A trial derivation of seismic plate coupling by focusing on the activity of shallow slow earthquakes
- 1.7k Downloads
To understand the effect of plate coupling on very low-frequency event (VLFE) activity resulting from megathrust earthquakes, we performed long-term multiscale earthquake cycle simulations (including a megathrust earthquake and slow earthquakes) on a 3-D subduction plate boundary model, based on a rate- and state-dependent friction law. Our simulation suggests that quiescence of shallow VLFEs off Miyagi may be explained by the location in the shallow central part of the 2011 Tohoku earthquake because of the locally strong coupling, while observed activation of VLFEs off Iwate (northern part of Tohoku district), Fukushima (southern part of Tohoku district), and Ibaraki (northern part of Kanto district) is explained by the location on the outer rim. The area and duration of the quiescence off Miyagi may be a new clue to evaluate the potential for plate coupling strong enough to cause the next megathrust earthquake.
KeywordsSubduction plate boundary Megathrust earthquake Rate- and state-dependent friction law Superhydrostatic pressure
Broadband seafloor seismometers near the trench (Sugioka et al. 2012) and dense inland networks of highly sensitive seismic stations (Matsuzawa et al. 2012) have enabled observation of very low-frequency earthquakes (VLFEs; Ito et al. 2007) in the shallow transition zone near the trench of subduction plate boundaries (Schwartz and Rokosky 2007) as well as a deep low-frequency earthquake (at a depth about 30 km) (Obara and Sekine 2009). VLFEs are thought to be of a slow earthquake group with a constant moment rate function (Ide et al. 2007).
From numerical simulations of VLFEs on the basis of a rate- and state-dependent friction (RSF) law (Dieterich 1979; Ruina 1983) derived from rock laboratory experiments, Ariyoshi et al. (2012) pointed out that the recurrence interval of VLFEs decreases and the moment release rate increases during the preseismic stage of megathrust earthquakes. Slow-slip events (SSEs), another type of slow earthquake with longer duration than VLFE, have also been modeled in recent studies, which have succeeded in reproducing SSE migration (Shibazaki and Shimamoto 2007) and have shown that the recurrence interval of SSEs also decreases during the preseismic stage of megathrust earthquakes (Matsuzawa et al. 2010).
These simulation results could not directly explain why the VLFEs were inactive off Miyagi, when large postseismic slip occurs around megathrust earthquakes, which is thought to promote VLFEs. Another question is whether the local difference in the shallow VLFE activity off Tohoku may be partly explained by the inhomogeneity of the frictional properties (Matsuzawa et al. 2012). This seems to be inconsistent with VLFE migration at depth, if the frictional property in the shallower part of the transition zone is largely homogenous along the strike direction in the same way as the deeper earthquake (e.g., Schwartz and Rokosky 2007), in which the VLFE migrates for long distances (Obara and Sekine 2009). This is because VLFE migration can be easily reproduced when the frictional properties are homogenous or weakly inhomogeneous along the strike direction, as shown in numerical simulations (e.g., Ariyoshi et al. 2009; Shibazaki and Shimamoto 2007). The stress shadow effect, which is the relationship of present observations of low seismicity rates with the occurrence of past large earthquakes in major asperities (e.g., Maccaferri et al. 2013), may be another factor of VLFE localization. Previous studies have not investigated this effect.
In this study, we perform numerical simulations of multiscale earthquake cycles, including a megathrust earthquake and VLFEs, on a three-dimensional subduction plate boundary with high spatiotemporal resolution to understand the change in activity of VLFEs after megathrust earthquakes.
Outline of modeling slow earthquakes under different plate coupling conditions
The numerical simulation employed in the present study is similar to that described by Ariyoshi et al. (2013). In our study, the motion equation for a subduction plate boundary is described by a quasi-dynamic equilibrium (e.g., Rice 1993) between the shear stress (due to reverse dip slip on the discretized faults) and the frictional stress based on an RSF law. It is well known that RSF laws can represent asperity models (Lay et al. 1982; Scholz 1990), where an asperity is an area on a stuck fault. An earthquake rupture usually begins at an asperity because of friction gap. From the various friction laws available, we adopted the aging (or slowness) law, which enabled us to reproduce slow earthquakes under the condition of weak frictional instability, which was relatively easier than the slip law, as pointed out by Ampuero and Rubin (2008). The details are described later.
