Dynamic rupture simulation of the 2011 off the Pacific coast of Tohoku Earthquake: Multi-event generation within dozens of seconds
© The Society of Geomagnetism and Earth, Planetary and Space Sciences (SGEPSS); The Seismological Society of Japan; The Volcanological Society of Japan; The Geodetic Society of Japan; The Japanese Society for Planetary Sciences; TERRAPUB 2012
Received: 22 December 2011
Accepted: 12 June 2012
Published: 28 January 2013
We focus on a complex rupture process, namely, multi-event generation within about 50 seconds during the 2011 off the Pacific coast of Tohoku Earthquake (Mw 9.0). We perform a 2D dynamic rupture simulation in order to explain physically the rupture process along the cross-section passing through the hypocenter. Realistic velocity structures are introduced into the simulation model. The dynamic parameters are selected by referring to kinematic source inversion results. The first significant event is generated on the deeper side of the fault. The scattered waves, mainly from the free surface, generate a stick-slip around the hypocenter, and then a second significant event is triggered. The synthetic waveforms consist of two major wave groups that are consistent with the observed ground motions.
On March 11, 2011, at 14:46 (JST, GMT+9), the off the Pacific coast of Tohoku Earthquake (Mw 9.0) hit the eastern part of mainland Japan; the earthquake and, in particular, the great tsunami that followed resulted in the death of more than ten thousand people. Strong ground motions during the earthquake were observed over almost the whole region of Japan by K-NET, KiK-net organized by the National Research Institute for Earth Science and Disaster Prevention (Aoi et al., 2011), and by other seismic networks. At least 18 stations observed over 9.8 m/s2 of peak ground accelerations in horizontal components, and two stations observed a seismic intensity over 6.5 on the Japan Meteorological Agency scale.
Several research groups have reported the source rupture process during the earthquake. GPS data and low-frequency ground motions imply that a large slip region is located on the shallower side of the seismic fault (Miyazaki et al., 2011; Yoshida et al., 2011). On the other hand, ground motions band-passed within about 0.1–5 Hz imply that several strong motion generation areas (SMGAs) are located on the deeper side of the seismic fault (Kurahashi and Irikura, 2011; Asano and Iwata, 2012), and the high-frequency radiation area, higher than about 0.5 Hz, is also estimated, by the back-projection method, to be located on the deeper side (Ishii, 2011; Wang and Mori, 2011; Zhang et al., 2011). This indicates that the locations of seismic wave radiation areas depend on the frequency bands (Koper et al., 2011).
Strong ground motions observed in the Miyagi and Iwate prefectures, close to the epicenter region, consist of two major wave groups with an interval of about 50 s (e.g. Kurahashi and Irikura, 2011; Nakahara et al., 2011). The origins of the wave groups are estimated to be on the deeper side of the hypocenter, and the origins are located close to each other (Kurahashi and Irikura, 2011; Asano and Iwata, 2012). However, there is a gap of about 50 s between the rupture times of the corresponding events. The important question is what is the physical mechanism that can explain the time lag. Hereafter, the events are referred to as a “multi-event” in a single earthquake (e.g. Kikuchi and Kanamori, 1982; Sawada et al., 1992).
Duan (2012) has performs 3D spontaneous rupture simulation by considering a subducting seamount just up-dip of the hypocenter. The result shows significant slip near the trench, and the seafloor deformation agrees well with observations (Sato et al., 2011). High-frequency radiations are modeled by three high-strength patches located at the down-dip portion; however, he does not attempt to reproduce details of the high-frequency radiation feature as compared to the actual observed ground motions.
In this study, we carry out a dynamic rupture simulation that models physically consistent slips with stress fields and interface frictions on the fault plane. We especially focus on the numerical modeling of multi-event generation along the dip direction. Ide et al. (2011) have proposed that the major effect can be attributed to a dynamic overshoot due to the fault crossing the free surface, whereas we demonstrate that simulation without the fault crossing the free surface can represent a rupture process consistent with the observed ground motions.
Several kinds of numerical methods have been applied to dynamic rupture simulation, and their accuracies have been discussed via a code validation project (e.g. Harris et al., 2009). Most of the methods are categorized as either domain-based methods (e.g., FDM and FEM), or boundary-based methods (e.g., BIEM). Goto et al. (2010) proposed an alternative hybrid approach: viz., the boundary-domain method (BDM). In this method, instantaneous traction change is represented by the combination of a direct term calculated by a boundary-based method for a homogeneous full space, and a residual term calculated from the differences between two solutions for a target heterogeneity model and for a homogeneous full space model by a domain-based method. BDM guarantees accuracy of the stress field close to the fault plane and its applicability to heterogeneous structures.
