Seismic imaging of the Southwest Japan arc from the Nankai trough to the Japan Sea

Physics of the Earth and Planetary Interiors 216 (2013) 59–73

Contents lists available at SciVerse ScienceDirect

Physics of the Earth and Planetary Interiors journal homepage: www.elsevier.com/locate/pepi

Seismic imaging of the Southwest Japan arc from the Nankai trough to the Japan Sea Xin Liu a,b,c,⇑, Dapeng Zhao a,⇑, Sanzhong Li b,c a

Department of Geophysics, Tohoku University, Sendai 980-8578, Japan Department of Marine Geosciences, Ocean University of China, Qingdao 266100, China c The Key Lab of Seabed Resource and Exploration Techniques, Ministry of Education, Qingdao 266100, China b

a r t i c l e

i n f o

Article history: Received 20 September 2012 Received in revised form 30 December 2012 Accepted 3 January 2013 Available online 10 January 2013 Edited by George Helffrich Keywords: Southwest Japan sP depth phase Megathrust zone Earthquakes Asperity Subduction zones Nankai trough

a b s t r a c t Detailed three-dimensional P- and S-wave velocity (Vp and Vs) models of the entire Southwest Japan arc from the Nankai trough to the Japan Sea are determined for the first time using a large number of highquality arrival-time data from local earthquakes. The suboceanic earthquakes used in the tomographic inversion were relocated precisely using sP depth phase data. Our results show that strong lateral heterogeneities exist in the interplate megathrust zone under the Nankai forearc. Large interplate earthquakes mainly occurred in or around high-velocity (high-V) patches in the megathrust zone. These high-V patches may represent asperities formed by the subducted oceanic ridges and seamounts. Low-velocity (low-V) zones in the megathrust zone may contain sediments and fluids associated with slab dehydration and so become weakly coupled areas. Our results also show that the coseismic slip distributions of some megathrust earthquakes are not limited in the high-V patches (asperities) where the ruptures initiated. Because of the weak interplate coupling in the low-V areas, the rupture of an interplate earthquake could unimpededly pass through the low-V anomalies and so result in a great megathrust earthquake. Ó 2013 Elsevier B.V. All rights reserved.

1. Introduction Along the Nankai trough off Southwest (SW) Japan, a relatively young (ca. 15–50 Ma) oceanic plate, the Philippine Sea plate, has been subducting beneath the Eurasian plate since ca. 15 Ma (Deschamps and Lallemand, 2002; Hall, 2002; Sdrolias et al., 2004) (Fig. 1). Many active arc volcanoes exist on this continental margin and form a clear volcanic front (Fig. 1). Earthquakes in the SW Japan region occur to a depth of 200 km under Kyushu Island and 80 km beneath Shikoku and SW Honshu. Deeper earthquakes under this region take place within the subducting Pacific slab, which is located beneath the Eurasian plate and the subducting Philippine Sea slab (Fig. 1). Many seismic tomographic studies have been made to characterize this active continental margin (see a recent review by Zhao et al., 2011a). A recent teleseismic tomography shows that the Philippine Sea slab has subducted aseismically down to 430 km depth under the Japan Sea and the East China Sea (Zhao et al., 2012). P-wave anisotropy tomography revealed trench-normal fast-velocity directions in the mantle wedge under Kyushu, which may reflect corner flow driven by the active subduction of the Philippine Sea plate (Wang and Zhao, 2012). These previous tomo⇑ Corresponding authors. Address: Department of Geophysics, Tohoku University, Sendai 980-8578, Japan (X. Liu). Tel.: +81 22 225 1950; fax: +81 22 264 3292. E-mail addresses: [email protected] (X. Liu), [email protected] (D. Zhao). 0031-9201/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.pepi.2013.01.003

