2011) put the
main slip up-dip of the anomalous zone. The structural variations
we have identified may have acted as asperities that allowed this
event to build up 30–40 m of strain in the near trench zone, making
it much bigger than expected.
The zone of negative R anomalies from the joint tomography lies
close to a zone of increased P wave velocity in the local tomography
undertaken by Matsubara and Obara (in press). Zhao et al. (2011)
have also identified significant wavespeed variations using the
detailed P velocity model of Huang et al. (2011); their analysis concentrates on the zone immediately above the subducting interface
where wavespeed variations are rather large. The variations are
suggested to be related to variations in frictional behaviour. The S
wavespeed model of Huang et al. (2011) is less detailed than that
for P, but also indicates significant variations within the subducted
plate. The positions of the structural anomalies are in good correspondence with our results shown in Figs. 1, 2.
The structural variations in the subducted plate are likely to be associated with structures ingested by the subduction zone, as in the
case of the seamount in the Philippine plate behind the Nankai
Trough (Cummins et al., 2002; Kodaira et al., 2000). Certainly seamounts are currently impinging on the Pacific plate margin at the
southern limit of the slip area in the 2011 Offshore-Tohoku event.
We note that north of 37.5°N topographic features on the seafloor
with a directional trend similar to the pattern of anomalies are in
the process of being subducted. Such variations can be expected to influence the way in which the slab responds as it bends into the trench,
opening lines of weakness for fluid infiltration and so alteration of the
physical properties, which can then be captured through the joint tomographic imaging.
The joint tomography (Gorbatov and Kennett, 2003) provides insight into the substructure of the subduction system that cannot be
seen directly in the results for a single wavetype. There is little visible
difference in the patterns of high wavespeed associated with the subducting slip in the separate images for P and S wavespeed constructed
from the same data set, yet the small variations can be captured
through the joint inversion. Most source inversions from seismic
and geodetic observations use rather simple models for the representation of the structure to calculate the necessary Green's functions.
Variations in local properties in the neighbourhood of the fault
plane would have the effect of distorting the inferred slip distribution
inferred from such simple models.
The negative anomalies in the ratio R displayed in Figs. 2 and 4 lie
at the up-dip edge of the zone of strong-coupling between the subduction zone and the overriding plate in the offshore Tohoku region
inferred by Loveless and Meade (2010) from geodetic observations.
Their zone of maximum coupling lies in the palest blue zone at 38–
39°N just down-dip of the orange anomalies in Fig. 2. The scale of
the subtle variations revealed in the joint tomography would be difficult to resolve by geodetic means.
The Jogan event in 869 CE appears to have had a similar distribution of tsunami deposits to that which occurred in March 2011, with a
tsunami magnitude of around 8.6 (Minoura et al., 2001). In the
1142 years since that event the slab appears to have been primed
for another great earthquake, but the available structural hints were
quite subtle and were missed.
The fine resolution in the joint tomographic results is only possible
because of the quantity and quality of seismic observations in the Japanese region. We note from Fig. 1 that weaker variations in the ratio of
variations of wavespeed can be seen near the trench in Hokkaido, and
to some extent to the south of Shikoku, both areas with currently strong
subduction zone coupling in the model of Loveless and Meade (2010),
and both historically the site of major earthquakes.
Another patch that shows a similar pattern to offshore Tohoku is
close to the island of Urup in the Kuriles (45.2°N, 150.4°E) where
again we see in Fig. 1 a zone of negative R behind the trench line.
The 1963 M8.3 event ruptured past this area, with a 4.5 m tsunami
run-up on Urup. The three main subevents associated with the breaking
of asperities (Ruff and Kanamori, 1983) lie on the down-dip edge of the
zone of negative R, in a similar configuration to the offshore-Tohoku
event. The M7.8 aftershock a week later produced a much larger tsunami
B.L.N. Kennett et al. / Earth and Planetary Science Letters 310 (2011) 462–467
on Urup with 15 m run-up 2011. In this case we have a temporal separation of days between down-dip and up-dip failure, whereas the
delay was no longer than 45 s for the Offshore-Tohoku event of 2011.
It is now worthwhile to look at the full suite of the subduction system across the globe to see whether other anomalous patches can be
recognised, which may be indicative that an earthquake cycle is
reaching maturity. We can use the Gorbatov and Kennett (2003) results for the subduction zones of the western Pacific and Indonesia,
but new detailed joint tomography will be required for the other
zones.