However, BDM is not applicable to problems with strong heterogeneities beside the fault plane, e.g. faults crossing the material interfaces and faults located on the interfaces of different materials. Actual seismic faults, including the subduction fault corresponding to the earthquake, fall under these conditions. On the other hand, domain-based methods are applicable to model the strong heterogeneities, whereas much denser grids and/or finer meshes than the usual wave propagation problems are required to ensure the accurate and stable simulation of the dynamic ruptures (e.g. Day et al., 2005). Also, special approaches are required to represent the time-dependent dislocations on the fault plane, such as the split-node technique (e.g. Oglesby and Archuleta, 2000; Dalguer and Day, 2007; Kaneko et al., 2011), and it is difficult to embed a low-angle dipping fault with such approaches. This is because a modification of the local coordinate system for FDM (e.g. Kase and Day, 2006) does not simulate well the low-angle case, and a mesh with a high aspect ratio and/or a triangular mesh for FEM give a worse accuracy (e.g. Yoshida, 2010). This means that the domain-based methods for a low-angle dipping fault remain a problem, and they also require great calculation costs to identify an appropriate parameter set for spontaneous rupture simulation.
3. Numerical Model
The tomography model by Yamamoto et al. (2011) is applied to the crust velocity structure on the cross-section. Figure 2 also shows the applied velocity model of a P -wave (V p ) and an S-wave (V s ). The plate boundary of the North American and Pacific plates (Ito et al., 2005; Miura et al., 2005) is also plotted in Fig. 2. We set a planar fault model with a width of 180 km, a dip of 12°, and a hypocenter depth of 17 km, so that the fault model fits the plate boundary around the hypocenter, as shown in Fig. 2. The dip angle and the hypocenter depth coincide with a unified source model by Koketsu et al. (2011). The top of the fault model is located beneath the free surface at a depth of 367 m. This means that the fault edge does not touch the free surface. This assumption may not be realistic, however if the high-frequency radiations can be simulated well without crossing the fault and free surface, the effect of crossing is not always required to simulate the high-frequency radiation. Also, there are no evidences of high-frequency radiations from the top of the fault (e.g. Koper et al., 2011; Asano and Iwata, 2012). If we focus on the high-frequency radiation, the crossing is of not much significance in the simulation.
The boundary integral equation method (BIEM) (Tada and Yamashita, 1997; Goto et al., 2010) is applied as the boundary-based method, and the finite-difference method (FDM) (Levander, 1988) is applied as the domain-based method in pBDM. The velocity models for BIEM are set to be 7.34 km/s for V p and 4.20 km/s for V s , which are the average values on the deeper side of the fault. Density models for both BIEM and FDM are evaluated from the empirical density-Vp relation of Ludwig et al. (1970), which has been validated by Miura et al. (2005). The numerical scheme of FDM is of fourth-order accuracy in space, and of second-order accuracy in time using staggered grids. The element size of BIEM is 87.9 m, and the grid intervals of FDM are homogeneously 100 m. The time step intervals of both methods are set at 4.77 × 10−3 s. The Courant-Friedrichs-Lewy conditions for BIEM and FDM are 0.399 and 0.40, respectively. The maximum support frequency of FDM is 1.34 Hz. Internal damping of the crust structure is set to be Q = 100f and installed by using the approach of Graves (1996).
Friction on the fault plane obeys a linear slip-weakening friction law (Ida, 1972), and the rate-dependency of friction is not considered. In such a case, the dynamic parameters consist of the initial stress ratio μ0, static and dynamic friction coefficients μ p , μ r , and the slip-weakening distance D C . The stress ratio is defined by dividing the initial shear traction by the normal stress. When the shear traction is unloaded, the value of the last friction strength is stored. After the traction reaches the friction strength again, the slip is allowed. This assumption is consistent with a standard elasto-plastic constitutive model.
4. Results and Discussions
At the same time, the scattered waves reach the fault plane after 40 s, and generate a stick-slip with low slip-rates around the hypocenter. Here, the stick-slip means that slips in the region repeat the stopping and rupturing. This causes a crack growth gradually expanding the ruptured area in the deeper direction, and then the second significant event is triggered at around −30 km after about 55 s. This event generates seismic waves of large amplitude propagating mainly in the landward direction.