graphic studies mainly focused on the inland region of the Nankai subduction zone, where the dense seismic network (Fig. 2a) allows for high-resolution seismic imaging. Although many two-dimensional seismic profiles and local three-dimensional (3-D) seismic reflection data have been acquired in the forearc region under the Philippine Sea (e.g., Kodaira et al., 2000, 2002; Bangs et al., 2009), detailed 3-D velocity images of the entire SW Japan arc, from the Nankai trough to the Japan Sea, have not been determined because few ocean-bottom-seismometer (OBS) stations are deployed in the Philippine Sea and the Japan Sea (Fig. 2). Therefore, the suboceanic earthquakes cannot be located accurately with the routine procedure of the land-based seismic network. The suboceanic earthquakes that occurred outside a seismic network could be located precisely (with hypocenter uncertainty <3 km) using the sP depth-phase data which are very sensitive to the focal depth because the bouncing point of the sP depth phase is very close to the earthquake epicenter (Umino et al., 1995; Zhao et al., 2002, 2007; Gamage et al., 2009; Huang et al., 2010). P-wave arrival times from the suboceanic earthquakes precisely relocated with sP depth phases were used to determine the 3-D Vp structure under the northeast Japan forearc region beneath the Pacific Ocean, and this approach was suggested to be a way of tomographic imaging outside a seismic network (Zhao et al., 2002, 2007). The sP depth-phase data were also used in suboceanic earthquake location (Bai et al., 2006; Tahara et al., 2006) and tomographic imaging (Wang and Zhao, 2006a,b) in some areas of the Nankai subduction

60

X. Liu et al. / Physics of the Earth and Planetary Interiors 216 (2013) 59–73

Fig. 1. Tectonic background of the Nankai subduction zone. The white bold sawtooth line denotes the Nankai trough. The inset map shows the simplified tectonic background of the study area (blue box). The black and dashed bold lines on the inset map denote the plate boundaries. The topography data are derived from the GEBCO_08 Grid, version 20100927, http://www.gebco.net. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

zone. Results obtained with this method have revealed a close relationship between structural heterogeneities and large earthquake nucleation in the megathrust zone, such as the great 2011 Tohoku-oki earthquake (Mw 9.0) (Zhao et al., 2011b). In this work we have determined, for the first time, detailed 3-D images of Vp and Vs under the entire Nankai subduction zone from the Nankai trough to the Japan Sea using a large number of highquality arrival-time data from many local inland earthquakes as well as suboceanic events which are relocated with sP depth phases. We have made great efforts to collect and use much more data than any of the previous tomographic studies in this region, and so our present results shed new light on the arc magmatism, seismotectonics and dynamics of the Nankai subduction zone.

2. Data and method In this study we used 876 permanent seismic stations (Fig. 2a) which belong to the Japanese Kiban seismic network (Okada et al., 2004). We selected 6622 earthquakes from all the available events that occurred in the study region during 1997–2012 (Fig. 2). These events are composed of four groups. The first group includes 2838 events that occurred during October 1997 to October 2007 under the seismic network on the Japan Islands (green dots in Fig. 2), which are selected from the data set used by Zhao et al. (2011a). This data set contains more P- and S-wave arrival times than those

released by the Japan Meteorological Agency (JMA) Unified catalog, thanks to the great efforts made by the staffs of Tohoku University who picked up all the clear P and S arrivals from the original seismograms. The second group consists of 1557 earthquakes (yellow dots in Fig. 2) that occurred under the seismic network on the Japan Islands from June 2002 to March 2012, which were directly selected from the JMA Unified catalog. The uncertainties of the hypocenter locations of these two groups are estimated to be smaller than 2 km. The third group contains 1675 shallow suboceanic events (light red dots in Fig. 2) that occurred beneath the East China Sea and the Philippine Sea during June 2002 to September 2011, which are selected from the data set used by Liu et al. (in review). This data set contains 4018 sP depth phases and more P- and Swave arrival times than those released by the JMA Unified catalog. The fourth group includes 552 shallow suboceanic earthquakes (red dots in Fig. 2) that occurred beneath the Philippine Sea and the Japan Sea during June 2002 to March 2012. Following the criteria established by the previous studies for the collection of sP depth phases (e.g., Umino and Hasegawa, 1994; Umino et al., 1995; Wang and Zhao, 2005, 2006a,b), we examined a large amount of threecomponent seismograms of over 2500 shallow earthquakes (MJMA P 2.0) that occurred beneath the Philippine Sea and the Japan Sea. As a result, we selected 552 events to form the fourth group for which 1559 sP depth phases were collected (Fig. 3). The picking accuracy of the sP depth phases is estimated to be 0.1–0.2 s. In order to improve the ray path coverage in the area outside the

61

X. Liu et al. / Physics of the Earth and Planetary Interiors 216 (2013) 59–73

Longitude 130˚

132˚

134˚

136˚

138˚

(a) 36˚

Japan Sea

Latitude

34˚

gh

32˚

a

nk

Na

rou iT

Philippine Sea 30˚

Depth (km) 130˚ 36˚

132˚

134˚

136˚

138˚

0

100

200

(b)

