Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
 Seismic velocity (Vp, Vs) and Poisson's ratio (σ) images show that the hypocenter of the 2008 Iwate-Miyagi earthquake (M7.2) is located in a distinctive zone with low-Vp, slightly low-Vs and low-σ within the upper 10 km depths, where the mechanical strength might be weaker than the normal seismogenic layer with high velocity and low Poisson's ratio immediate below. A prominent low-Vp, low-Vs and high-σ zone is revealed in the lower crust and uppermost mantle under the Iwate mainshock hypocenter, which may reflect fluids resulting from the dehydration of the subducting Pacific slab under northeast Japan. The aqueous component of fluids intruded into the source region may have reduced the mechanical strength of the rock matrix and so triggered the Iwate earthquake. Our results indicate that crustal fluids beneath the seismogenic layer can play an important role in the initiation of large crustal earthquakes.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 On June 14, 2008, a crustal earthquake of JMA (Japan Meteorological Agency) magnitude 7.2 occurred in Iwate Prefecture, the central portion of northeast Japan, which resulted in as much as 4–7 m coseismic slips along a N–S trending fault located east of the volcanic front (H. Suzuki et al., Coseismic slip for the 2008 Iwate earthquake, 2008, available at http://www.k-net.bosai.go.jp/k-net/topics/Iwatemiyaginairiku_080614/inversion/Figs/fig2.html). Focal mechanism solutions of the earthquakes show a WNW–ESE trending axis of compression in the region (http://www.hinet.bosai.go.jp/topics/). The hypocenter area is characterized by mutual interactions of three plates: the Pacific plate, the Eurasian plate and the Okhotsk plate (Figure 1). On the eastern margin of the Japan Sea, the plate boundary between the Eurasian plate and the Okhotsk plate is located near the coastline. Beneath the Iwate earthquake area the Pacific slab is subducting toward WNW with a dipping angle of 30° from the Japan Trench [Zhao et al., 1992; Wang and Zhao, 2005]. The Iwate hypocenter is located at a depth of 8 km in a region with high seismicity. Within 100 km of the source area there have been three earthquakes with JMA magnitude larger than 7 since 1900 (Figure 1). The frequent occurrence of large crustal earthquakes close to the densely populated areas in this region has caused significant damage and loss of human life, which demonstrated an urgent need for better knowledge of crustal heterogeneities for understanding the earthquake nucleation. In order to investigate the deep structure under the source area and its influence on the generation of the 2008 Iwate earthquake, we used a high-quality data set to determine high-resolution 3-D seismic images of the crust and upper mantle under NE Japan, which provides us with useful information for understanding the generating mechanism of the Iwate earthquake.
2. Data and Method
 For tomographic imaging, we used a large number of P- and S-wave arrival times from two groups of earthquakes (Figure 2). One group includes 2055 aftershocks of the 2008 Iwate earthquake which are shallower than 20 km depth and were recorded by the dense Hi-net (High Sensitivity Seismograph Network Japan) and other seismic networks in Japan from June 14 to August 10, 2008. The other group includes 8264 earthquakes at depths of 0–180 km with JMA magnitude greater than 1.5, which were recorded by Hi-net and J-array (J-array Seismogram Data Sets of Japan) from June 1, 2002 to August 10, 2008. For the 2055 aftershocks, the picking accuracy varies from 0.05 to 0.1 s for P-wave data and about 0.1 s for S-wave data. All the aftershocks were recorded by more than 14 stations, and they were selected carefully with a criterion that the uncertainty of the epicenters is <0.5 km while that of the focal depths is <1.5 km. For the second group earthquake, all of them were recorded by more than 30 stations, and the uncertainty of the epicenters is <1.0 km while that of the focal depths is <2.0 km. As a result, 255,394 P-wave and 245,903 S-wave arrival times are selected for the determination of the 3-D seismic velocity (Vp, Vs) and Poisson's ratio (σ) structures in and around the Iwate source area by using a tomographic method [Zhao et al., 1992]. The spatial distribution of the ray paths of the P- and S-wave data indicates that the lower crust and uppermost mantle are well sampled by numerous rays (P*, S*, Pn and Sn) from shallow earthquakes and rays from the intermediate-depth earthquakes (auxiliary material). In our initial velocity model (auxiliary material) we adjusted the depths of the Conrad and Moho discontinuities to those derived from a previous study [Zhao et al., 1992], which resulted in a better 3-D velocity model for the present study area.