The barrier set in the deeper side of the fault may play an important role in the generation of the multi-event. In contrast, no clear barriers are seen in the rupture processes estimated by kinematic source inversions. We suppose that the barrier of our model is consistent with the rupture processes because the spatial resolutions in the inversions may not be sufficient to reveal it, this has been discussed in the Source Inversion Validation project (Page et al., 2011). For the earthquake, we select four inversion results by Ide et al. (2011), Miyazaki et al. (2011), Suzuki et al. (2011), and Yoshida et al. (2011), to compare the maximum spatial resolutions in their inversion schemes. The slip distributions are represented by the superposition of the basis functions such as the box-car basis (Suzuki et al., 2011; Yoshida et al., 2011) and the triangle basis (Ide et al., 2011; Miyazaki et al., 2011). The spatial sizes, e.g. the size of sub-faults (Suzuki et al., 2011; Yoshida et al., 2011), are usually selected such that the dominant frequency caused by the spatial periodicity is well outside the frequency range of analysis (e.g. Sekiguchi et al., 2000). In addition, the coefficients for each basis are constrained by prior information in the Bayesian approach such as Akaike’s Bayesian Information Criterion (ABIC) (e.g. Ide and Takeo, 1997; Sekiguchi et al., 2000), and the constraint results in the slip distribution being smoothed.
Kurahashi and Irikura (2011) and Asano and Iwata (2012), have estimated the locations of SMGAs. Constant stress drops and square areas of SMGAs are assumed, and the locations, rise time, rupture starting point, and the stress drop are estimated by a trial-and-error approach (Kurahashi and Irikura, 2011), and a grid search (Asano and Iwata, 2012). The procedures infer that a spatial variation within the size of SMGA is averaged. In addition, the relative locations of the two SMGAs are opposite to each other. This implies that the accuracy of SMGA locations is the order of SMGA size. Therefore, it may be difficult to distinguish the barrier based on the results from kinematic source inversions.
In order to clarify the effect of the free surface on the generation of the second significant event, we perform two other reference simulations, case 1: the same dynamic parameters without a free surface, and case 2: different dynamic parameters with a free surface. The velocity model beneath a depth of 0 km is set to be the same model as in the previous simulation, and the model above 0 km for case 1 is equal to the velocities of the uppermost layer of the previous model. The dynamic parameters, fault geometry, and calculation conditions are the same as in the previous simulation for case 1, while both μ0 and μ r from −40 to −20 km are reduced to 0.35 for case 2.
The accuracy of the simulated result in the shallower region may not be well guaranteed. Furthermore, the modeled fault does not reach the free surface. These factors result in an underestimation of the final slips in the shallower region compared to the inversion results. However, we point out that a significant effect from the shallower region is not required for the multi-event generation.
Notice that the dynamic parameters are not validated by comparing the amplitudes between observed and synthetic waveforms because the 2D P-SV wave field does not simulate the actual geometric spreading in 3D. The parameter set of our original model is the best we have found, but this may not be the optimum set. Therefore, it is worth showing the justifiability of our choice of parameter set.
We simulate the dynamic rupture process that occurred during the 2011 off the Pacific coast of Tohoku Earthquake on the cross-section passing through the hypocenter. The first significant event occurred at around −60 km along the fault width direction. The waves, scattered mainly from the free surface, come back to the fault plane in about 25 s, and generate a stick-slip with small amplitude around the hypocenter. The stick-slips expand the ruptured area to the deeper side, and then the second significant event is triggered. The synthetic waveforms consist of two major wave groups, which are similar to the observations.
We thank two anonymous reviewers and the editor Prof. Takashi Furumura for their thorough comments. We are grateful to the following persons for the useful information they provided us and for helpful discussions: Professor Kazuro Hirahara, Professor Sumio Sawada, Professor Shin’ichi Miyazaki, Dr. Kimiyuki Asano, Dr. Wataru Suzuki, Dr. Hiroe Miyake, and many other researchers. This study was supported in part by the Grant-in-Aid for Young Scientists B of the Japan Society for the Promotion of Science. We offer our thanks to NIED for the contributions to the strong ground motion observations, namely, K-NET, KiK-net.