36

34

32˚

32

Latitude

34˚

(d)

Depth (km)

30˚ 0

30 40˚

100

200

30˚

(c)

(e) 130

132

134

136

138

130˚

140˚

Fig. 2. (a) Distribution of the 876 seismic stations (blue squares) used in this study. (b) Distribution of the 6622 earthquakes (dots) used in this study. Green dots (Group-1) denote 2838 events selected from the data sets used by Zhao et al. (2011a). Yellow dots (Group-2) show 1557 events that occurred under the seismic network. Light red dots (Group-3) denote 1675 suboceanic earthquakes selected from the data sets used by Liu et al. (in review). Red dots (Group-4) denote 552 suboceanic earthquakes that are relocated by using the sP depth-phase data. (c) East–west and (d) north–south vertical cross-sections of the earthquakes shown in (b). (e) Simplified tectonic background of the study area (green box). The black and dashed bold lines denote the plate boundaries. The red solid triangles denote the active arc volcanoes. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

seismic network, we also collected clear P- and S-wave arrival times from the original three-component seismograms of these 552 events recorded by the seismic stations on the Japan Islands, in addition to the P- and S-wave data released by the JMA Unified catalog. Thus we have 44,944 P- and 26,295 S-wave arrival times

from the 552 events, and so the average number of arrival times for each event is 129.1. In contrast, the 1557 events in the second group have 44,136 P- and 39,949 S-wave arrival times which are released by the JMA Unified catalog, and their average number of data is 54.0, nearly a half of that for the events in the fourth group.

62

X. Liu et al. / Physics of the Earth and Planetary Interiors 216 (2013) 59–73 o

o

Event: 2004.04.05 17:41:13.57 s 31.355 N 133.787 E

27.00 km

M 2.5 36˚

N.TSSH 176.2 km

34˚

TOSASH 191.9 km

32˚

Event

N.OOTH

Vertical component

194.3 km

130˚

132˚

134˚

136˚

138˚

N.NAKH 206.8 km

sP

P

KC.MUT

S

215.2 km KC.KUB 218.6 km N.TSMH 221.9 km N.TSYH 225.5 km KUBOKA 228.0 km 10 s

0s

30 s

20 s

40 s

Fig. 3. An example of vertical-component seismograms of a suboceanic earthquake that occurred beneath the Philippine Sea. Hypocenter parameters of this earthquake are shown above the seismograms. The station codes and epicentral distances are shown on the left. Clear sP depth phases labeled by blue dots are visible at 7 s after the first Pwave arrivals labeled by black dots. The inset map shows the location of the epicenter (blue star) and the solid triangles denote the active arc volcanoes. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 1 Data sets used in this study (October 1997 to March 2012). Southwest Japan

Events

P-wave

S-wave

sP depth phase

Average number of P- and S-wave data per event

MJMA

Group-1a Group-2b Group-3c Group-4d

2838 1557 1675 552

155,356 44,136 71,670 44,944

116,519 39,949 63,218 26,295

4018 1559

95.8 54.0 80.5 129.1

P0.0 P0.0 P2.0 P2.0

Total

6622

316,106

245,981

5577

a

Earthquakes that occurred under the seismic network during October 1997 to October 2007 (green dots in Fig. 2), selected from the data sets used by Zhao et al. (2011a). b Earthquakes that occurred under the seismic network during June 2002 to March 2012 (yellow dots in Fig. 2), selected from the JMA Unified catalog. c Earthquakes that occurred out of the seismic network during June 2002 to September 2011 (light red dots in Fig. 2), selected from the data sets used by Liu et al. (in review). d Earthquakes that occurred out of the seismic network during June 2002 to March 2012 (red dots in Fig. 2), selected from the JMA Unified catalog.