To evaluate the stress shadow effect around megathrust earthquakes, we had to estimate the state of plate coupling, where plate coupling χ is defined as Coseismic slip/Total slip for one cycle of the megathrust earthquake (e.g., Kato and Hirasawa 1997) or 1 − Present slip rate V/Plate convergence rate Vpl (e.g., Hyndman et al. 1997). On the relation between the plate coupling and the frictional parameters of RSF law, it is well known that the degree of the plate coupling has a strongly negative correlation with the value of Lmin (e.g., Kato and Hirasawa 1997), if we assume a fixed size of asperity. In this study, we chose two models: a strong coupling model (SCM) and a weak coupling model (WCM). The WCM is the model adopted by Ariyoshi et al. (2013) for the 1944 Tonankai earthquake. The SCM is the same as the WCM, except that the frictional stability parameter γ for large asperity is about 2.6 times greater under the higher value of the frictional parameter b in the RSF law.
For generating slow earthquakes that migrate along the strike direction in both the deep and shallow belts of small asperities, we assumed that σ is described as σ = κ (σlitho − σhydro), where κ, σlitho, and σhydro are a superhydrostatic factor (Ariyoshi et al. 2007, 2012), lithostatic pressure, and hydrostatic pressure, respectively. We also assumed that, locally, σ has a low value of about 30 MPa due to the existence of high pore pressure at a depth near 30 km; we did this by setting the local value of κ to about 0.1 (Ariyoshi et al. 2007), as shown in Figure 2. We will discuss this in more detail in the ‘Discussion’ section.
There is a large difference in scale between large and small asperities, and unless one has an effective strategy, there is a huge calculation cost to repeatedly generating megathrust earthquakes (e.g., Rice 1993). To simplify the model, we assumed that the direction of plate bending and shear displacement was only along the dip direction. Considering previous structural surveys (e.g., Nakanishi et al. 2008), we assume that the dip angle gradually increases from 4° (at the surface) to 27° (at the bottom of the target region), and plate shape is drawn by the B-spline curve, as shown in Figure 2c. Subduction plate motion was represented by a back-slip model (Savage 1983), and it was assumed to be uniform at a rate of Vpl = 4 cm/year, which means that the plate boundary beneath the model region shown in Figure 2a is also subducting at a rate of Vpl. This is seen in the oceanic trenches of Nankai, Sumatra, and Chile, where there are several sources of megathrust earthquakes adjacent to each other along the trenches (e.g., Furumura et al. 2011; Wiseman and Bürgmann 2011; Hayes et al. 2013). To represent the migration of slow earthquakes over the model region, as well as the effect of neighboring megathrust earthquakes, we assumed a periodical condition along the strike direction.
These simplifications enabled us to make great calculation savings by taking advantage of fast Fourier transforms (FFT) based on Green’s function (Okada 1992) when calculating the shear stress, for the dip slip on the array of subfaults along the strike direction (Rice 1993; Ariyoshi et al. 2012). By simplifying the model conditions and using the vectorized multi-FFT algorithm on the Earth Simulator supercomputer, we were able to conduct large-scale simulations of a multiscale earthquake cycle on the subduction plate, divided into 1,210,368 subfaults [2,048 (strike) × 591 (dip)], where cell sizes (hstrike, hdip) for the small asperities and the large asperity are (0.1, 0.5) and (0.1, 0.1) [km], respectively. This was smaller than critical mesh size L b = ηGdc/σb (Rubin and Ampuero 2005), where η = 1.38 (Shibazaki and Shimamoto 2007), G = 30 GPa, d c = dc 1 = 10 and dc 2 = 0.3 [mm], σ = 30 MPa, b = a1 − γ3 = 2.3 [×10−3], L b large = 6 [km], and L b small = 0.18 [km], and sufficiently fine that the numerical simulation results are independent of the mesh size, as determined by comparing with several other models with the mesh size changed.
Effect of a strong coupling model on the activity of shallow VLFEs
From Figure 3, we see that the megathrust earthquake of the SCM ruptures in and around the large asperity, penetrating to the trench, as was observed in the 2011 Tohoku earthquake (Yagi and Fukahata 2011), while that of the WCM ruptures only part of the large asperity, as was seen in the 1944 Tonankai earthquake (Kikuchi et al. 2003) and the 2003 Tokachi earthquake (Yamanaka and Kikuchi 2003).