 In this study, a total number of 110 seismic stations are used (Figure 2). These stations belong to the Japan University Seismic Network, JMA network, Hi-net, and J-array.
3. Results and Resolution Analyses
 The final images of velocity perturbations at each depth are inverted by using the 1-D calculated velocity model (auxiliary material). The Iwate hypocenter is located in a low-Vp, slightly low-Vs and low-σ zone in the upper crust (Figures 3 and 4) , which may indicate the water-injected segment [Takei, 2002] embedded in the high-Vp, high-Vs and low-σ brittle seismogenic layer [e.g., Brace and Kohlstedt, 1980; Sibson, 1992; Wang and Zhao, 2006a]. Most of the Iwate aftershocks occurred within the presumable water-injected zone (Figures 4a and 4b), coinciding with the previous studies on inland earthquakes that the cutoff-depth of the microseismic activity in the crust is located at 10–15 km depth [e.g., Ito, 1999]. Although the Vp image is uncorrelated with the Vs image along the section AB in the epicenter area, similar features of Vp and Vs images are observed in most of the regions surrounding the source area (Figure 4a). Low-Vp (3–6%), low-Vs (4–6%) and high-σ (5–10%) anomalies are clearly imaged under the Iwate hypocenter at depths of 18–40 km that extends laterally 50–80 km (Figures 4a and 4b). The Vp image is well correlated with the Vs image at these depths. Our results of the low-velocity and high-Poisson's ratio anomalies in the crust below the epicenter area are generally consistent with the previous seismic velocity and attenuation results [e.g., Tsumura et al., 2000; Nakajima et al., 2001, Wang and Zhao, 2005; Matsubara et al., 2008] as well as the geoelectricity studies that revealed a high-conductivity layer in the crust under the study area [e.g., Nakazato et al., 2003].
 To evaluate the resolution of the obtained 3-D velocity model, we conducted the checkerboard resolution tests (CRT) in the study region. The CRT results show that the resolution is high at all the depths under the Iwate source area (auxiliary material), suggesting that the velocity anomalies were adequately resolved by the inversion with the present grid spacing. To confirm the reliability of the inverted images, we also examined the ray path distribution in and around the source area. There are numerous rays crisscrossing the source area from earthquakes in the crust and the subducting Pacific slab outside of the Iwate source area (auxiliary material). We also conducted tomographic inversions and synthetic tests with different grid intervals and initial velocity models. Similar features of the velocity and Poisson's ratio anomalies were revealed in the source area by these inversions.
 Many researchers have suggested that fluids occur widely in the crust and uppermost mantle in subduction zones [e.g., Peacock, 1990; Iwamori, 1998; Nakajima et al., 2001; Nakajima and Hasegawa, 2003; Zhao et al., 2002; Wang and Zhao, 2006b, 2006c]. There are two possible origins of fluids in the crust and uppermost mantle below the 2008 Iwate hypocenter. One is shallow origin such as meteoric water, pore fluids and mineral dehydration in the crust [e.g., Kerrich et al., 1984]. The other is deep origin such as dehydration of the subducting oceanic plate [Zhao et al., 1992, 1996; Wang and Zhao, 2005, 2006a, 2006b, 2006c]. High-resistivity anomaly in upper crust and low-resistivity anomaly in lower crust were revealed in northeast Japan [Mitsuhata et al., 2001; Nakazato et al., 2003], which coincides grossly with the features of the velocity and Poisson's ratio anomalies in the crust revealed by this study. They suggested that the low-resistivity zone at depths below 15 km in the crust can be attributed to fluids. Our seismic images indicate that the low-velocity and high Poisson's ratio anomalies below the Iwate mainshock hypocenter are closely related to the Pacific slab subduction (Figure 4c). In views of the significant Vp and Vs reductions, high Poisson's ratio, and their correlation with a layer of high conductivity in the crust [e.g., Mitsuhata et al., 2001], we consider that the low velocity and high Poisson's ratio features are mainly attributed to fluids associated with dehydration of the subducting Pacific slab.