As a result, in this work we have used a total of 6622 events (Fig. 2) which generated 316,106 P- and 245,981 S-wave arrival times recorded by 876 seismic stations on the Japan Islands (Table 1). All the events were recorded by over 10 seismic stations. The picking accuracy of the arrival times is estimated to be 0.05– 0.15 s for P-wave data, and 0.1–0.2 s for S-wave data. The tomographic method of Zhao et al. (1992, 2007) was used to determine 3-D Vp and Vs models under the study region (Fig. 1). To express the 3-D velocity structure, a 3-D grid net is set up in the modeling space, following the approach of Zhao et al. (1992). The starting one-dimensional velocity model for the tomographic inversion is derived from Zhao et al. (2007). Vp and Vs are 6.0 and 3.5 km/s in the upper crust, and 6.7 and 3.8 km/s in the lower crust. The J-B velocity model (Jeffreys and Bullen, 1940) is adopted for the mantle in the starting model. For the subducting Philippine Sea slab, the initial Vp and Vs are assigned to be 3% and 5% faster than the mantle velocity at the same depth, respectively, according to the results of the reductions of P- and S-wave root-mean-square (RMS) travel-time residuals (Fig. S1a). Hypocentral parameters and velocity perturbations at the grid nodes from the starting velocity model are taken as unknown parameters. The velocity perturbation at any point in the model is calculated by linearly interpolating the velocity perturbations at the eight grid nodes surrounding

that point. An efficient 3-D ray tracing technique (Zhao et al., 1992) is used to compute travel times and ray paths accurately. Lateral depth variations of the Conrad and Moho discontinuities modified from Katsumata (2010) and the upper boundary of the subducting Philippine Sea slab derived from Zhao et al. (2012) are taken into account in the 3-D ray tracing and the tomographic inversion, because the existence of these three discontinuities under the Japan Islands has been established well and their geometries have been determined. The sediment layer depth from the model CRUST2.0 (Laske et al., 2003), station elevations and the topography are also taken into account in the 3-D ray tracing, earthquake relocation, and the tomographic inversion. The damped least-squares method is used to solve the large and sparse system of observation equations that relate the observed arrival times to the unknown parameters. Hypocenter parameters are re-determined for the obtained 3-D Vp and Vs models in an iterative process using the P, S and sP arrival-time data (Zhao et al., 2007). Smoothing and damping regularizations are adopted to suppress the dramatic short-scale variations of the velocity anomalies (Zhao et al., 2007). To minimize the effect of the uncertainties of the initial velocity model, the final velocity perturbation at each grid node was calculated from the average of the obtained velocity perturbations at each depth. The P-wave RMS travel-time residuals before and after the

X. Liu et al. / Physics of the Earth and Planetary Interiors 216 (2013) 59–73

tomographic inversion are 0.644 and 0.359 s, and the corresponding S-wave RMS residuals are 0.723 and 0.509 s, respectively. The variance reductions of P- and S-wave travel-time residuals are 69% and 50%, respectively.

3. Analysis and results Similar to the results of Umino et al. (1995) and Gamage et al. (2009), after the relocation of the suboceanic events with sP depth phases, the observed differences between sP and P arrival times (sP–P) exhibit a better linear relationship with focal depths than that before the relocation (Fig. 4a), and the P- and S-wave traveltime residuals are reduced significantly for the 2227 suboceanic events (Group-3 and Group-4 in Table 1) (Fig. 4b–e). The formal uncertainties of the relocated hypocenters are estimated to be smaller than 3 km for the 2227 events, thanks to the use of sP depth phase data. After relocation, the uncertainty in focal depth of these suboceanic events has been reduced significantly as compared with that of the JMA Unified catalog, especially for the events under the Philippine Sea (Fig. S2). For the events under the East China Sea and the Japan Sea, the relocated hypocenters become shallower than 15 km depth, while the relocated events beneath the Philippine Sea are shallower than 30 km depth (Fig. S2). These results are generally consistent with many previous studies using OBS data or double-difference location method (e.g., Obana et al., 2003, 2004, 2005, 2009; Sakai et al., 2005; Bai et al., 2006). We also conducted extensive tests to assess the effects of earthquake location with P, S and sP depth phase data recorded by the seismic stations on the Japan Islands as compared with those with the OBS data. The test results show that the hypocenters can be best determined with P, S, sP depth phase and OBS data together, but the hypocenters can be also determined quite reliably using P, S and sP depth phase data (Fig. S3). Fig. 5 shows the distribution of ray paths of the P- and S-wave data from the 2227 suboceanic earthquakes relocated with sP depth phases, indicating that the rays crisscross very well in the crust and uppermost mantle under the forearc region beneath the Philippine Sea and the back-arc region beneath the Japan Sea. We conducted many checkerboard resolution tests (CRTs) following the approach of Zhao et al. (1992) to assess the adequacy of the ray coverage and to evaluate the resolution of the tomographic image. To perform a CRT, we first assigned positive and negative velocity perturbations of 3% to all the 3-D grid nodes, then calculated synthetic arrival times for the checkerboard model with the same numbers of seismic stations, events and ray paths as those in the real data set, and then inverted the synthetic data to see whether the assigned velocity anomalies could be recovered or not. To simulate the picking errors contained in the observed data, random errors with a normal distribution having a standard deviation of 0.1 s were added to the synthetic arrival times before the tomographic inversion. Fig. S4 shows the test results for the Vp and Vs structures with different lateral grid intervals at six depths in the crust and upper mantle, which indicate that our tomographic model has a spatial resolution of about 30 km in the horizontal direction and 10–30 km in depth. To determine the optimal values of damping and smoothing parameters, we conducted many tomographic inversions (Fig. S5). The results with the optimal values of damping and smoothing are shown in Figs. 6–10. In the optimal tomographic model, the grid interval is 0.33° in the lateral direction and 10– 30 km in depth. The tomographic results show that the Vp and Vs images are generally similar to each other, and strong lateral heterogeneities exist in the crust and uppermost mantle under the study region. Under the land areas, our present results (Figs. 6– 10) are generally consistent with the previous tomography results