Acknowledgements
This paper was markedly improved by the critical comments of
two anonymous reviewers. We are grateful to C.J. Ammon for the
public provision of his model for the rupture process of the 2011 Offshore Tohoku earthquake, and to H. Kanamori for information on earlier events.
References
Ammon C.J., Lay T., Kanamori H., and Cleveland M., in press. A rupture model of the
great 2011 Tohoku earthquake, Earth Planets Space, 63.
Cummins, P.R., Baba, T., Kodaira, T., Kaneda, Y., 2002. The 1946 Nankai earthquake and
segmentation of the Nankai Trough. Phys. Earth Planet. Inter. 132, 75–87.
Fujii Y., Satake K., Sakai S., Shinohara M., and Kanazawa M., in press. Tsunami source of
the 2011 off the Pacific coast of Tohoku, Japan earthquake, Earth Planets Space, 63.
Furumura, T., Takemura, T., Noguchi, S., Takemoto, T., Maeda, T., Iawai, K., Padhyi, S., 2011.
Strong ground motions from the 2011 Off the Pacific Coast of Tohoku, Japan (Mw 9.0
earthquake obtained from a dense nation-wide seismic network. Landslides 8,
333–338.
Gorbatov, A., Kennett, B.L.N., 2003. Joint bulk-sound and shear tomography for Western
Pacific subduction zones. Earth Planet. Sci. Lett. 210, 527–543.
Honda R., Yukutake Y., Ito H., Harada M., Aketagawa T., Yoshida A., Sakai S., Nakagawa S.,
Hirata N., Obara K., and Kimura H., in press. A complex rupture image of the giant
earthquake which struck east Japan revealed by a dense seismograph network,
MeSO-net, Earth Planets Space, 63.
Huang, Z., Zhao, D., Wang, L., 2011. Seismic heterogeneity and anisotropy of the Honshu arc from the Japan Trench to the Japan Sea. Geophys. J. Int. 184, 1428–1444.
doi:10.1111/j.1365-246X.2011.04934.x.
Ishii, M., Shearer, P.M., Houston, H., Vidale, J.E., 2005. Extent, duration and speed of the
2004 Sumatra-Andaman earthquake imaged by the Hi-net array. Nature 435,
933–936. doi:10.1038/nature03675.
Kanamori, H., Miyazawa, M., Mori, J., 2006. Investigation of the earthquake sequence
off Miyagi prefecture with historical seismograms. Earth Planets Space 58,
1533–1541.
Kennett, B.L.N., Cummins, P.R., 2005. Relationship of the seismic source and subduction
zone structure for the 2004 Dec 26 Sumatra-Andaman Earthquake. Earth Planet
Sci. Lett. 239, 1–8.
Kennett, B.L.N., Engdahl, E.R., Buland, R., 1995. Constraints on seismic velocities in the
Earth from travel times. Geophys. J. Int. 122, 108–124.
Kiser, E., 2011. Preliminary Rupture Modelling of the March 11, 2011 Tohoku-Chiho
Taiheiyo-Oki Earthquake and Sequence of Events using the USArray Transportable
Array. http://www.seismology.harvard.edu/research_japan.html.
Kiser, E., Ishii, M., 2011. The 2010 Mw 8.8 Chile earthquake: triggering on multiple segments and frequency-dependent rupture behavior. Geophys. Res. Lett. 38, L07301.
doi:10.1029/2011GL047140.
467
Kodaira, S., Takahashi, N., Nakanishi, A., Miura, S., Kaneda, Y., 2000. Subducted seamount imaged in the rupture zone of the 1946 Nankaido earthquake. Science
289, 104–106.
Koper K.D., Hutko A.R., Lay T., Ammon C.J., and Kanamori H., in press. Frequencydependent rupture process of the 11 March 2011 MW 9.0 Tohoku earthquake:
Comparison of short-period P wave backprojection images and broadband seismic
rupture models, Earth Planets Space, 63.
Lay, T., Ammon, C.J., Kanamori, H., Koper, K.D., Sufri, O., Hutko, A.R., 2010. Teleseismic
inversion for rupture process of the 27 February 2010 Chile (Mw 8.8) earthquake.