- Aochi, H. and E. Fukuyama, Three-dimensional nonplanar simulation of the 1992 Landers earthquake, J. Geophys. Res., 107, 2039, 2002.View ArticleGoogle Scholar
- Aoi, S., T. Kunugi, W. Suzuki, N. Morikawa, H. Nakamura, N. Pulido, K. Shiomi, and H. Fujiwara, Strong motion characteristics of the 2011 Tohoku-oki earthquake from K-NET and KiK-net, Proc. of 2011 SSA Annual Meeting, 2011.Google Scholar
- Asano, K. and T. Iwata, Source model for strong ground motion generation in the frequency range 0.1–10 Hz during the 2011 Tohoku earthquake, Earth Planets Space, 64, this issue, 1111–1123, 2012.View ArticleGoogle Scholar
- Dalguer, L. A. and S. M. Day, Staggered-grid split-node method for spontaneous rupture simulation, J. Geophys. Res., 112, B02302, 2007.Google Scholar
- Day, S. M., L. A. Dalguer, N. Lapusta, and Y. Liu, Comparison of finite difference and boundary integral solutions to three-dimensional spontaneous rupture,J. Geophys. Res., 110, B12307, 2005.View ArticleGoogle Scholar
- Duan, B., Dynamic rupture of the 2011 Mw9.0 Tohoku-oki earthquake: roles of a possible subducting seamount, J. Geophys. Res., 2012 (in printing).Google Scholar
- Fukuyama, E. and R. Madariaga, Rupture dynamics of a planar fault in a 3D elastic medium: rate- and slip-weakening friction, Bull. Seismol. Soc. Am., 88, 1–17, 1998.Google Scholar
- Fukuyama, E., R. Ando, C. Hashimoto, S. Aoi, and M. Matsu’ura, A physics-based simulation of the 2003 Tokachi-oki, Japan, earthquake to predict strong ground motions, Bull. Seismol. Soc. Am., 99, 3150–3171, 2009.View ArticleGoogle Scholar
- Goto, H., L. Ramírez-Guzmán, and J. Bielak, Simulation of spontaneous rupture based on a combined boundary integral equation method and finite element method approach: SH and P-SV cases, Geophys. J. Int., 183, 975–1004, 2010.View ArticleGoogle Scholar
- Graves, R. W., Simulating seismic wave propagation in 3D elastic media using staggered-grid finite differences, Bull. Seismol. Soc. Am., 86, 1091–1106, 1996.Google Scholar
- Harris, R. A., M. Barall, R. Archuleta, E. Dunham, B. Aagaard, J. P. Ampuero, H. Bhat, V. Cruz-Atienza, L. Dalguer, P. Dawson, S. Day, B. Duan, G. Ely, Y Kaneko, Y Kase, N. Lapusta, Y Liu, S. Ma, D. Oglesby, K. Olsen, A. Pitarka, S. Song, and E. Templeton, The SCEC/USGS dynamic earthquake rupture code verification exercise, Seismol. Res. Lett., 80, 119–126, 2009.View ArticleGoogle Scholar
- Ida, Y, Cohesive force across the tip of a longitudinal-shear crack and Griffith’s specific surface energy, J. Geophys. Res., 77, 3796–3805, 1972.View ArticleGoogle Scholar
- Ide, S. and M. Takeo, Determination of constitutive relations of fault slip based on seismic wave analysis, J. Geophys. Res., 102, 27379–27391, 1997.View ArticleGoogle Scholar
- Ide, S., A. Baltay, and G. C. Beroza, Shallow dynamic overshoot and energetic deep rupture in the 2011 Mw 9.0 Tohoku-Oki earthquake, Science, 332, 1426–1429, 2011.View ArticleGoogle Scholar
- Ishii, M., High-frequency rupture properties of the Mw 9.0 off the Pacific coast of Tohoku Earthquake, Earth Planets Space, 63, 609–614, 2011.View ArticleGoogle Scholar
- Ito, A., G. Fujie, S. Miura, S. Kodaira, R. Hino, and Y. Kaneda, Bending of the subducting oceanic plate and its implification for rupture propagation of large interplate earthquake off Miyagi, Japan, in the Japan trench subduction zone, Geophys. Res. Lett., 32, L05310, 2005.Google Scholar
- Kaneko, Y, J. P. Ampuero, and N. Lapusta, Spectral-element simulations of long-term fault slip: effect of low-rigidity layers on earthquake-cycle dynamics, J. Geophys. Res., 116, B10313, 2011.View ArticleGoogle Scholar
- Kase, Y and S. M. Day, Spontaneous rupture processes on a bending fault, Geophys. Res. Lett., 33, L10302, 2006.Google Scholar
- Kikuchi, M. and H. Kanamori, Inversion of complex body waves, Bull. Seismol. Soc. Am., 72, 491–506, 1982.Google Scholar
- Koketsu, K., Y Yokota, N. Nishimura, Y Yagi, S. Miyazaki, K. Satake, Y Fujii, H. Miyake, S. Sakai, Y Yamanaka, and T. Okada, A unified source model for the 2011 Tohoku earthquake, Earth Planet. Sci. Lett., 310, 480–487, 2011.View ArticleGoogle Scholar