63

(e.g., Wang and Zhao, 2006a,b, 2012; Zhao et al., 2011a). But our present images are much improved in the forearc region beneath the Philippine Sea and the back-arc region under the Japan Sea, thanks to the use of precisely relocated suboceanic events with P, S and sP arrive-time data. To confirm the effect of the subducting Philippine Sea slab, we conducted detailed synthetic tests using the restoring resolution test (RRT) (Zhao et al., 1992). The procedure of the RRT synthetic tests is the same as that of the CRTs. The only difference is in the input model. For RRT, the input model is constructed with the main features of the obtained tomographic results (Figs. 6–10). The subducting Philippine Sea slab was set as a continuous high-velocity (high-V) zone in the input model (Figs. S6 and S7). The test results show that this high-V zone can be generally recovered, but it becomes intermittent and scattered in many places (Figs. S6 and S7). This feature was caused by the imperfect ray path coverage in and around the Philippine Sea slab which may be too thin to be fully recovered due to the limited ray coverage of our data set. We also conducted a tomographic inversion using a homogeneous 1-D starting velocity model that does not include the subducting Philippine Sea slab. The obtained results (Figs. S8 and S9) are similar to that shown in Figs. 8 and 9, except for the subducting Philippine Sea slab. In Figs. S8 and S9, the Philippine Sea slab is generally imaged as a high-V zone with a thickness of 40 km, but the high-V zones are much more intermittent and even disappeared in many places than that in Figs. 8 and 9. According to the above-mentioned previous studies of the study region, the Philippine Sea slab has a thickness of 30–50 km under SW Japan. Fig. S1 shows that introducing the Philippine Sea slab with proper velocity perturbations into the starting velocity model can lead to a significant reduction in the P- and S-wave RMS travel-time residuals, because the theoretical travel times and ray paths can be calculated more accurately when the slab is considered (Zhao et al., 1992, 2012). However, different initial values (20–50 km) of the slab thickness can only cause changes of 0.1–0.3% in the RMS travel-time residuals (Fig. S1b), suggesting that our data set is not sensitive to the thickness of the Philippine Sea slab because few intraslab earthquakes occur under many parts of the study region. Therefore, in the optimal tomographic model (Figs. 6–10), we introduced the Philippine Sea slab as a high-V zone with a thickness of 40 km by considering the results of these detailed analyses (e.g., Figs. S1, S8 and S9). Recently, some of the other tomographic studies have also adopted a starting velocity model including the subducting Philippine Sea slab and so they obtained reasonable tomographic results for understanding the subduction dynamics under the Kyushu subduction zone (e.g., Wang and Zhao, 2006a).

4. Discussion 4.1. Nature of low-velocity anomalies Although low-velocity (low-V) anomalies in the Nankai subduction zone vary from the east to west, they can be divided into four groups distributing in different areas (Figs. 6–9). The low-V zones are not consequences of prescribing a fast slab because they also appear in the tomographic images obtained with a 1-D homogeneous starting velocity model (Figs. S8 and S9). The results of our detailed synthetic tests (Figs. S6 and S7) indicate that these lowV zones are reliable features. The first group of low-V anomalies is located under the volcanic front. Beneath Kyushu Island, this low-V zone generally dips toward the back-arc side, whereas this feature is not clear under southwestern Honshu due to the absence of deep earthquakes there (Figs. 6–9). Recently, a regional electrical-conductivity study revealed a conductor beneath the volcanic front in Kyushu, which