Geophys. Res. Lett. 37, L13301. doi:10.1029/2010GL043379.
Lay T., Ammon C.J., Kanamori H., Xue L., and Kim M.J., in press. Possible large neartrench slip during the great 2011 Tohoku (MW 9.0) earthquake, Earth Planets
Space, 63.
Loveless, J.P., Meade, B.J., 2010. Geodetic imaging of plate motions, slip rates and partitioning of deformation in Japan. J. Geophys. Res. 115, B02410. doi:10.1029/
2008JB006248.
Matsubara M. and Obara K., in press. The 2011 off the Pacific coast of Tohoku Earthquake related to a strong velocity gradient with the Pacific plate, Earth Planets
Space, 63.
Minoura, K., Imamura, F., Sugawara, D., Kono, Y., Iwashita, T., 2001. The 869 Jogan tsunami deposit and recurrence interval of large-scale tsunami on the Pacific coast of
northeast Japan. J Natural Disaster Sci. 23 (2), 83–88.
Nishimura, T., Miura, S., Tachibana, K., Hashimoto, K., Sato, T., Hori, S., Murakami, E.,
Kono, T., Nida, K., Mishina, M., Hirasawa, T., Miyazaki, S., 2000. Distribution of seismic coupling on the subducting plate boundary in northeastern Japan inferred
from GPS observations. Tectonophysics 323, 217–238. doi:10.1016/S0040-1951
(00)00108-6.
Nishimura, T., Hirasawa, T., Miyazaki, S., Sagiya, T., Tada, T., Miura, S., Tanaka, K., 2004.
Temporal change of interplate coupling in northeastern Japan during 19952002 estimated from continuous GPS observations. Geophys. J. Int. 157, 901–916.
doi:10.1111/j.1365-246X.2004.02159.x.
Richter, C., 1956. Elementary Seismology. W.H. Freeman, San Francisco.
Ruff, L., Kanamori, H., 1983. Seismic coupling and uncoupling at subduction zones. Tectonophysics 99, 99–117.
Simons, M., Minson, S.E., Sladen, A., Ortega, F., Jiang, J., Owen, S.E., Meng, L., Ampuero, J.-P.,
Wei, S., Chu, R., Helmberger, D.V., Kanamori, H., Hetland, E., Moore, A.W., Webb, F.H.,
2011. The 2011 magnitude 9.0 Tohoku-Oki earthquake: mosaicking the megathrust
from seconds to centuries. Sci. Express. doi:10.1126/science.1206731.
van der Hilst, R.D., Engdahl, E.R., Spakman, W., Nolet, G., 1991. Tomographic imaging of
subducted lithosphere below northwest Pacific island arcs. Nature 353, 37–43.
Wang D. and Mori J., in press. Rupture process of the 2011 off the Pacific coast of
Tohoku Earthquake (Mw 9.0) as imaged with back-projection of teleseismic Pwaves, Earth Planets Space, 63.
Widiyantoro, S., Kennett, B.L.N., van der Hilst, R.D., 1999. Seismic tomography with P
and S data reveals lateral variation in the rigidity of deep slabs. Earth Planet Sci.
Lett. 173, 91–100.
Yao, H., Shearer, P., Uchide, T., 2011. Iterative back projection of the rupture and
spatial-temporal distribution of subevents of the 2011 Tohoku. Japan Earthquakehttp://igppweb.ucsd.edu/~huyao/TohokuEarthquake.html Viewed 2011 April
27.
Yoshida Y., Ueno H., Muto D., and Aoki S., in press. Source process of the 2011 Off the
Pacific Coast of Tohoku earthquake with the combination of teleseismic and strong
motion data, Earth Planets Space, 63
Zhang H., Ge Z., and Ding L., in press. Three sub-events composing the 2011 off the Pacific coast of Tohoku Earthquake (Mw 9.0) inferred from rupture imaging by back
projecting teleseismic P waves, Earth Planets Space, 63.
Zhao, D., Huang, Z., Umino, N., Hasegawa, A., Kanamori, H., in press. Structural heterogeneity in the megathrust zone and mechanism of the 2011 Tohoku-oki earthquake (Mw 9.0), Geophys. Res. Lett. 38, L17308. doi:10.1029/2011GL048408.