In Figure 5a,b,d, the migration of slow earthquakes is shown by streaks of warm colors (i.e., yellow through red; see the close-up in Figure 5e,f,g,h), which are mainly composed of VLFEs, as demonstrated by Ariyoshi et al. (2012, 2013). As confirmed by other trial simulations, if frictional instability in the small asperities is greater - that is, the value of γ3 and/or κ2 (κ3) is greater - slip events in the belt of the small asperities approach from slow earthquake to regular earthquake, which is consistent with results of recent simulations (e.g., Shibazaki and Shimamoto 2007; Matsuzawa et al. 2010; Ariyoshi et al. 2009; Ariyoshi et al. 2012) and theoretical analysis of nucleation sizes Lmin and L∞ (Ampuero and Rubin 2008).
With the WCM, Figure 5a,b shows that migration occurs in the shallow belt of small asperity as well as in the deep belt, which is more active from about 10 years before until several years after a megathrust earthquake, due to stress field promotion driven by preseismic and postseismic slips around the large asperity (Ariyoshi et al. 2012). With the SCM, however, Figure 5c shows there is a quiescence in slow earthquakes that appears on the shallow belt during the interseismic stage (about 0 to +40 years), where the whole region is aqua except for the central part, which is blue (|Strike| < 30 km), and there are no streaks of warm colors. This is due to strong plate coupling, as shown in Figure 4a. From other additional simulations, in which we varied the nucleation length of Lmin and L∞ in the large asperity by changing the frictional parameters b = (a1 − γ4) and dc 1 for the large asperity, the lower frictional instability of the large asperity caused the quiescence of the VLFEs to be shorter in time and smaller in space because of the weaker plate coupling (e.g., Kato and Hirasawa 1997). This will be further discussed later.
For slow earthquakes, when comparing shallow (Figure 5e,g) and deep (Figure 5f,h) belts of small asperity, we find that isolated slow earthquakes occur on the shallow belt, as indicated by the green ellipses and that slow earthquake migration becomes active, especially away from the central region (|Strike| > 80 km). We also see that slow earthquake migration occurs widely and continuously on the deep belt, with the approaching onset of a megathrust earthquake, as pointed out by Ariyoshi et al. (2012). To solve the questions raised in the ‘Background’ section, we will apply these results (in the next section) to the VLFE activity observed off Tohoku.
Interpretation of activity change of VLFEs off Tohoku
Figure 1 shows the coseismic slip distribution estimated by Yagi and Fukahata (2011) for the 2011 Tohoku earthquake, the epicenters of the following VLFEs, and our interpretations. In Figure 1, the epicenters of the very low-frequency (VLF) after-events off Fukushima and Ibaraki were primarily located on the outer rim of the coseismic slip of the 2011 Tohoku earthquake. This location of the VLFEs may be explained as a reactivation in the outer part (|Strike| > 80 km), as shown in Figure 4c. Figures 3a and 4c also suggest a quiescence of the VLFEs around the center of strong plate coupling, which may correspond to the fact that VLF after-events were not observed off Miyagi (Matsuzawa et al. 2012). We can also expect activation off Iwate, on the opposite side of the main source region, which gives us a clue for finding previously undiscovered VLFEs from very large events.
As mentioned in the ‘Results’ section, our simulation results show that VLFE quiescence correlates with the degree of frictional instability in the large asperity. From Figure 5, we see that in the SCM, the duration time of the shallow VLFE quiescence (Tq) is about 40 years, with Tr about 140 years. The ratio Tq/Tr seems high, which is due to the strong frictional instability that covers the entire large asperity, as shown in Figure 2a. This assumption may differ from the situation in the 2011 Tohoku earthquake, because the estimated strong-coupled region is thought to be a portion of the main fault off Miyagi, as shown in Figure 1. With the SCM, the frictional parameters [a1, b = (a1 − γ4)] in the large asperity were (2.0, 2.98) [×10−3], which is thought to be a case of the possible largest value of b, because laboratory experiments suggest that b is comparable with a (e.g., Blanpied et al. 1998). In addition to these conditions, considering that Tq/Tr becomes higher when the nucleation lengths Lmin and L∞ are shorter (or frictionally more unstable) in a wider range, we think that the SCM is a possible case that produces the highest value for Tq/Tr.
Considering these results, we propose that the duration and the spatial range of the VLFE quiescence observed off Miyagi may be a new tool to monitor the state of plate coupling by applying the value of Tq/Tr (about 1/3 for the highest case) to actual megathrust earthquakes. If the quiescence lasts longer than 100 years (Tq = 100 years), it would be further evidence that the 2011 Tohoku earthquake is a recurrence of the 869 Jogan earthquake (e.g., Ozawa et al. 2011), because Tr should be greater than 300 years. Note that this monitoring is independent of marine geodetic observations (e.g., Sato et al. 2011; Kido et al. 2011) which have a limited number of seafloor stations.