 Structural heterogeneities in the source area may reflect the physical properties of crustal rocks such as temperature, pressure, composition and fluid content [O'Connell and Budiansky, 1974; Sibson, 1992; Zhao et al., 1996; Wang and Zhao, 2006b, 2006c], which might have influenced the initiation and rupture propagation of this earthquake. The relationship between Poisson's ratio and temperature is still not very clear though it is known that Vp and Vs decrease when temperature increases [e.g., Kern and Richter, 1981; Christensen, 1996]. Poisson's ratio is proved to be very effective for the clarification of the seismogenic behavior of the crust, especially the role of crustal fluids in the nucleation and growth of earthquake rupture. Takei  revealed that Poisson's ratio decreases when water is penetrated into the shallow crust and is texturally equilibrated with the surrounding rocks. Our tomographic images in the source area (Figures 4a and 4b) suggest that the aqueous component segregated from lower crustal fluids is intruded into a shallow brittle zone through texturally equilibrated pores to reduce the mechanical strength of this zone. Continuation of high-σ, low-Vp and low-Vs anomalies below the hypocenter suggests that the aqueous component of fluids ascending from the lowermost crust and uppermost mantle is maintained in the relatively permeable brittle layer at relatively high pressures [Hardebeck and Hauksson, 1999; Takei, 2002]. The rupture initiated at the center of the rupture zone and propagated to southwest and northeast, respectively. The southwestern segment released greater seismic energy than the northeastern segment (http://www.k-net.bosai.go.jp/k-net/topics/Iwatemiyaginairiku_080614/inversion/Figs/fig2.html). The coseismic slip zone coincides well with the high-velocity and low-Poisson's ratio layer identified from our tomographic images (Figure 4b), which we interpret as the brittle seismogenic layer. We suggest that the rupture initiated in a brittle part of the upper crust weakened by intrusion of aqueous fluids from the lower crust and uppermost mantle associated with the slab dehydration (Figure 4).
 Previous studies revealed that the nucleation zones of the earthquakes exhibited low velocity which reflected the presence of overpressurized fluids [e.g., Eberhart-Phillips and Michael, 1993]. When aqueous fluids are penetrated to the brittle seismogenic layer from the lower crust and uppermost mantle, the effective rock strength or fault friction in the intruded portion would be reduced by increasing the pore pressure [Nakajima and Hasegawa, 2003]. A geochemical study indicated that the average amount of fluids from the Pacific slab dehydration is of 0.010 kg fluid/kg mantle, implying fluid extrusion in the crust due to high pore pressure in the lower crust and uppermost mantle in subduction zone [Nakamura et al., 2008]. The Iwate mainshock and most of its aftershocks are located in the low-velocity and low-Poisson's ratio zone (Figures 4a and 4b), where the mechanical strength of materials should be weaker than the normal section of the seismogenic layer with high Vp, high Vs and low σ immediate below. This idea is consistent with the proposed mechanisms for the crust weakening from uprising high-pressure fluids [Zencher et al., 2006] and for the stress concentration between a seismic asperity and the adjacent weakened areas to trigger a major earthquake [Bonafede et al., 2007]. Recent results of detailed petrologic analyses for northeastern Japan also support this scenario [Kimura and Yoshida, 2006]. Therefore, we conclude that the hypocenter area was intruded from below by aqueous fluids, which reduced the strength of crustal rocks and thereby triggered the 2008 Iwate earthquake.
 Our tomographic images show that the M 7.2, 2008 Iwate earthquake and most of its aftershocks occurred in a low-Vp, slightly low-Vs and low-σ belt within the upper 10 km depths, which might reflect a water-injected segment embedded in the brittle seismogenic layer with high velocity and low Poisson's ratio immediate below. Prominent low-Vp, low-Vs and high-σ anomalies are clearly imaged under the Iwate hypocenter area at depths of 18–40 km, indicating the presence of fluids liberated from the Pacific slab dehydration. These results suggest that the mechanical strength in the source area was reduced by the aqueous fluid segregated and ascent from the lower crust and uppermost mantle, which thereby brought the source rock matrix into brittle failure. Our study indicates that aqueous fluid intrusion to the brittle layer in the upper crust extrusion played an important role in the generation of the 2008 Iwate earthquake.
 We appreciate F. Florindo, J. Famiglietti, D. Zhao, M. Bonafede and one anonymous reviewer for their constructive comments and suggestions on the present work. Arrival time data from the JMA, Hi-net and J-array seismic networks are used in this study. This work was partially supported by the projects sponsored by NSFC (50539050) and by SRF for ROCS. SEM (ZS0021, HS0025, 2008DTKT004).