- Koper, K. D., A. R. Hutko, T. Lay, C. J. Ammon, and H. Kanamori, Frequency-dependent rupture process of the 2011 Mw 9.0 Tohoku Earthquake: Comparison of short-period P wave backprojection images and broadband seismic rupture models, Earth Planets Space, 63, 599–602, 2011.View ArticleGoogle Scholar
- Kurahashi, S. and K. Irikura, Source model for generating strong ground motions during the 2011 off the Pacific coast of Tohoku Earthquake, Earth Planets Space, 63, 571–576, 2011.View ArticleGoogle Scholar
- Levander, A. R., Fourth-order finite-difference P-SV seismograms, Geophysics, 53, 1425–1436, 1988.View ArticleGoogle Scholar
- Ludwig, W. J., J. E. Nafe, and C. L. Drake, Seismic refraction, in The Sea, vol. 4, 54–84, Wiley-Inter-science, 1970.Google Scholar
- Miura, S., N. Takahashi, A. Nakanishi, T. Tsuru, S. Kodaira, and Y. Kaneda, Structural characteristics off Miyagi forearc region, the Japan Trench seismogenic zone, deduced from a wide-angle reflection and refraction study, Tectonophysics,407, 165–188, 2005.View ArticleGoogle Scholar
- Miyazaki, S., J. J. McGuire, and P. Segall, Seismic and aseismic fault slip before and during the 2011 off the Pacific coast of Tohoku Earthquake, Earth Planets Space, 63, 637–642, 2011.View ArticleGoogle Scholar
- Nakahara, H., H. Sato, T. Nishimura, and H. Fujiwara, Direct observation of rupture propagation during the 2011 off the Pacific coast of Tohoku Earthquake (Mw 9.0) using a small seismic array, Earth Planets Space, 63, 589–594, 2011.View ArticleGoogle Scholar
- Oglesby, D. and R. Archuleta, Dynamics of dip-slip faulting: explorations in two dimensions, Geophys. J. Int., 105, 13643–13653, 2000.View ArticleGoogle Scholar
- Page, M., P. M. Mai, and D. Schorlemmer, Testing earthquake source inversion methodologies, Eos Trans. AGU, 92, 75, 2011.View ArticleGoogle Scholar
- Sato, M., T. Ishikawa, N. Ujihara, S. Yoshida, M. Fujita, M. Mochizuki, and A. Asada, Displacement above the hypocenter of the 2011 Tohoku-Oki earthquake, Science, 332, 1395, 2011.View ArticleGoogle Scholar
- Sawada, S., C. S. Fu, T. Kitano, and S. Yoshikawa, Multi-event inversion analysis for simpler representation of source mechanism, Proc. of 10th WCEE, 757–760, 1992.Google Scholar
- Sekiguchi, H., K. Irikura and T. Iwata, Fault geometry at the rupture termination of the 1995 Hyogo-ken Nanbu earthquake, Bull. Seismol. Soc. Am., 90, 117–133, 2000.View ArticleGoogle Scholar
- Suzuki, W., S. Aoi, H. Sekiguchi, and T. Kunugi, Rupture process of the 2011 Tohoku-Oki mega-thrust earthquake (M9.0) inverted from strong-motion data, Geophys. Res. Lett., 38, L00G16, 2011.Google Scholar
- Tada, T. and T. Yamashita, Non-hypersingular boundary integral equations for two-dimensional non-planar crack analysis, Geophys. J. Int., 130, 269–282, 1997.View ArticleGoogle Scholar
- Wada, K. and H. Goto, Generation mechanism of surface and buried faults: effect of plasticity in a shallow crust structure, Bull. Seismol. Soc. Am., 2012 (in printing).Google Scholar
- Wang, D. and J. Mori, Rupture process of the 2011 off the Pacific coast of Tohoku Earthquake (Mw 9.0) as imaged with back-projection of teleseismic P-waves, Earth Planets Space, 63, 603–607, 2011.View ArticleGoogle Scholar
- Yamamoto, Y., R. Hino, and M. Shinohara, Mantle wedge structure in the Miyagi Prefecture forearc region, central northeastern Japan arc, and its relation to corner-flow pattern and interplate coupling, J. Geophys. Res., 116, B10310, 2011.View ArticleGoogle Scholar
- Yoshida, K., K. Miyakoshi, and K. Irikura, Source process of the 2011 off the Pacific coast of Tohoku Earthquake inferred from waveform inversion with long-period strong-motion records, Earth Planets Space, 63, 577–582, 2011.View ArticleGoogle Scholar
- Yoshida, N., Earthquake Response Analysis of Ground, Kajima, Tokyo, 2010 (in Japanese).Google Scholar
- Zhang, H., Z. Ge, and L. Ding, 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, 595–598, 2011.View ArticleGoogle Scholar