As mentioned in the ‘Methods’ section, the WCM is based on the 1944 Tonankai earthquake (Ariyoshi et al. 2013), and the SCM is based on the same location of asperity and plate shape as the weak coupling model. This means that the location and size of the large asperity are not based on the 2011 Tohoku earthquake, where the peak of the coseismic slip was located near the trench and had a significantly greater amount of slip, as shown in Figure 1. To discuss quantitatively the change in activity of the VLFEs, we need a more realistic three-dimensional model (e.g., Shibazaki et al. 2011). We intend to pursue this topic in a future study.
Focusing on the shallower part of the plate coupling in Figure 4, we find that the stress shadow effect depends on the distance along the dip as well as along the strike, as discussed in Figure 1. Recently, shallow VLFEs with migration have been observed off Tokachi (e.g., NIED 2014), northward from off Tohoku, where megathrust earthquakes have occurred in 1952 and 2003 (Yamanaka and Kikuchi 2003). The estimated hypocenters of VLFEs determined by NIED (2014) seem to have some dispersion along the dip direction as well as the strike direction, though they use only inland observations. If this is true, it means that we have to consider the dependency of the along-dip distance on the stress shadow effect as well as the along-strike distance from the source region of megathrust earthquakes.
On the deep belt of VLFEs, Figure 5 shows that the VLFEs migrate continuously and widely, which is consistent with observations off southwestern Japan at a depth of about 30 km (Obara and Sekine 2009), but deep VLFEs have not yet been observed off Tohoku. Figure 1 shows the down-dip limit of the interplate earthquakes off Tohoku. Their depth is about 60 km, as shown by the pink bold line. The M 7-class interplate earthquakes occur in the locations indicated by gray contours near the down-dip limit lines. These results are significantly deeper than off southwestern Japan, where deep VLFEs occur at about 30 km (Obara and Sekine 2009). Since large (M 7-class) earthquakes and slow earthquakes (including VLFEs) are thought to occur in the frictionally unstable (γ < 0) and transition zones (γ ~ 0) of subduction plate boundaries (e.g., Schwartz and Rokosky 2007), respectively, this may be the reason for the difference in the observed deep VLFE activity between off Tohoku and southwestern Japan, from the view of frictional transition depth (Evans and Meade 2012).
As described in the ‘Methods’ section, the generation of VLFE depends on the nucleation sizes Lmin and L∞ described as a function of the effective normal stress σ and the characteristic slip distance d c , in addition to frictional stability γ. However, we have not yet robustly estimated the actual value of the characteristic slip distance because the characteristic slip distance is very different in a rock laboratory experiment and in an actual fault of a megathrust earthquake, while the frictional parameters a and b are dimensionless. Therefore, the existence of a zone with high pore pressure is also thought to play an important role in constraining the condition of slow earthquake generation (Obara and Sekine 2009).
In our model, the characteristic slip distance for numerous small asperities (dc 2, dc 3) is assumed to be 0.3 to 0.4 mm, which is significantly shorter than other simulation studies (e.g., Matsuzawa et al. 2010; Shibazaki and Shimamoto 2007), which assume a characteristic distance to be 4 to 10 mm. This difference is derived from uncertainty about the effective normal stress and pore pressure, where the effective normal stress in an SSE patch at a depth of 30 km for Matsuzawa et al. (2010) and Shibazaki and Shimamoto (2007) is assumed to be about 2 to 5 MPa, so as to keep the value of Lmin and L∞ nearly constant.
Off southwestern Japan, high pore pressure may exist locally around a depth of 30 km, based on tomographic studies (Kato et al. 2010) and rock experiments (Katayama et al. 2012). Off Tohoku, Hasegawa et al. (2011) estimated that the fault strength of the 2011 Tohoku earthquake (M w 9: fault size about 200 km × 500 km) was macroscopically 21 to 22 MPa. Because the depth range of the fault for the 2011 Tohoku earthquake is about 10 to 50 km, the estimated fault strength is significantly weak for their depth, from the perspective of rock laboratory experiments under dry conditions (e.g., Kohlstedt et al. 1995). Ariyoshi et al. (2007) pointed out that the observed postseismic slip propagation, at a rate of about 10 km/month in the deep region (>30-km depth) for large interplate earthquakes, can be quantitatively explained by the lower value of the superhydrostatic factor κ (about 0.1 to 0.2).
These results suggest that the zone of high pore pressure in the subduction plate boundary off Tohoku may be widespread rather than local, as they are off southwestern Japan. Thus, detailed mapping of the geofluid (e.g., Iwamori and Nakakuki 2013) is necessary to understand the differences between the slow earthquake activity off Tohoku, southwestern Japan, and the slow earthquake activity of other subduction zones across the globe (e.g., Schwartz and Rokosky 2007). These data can then be incorporated into more fully developed numerical simulations in future studies.
For a megathrust earthquake in which the coseismic slip penetrates to the trench, plate coupling in the postseismic stage will be strong in the region from the central part of the source region to the shallower part toward the trench, which will cause the shallow VLF after-events to be quiescent or to occur infrequently in isolation. On the outer rim, shallow VLF after-events will be reactivated earlier than they will be in the center because of weak plate coupling.
The local differences in shallow VLFE activity can be explained by plate coupling perturbations resulting from the stress shadow effect of megathrust earthquakes with strong plate coupling, even if the frictional property is macroscopically homogeneous in the belt of small asperities along the strike direction.
When the down-dip limit of an interplate earthquake or a frictional transition zone is as deep as 30 km, deep VLFEs occur repeatedly and migrate for long distances along the strike direction. This is because the free-surface effect that amplifies the stress shadow is significant only in the shallow part.
The activated region around the hypocenters of shallow VLF after-events off Iwate, Fukushima, and Ibaraki may indicate the outer rim of the source region of the 2011 Tohoku earthquake, while the quiescent zone off Miyagi may indicate the center side of the source region.
The duration and the spatial range of future quiescent regions off Miyagi are expected to reveal the strength and size of plate coupling, which may be a new clue to evaluate the potential for local plate coupling to be strong enough to cause the next megathrust earthquake.
The authors thank the DONET project members for fruitful discussions. The present study used the Earth Simulator and the supercomputing resources at the Cyberscience Center of Tohoku University. GMT software (Wessel and Smith 1998) was used to draw a number of the figures. The comments of two anonymous reviewers helped improve our manuscript. We are obliged to the editor Junichi Nakajima for expeditious handling. This study was partly supported by the MEXT for Geofluids program and the JSPS for Young Scientists (B), 23710212, and for Specially Promoted Research, 26000002. We thank Midori Hasegawa for the final proofreading.
- Ampuero JP, Rubin AM: Earthquake nucleation on rate state faults – aging and slip laws. J Geophys Res 2008., 113: B01302 doi:10.1029/2007JB005082 B01302 doi:10.1029/2007JB005082Google Scholar
- Ariyoshi K, Hori T, Ampuero JP, Kaneda Y, Matsuzawa T, Hino R, Hasegawa A: Influence of interaction between small asperities on various types of slow earthquakes in a 3-D simulation for a subduction plate boundary. Gondwana Res 2009, 16(3–4):534–544. doi:10.1016/j.gr.2009.03.006CrossRefGoogle Scholar
- Ariyoshi K, Matsuzawa T, Ampuero JP, Nakata R, Hori T, Kaneda Y, Hasegawa A: Migration process of very low-frequency events based on a chain-reaction model and its application to the detection of preseismic slip for megathrust earthquakes. Earth Planets Space 2012, 64(8):693–702. doi:10.5047/eps.2010.09.003CrossRefGoogle Scholar
- Ariyoshi K, Nakata R, Matsuzawa T, Hino R, Hori T, Hasegawa A, Kaneda Y: The detectability of shallow slow earthquakes by the Dense Oceanfloor Network system for Earthquakes and Tsunamis (DONET) in Tonankai district. Japan Mar Geophys Res 2013. doi:10.1007/s11001–013–9192–6Google Scholar
- Evans EL, Meade BJ: Geodetic imaging of coseismic slip and postseismic afterslip: sparsity promoting methods applied to the great Tohoku earthquake. Geophys Res Lett 2012., 39: L11314 doi:10.1029/2012GL051990Google Scholar
- Furumura T, Imai K, Maeda T: A revised tsunami source model for the 1707 Hoei earthquake and simulation of tsunami inundation of Ryujin Lake, Kyushu, Japan. J Geophys Res 2011., 116: B02308 doi:10.1029/2010JB007918Google Scholar
- Igarashi T, Matsuzawa T, Umino N, Hasegawa A: Spatial distribution of focal mechanisms for interplate and intraplate earthquake associated with the subducting Pacific plate beneath the northeastern Japan arc: a triple-planed deep seismic zone. J Geophys Res 2001, 106(B2):2177–2191. doi:10.1029/2000JB900386CrossRefGoogle Scholar
- Iwamori H, Nakakuki T: Fluid processes in subduction zones and water transport to the deep mantle. In Physics and chemistry of the deep Earth, chap 13. Edited by: Karato S. Chichester: Wiley; 2013. doi:10.1002/9781118529492.ch13Google Scholar
- Kato A, Iidaka T, Ikuta R, Yoshida Y, Katsumata K, Iwasaki T, Sakai S, Thurber C, Tsumura N, Yamaoka K, Watanabe T, Kunitomo T, Yamazaki F, Okubo M, Suzuki S, Hirata N: Variations of fluid pressure within the subducting oceanic crust and slow earthquakes. Geophys Res Lett 2010., 37: L14310 doi:10.1029/2010GL043723 L14310 doi:10.1029/2010GL043723Google Scholar
- Kido M, Osada Y, Fujimoto H, Hino R, Ito Y: Trench-normal variation in observed seafloor displacements associated with the 2011 Tohoku-Oki earthquake. Geophys Res Lett 2011., 38: L24303 doi:10.1029/2011GL050057Google Scholar
- Lay T, Kanamori H, Ruff L: The asperity model and the nature of large subduction zone earthquakes. Earthq Pred Res 1982, 1: 3–71.Google Scholar
- Matsuzawa T, Hirose H, Shibazaki B, Obara K: Modeling short- and long-term slow slip events in the seismic cycles of large subduction earthquakes. J Geophys Res 2010., 115(B12): B12301 doi:10.1029/2010JB007566Google Scholar
- Matsuzawa T, Asano Y, Obara K: Shallow very low frequency earthquakes off the Pacific coast of Tohoku, Japan. 2012. Abstract of 2012 Fall Meeting AGU, S41D-07Google Scholar
- Murotani S: Rupture processes of large Fukushima-Oki earthquakes in 1938. University of Tokyo, MSc thesis; 2003.Google Scholar
- Nakanishi A, Kodaira S, Miura S, Ito A, Sato T, Park JO, Kido Y, Kaneda Y: Detailed structural image around splay-fault branching in the Nankai subduction seismogenic zone: results from a high-density ocean bottom seismic survey. J Geophys Res 2008., 113: B03105 B03105Google Scholar
- NIED (National Research Institute for Earth Science and Disaster Prevention): Activity of shallow very-low-frequency earthquakes in and around Japan (June–October, 2011). Report of The Coordinating Committee for Earthquake Prediction (CCEP) 2012, 87: 5–7. written in Japanese with captions in English written in Japanese with captions in EnglishGoogle Scholar
- NIED (National Research Institute for Earth Science and Disaster Prevention): Activity of shallow very-low-frequency earthquakes in and around Japan (May–October, 2013). Report of The Coordinating Committee for Earthquake Prediction (CCEP) 2014, 89: 5–7. written in Japanese with captions in English written in Japanese with captions in EnglishGoogle Scholar
- Okada Y: Internal deformation due to shear and tensile faults in a halfspace. Bull Seism Soc Am 1992, 82: 1018–1040.Google Scholar
- Rubin AM, Ampuero JP: Earthquake nucleation on (aging) rate and state faults. J Geophys Res 2005., 110: B11312 doi:10.1029/2005JB003686 B11312 doi:10.1029/2005JB003686Google Scholar
- Scholz CH: The mechanics of earthquakes and faulting. Cambridge University Press, Cambridge; 1990:439.Google Scholar
- Schwartz SY, Rokosky JM: Slow slip events and seismic tremor at circum-Pacific subduction zones. Rev Geophys 2007., 45: RC3004 doi:10.1029/2006RG000208Google Scholar
- Shibazaki B, Matsuzawa T, Tsusumi A, Ujiie K, Hasegawa A, Ito Y: 3D modeling of the cycle of a great Tohoku-oki earthquake, considering frictional behavior at low to high slip velocities. Geophys Res Lett 2011., 38: L21305 doi:10.1029/2011GL049308Google Scholar
- Yagi Y, Fukahata Y: Rupture process of the 2011 Tohoku-oki earthquake and absolute elastic strain release. Geophys Res Lett 2011., 38: L19307 doi:10.1029/2011GL048701Google Scholar
- Yamanaka Y, Kikuchi M: Asperity map along the subduction zone in northeastern Japan inferred from regional seismic data. J Geophys Res 2004., 109: B07307 doi:10.1029/2003JB002683Google Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.