64

X. Liu et al. / Physics of the Earth and Planetary Interiors 216 (2013) 59–73

sP−P (s) 0

0

4

8

12

Depth (km)

2227 events

10 20 30 40

(a)

12 Before relocation

P−wave (b)

Before relocation

S−wave (c)

After relocation

P−wave (d)

After relocation

S−wave (e)

9

6

Frequency (%)

3

0 12

9

6

3

0 −2

−1

0

1

2

−2

P−wave travel−time residual (s)

−1

0

1

2

S−wave travel−time residual (s)

Fig. 4. (a) The relationship between the observed differences between sP and P phases arrival times (sP–P) and focal depths for 2227 suboceanic earthquakes. Grey dots and open circles with the error bars show this relationship before and after earthquake relocation with sP depth phase data. (b and c) Distribution of P- and S-wave travel-time residuals of the 2227 suboceanic earthquakes before the earthquake relocation. (d and e) Distribution of P- and S-wave travel-time residuals of the 2227 suboceanic earthquakes after the earthquake relocation.

also extends to the back-arc side (Hata et al., 2012). The tomographic results from a simultaneous inversion of local and teleseismic data (Zhao et al., 2012) show that the down-dip limit of these low-V anomalies in the mantle wedge is greater than 300 km

depth under the back-arc region. Many previous tomographic studies have also revealed a low-V zone in the mantle wedge under the volcanic front of SW Japan and suggested that the low-V zone represents the hot upwelling flow associated with the slab dehy-

X. Liu et al. / Physics of the Earth and Planetary Interiors 216 (2013) 59–73

65

Fig. 5. Distribution of (A) P- and (B) S-wave ray paths (gray lines) from the 2227 suboceanic earthquakes (white circles) beneath the Philippine Sea, the East China Sea and the Japan Sea. Map view (a and a0 ) and east–west (b and b0 ) and north–south (c and c’) vertical cross sections. The black squares show the seismic stations used.

66

X. Liu et al. / Physics of the Earth and Planetary Interiors 216 (2013) 59–73

Fig. 6. P-wave velocity images at depths of 5, 15, 25, 40, 65 and 90 km under the study area. Red and blue colors denote low and high velocities, respectively. Red stars denote large earthquakes (M P 6.0) that occurred during 416–2011 (Utsu, 1982; Usami, 2003). The black and grey triangles denote active and Quaternary arc volcanoes, respectively. The velocity perturbation scale and the earthquake magnitude scale are shown at the bottom. The sawtooth and dashed lines indicate the Nankai trough and the estimated boundary between the Eurasia and the subducting Philippine Sea plates, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

dration and corner flow in the mantle wedge that forms the source of arc magmatism and volcanism (e.g., Zhao et al., 2000, 2011a; Wang and Zhao, 2006a,b, 2012). The second group of low-V anomalies is visible in the lower crust and uppermost mantle under the back-arc region (Figs. 6– 9). These low-V zones are generally subhorizontal and much stronger in S-wave image than those in P-wave image. In the horizontal

direction, such a low-V zone is mainly linked with the low-V zone under the volcanic front and extends westward under the East China Sea and the Japan Sea. As shown in Figs. 8 and 9, beneath the Unzen volcano in the back-arc area of Kyushu, such a low-V zone is connected with the low-V zone under the volcanic front. Similar features have been revealed by previous tomography (Wang and Zhao, 2006a). However, the back-arc low-V zone could not be

X. Liu et al. / Physics of the Earth and Planetary Interiors 216 (2013) 59–73

67

Fig. 7. The same as Fig. 6, but for S-wave velocity images. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

distinguished from the volcanic-front low-V zone by the previous studies, such as surface-wave tomography (Yoshizawa et al., 2010), ambient-noise tomography (Zheng et al., 2011), and joint inversion of local and teleseismic data (Zhao et al., 2012), perhaps because these large-scale tomographic models have a low resolution, while our present tomography has a much higher resolution. The back-arc low-V zone may reflect a fluid-filled or partial-melting zone associated with convective circulation in the mantle

wedge and dehydration of the Philippine Sea slab which may have led to the formation of the back-arc basin, though they may be also affected by seawater permeating down to the deep crust through many normal faults developed during the back-arc extension (Zhao et al., 2002, 2012; Huang et al., 2010). The third group of low-V anomalies is mainly visible under the forearc region adjacent to the megathrust zone (Figs. 8 and 9), especially off Kyushu and Shikoku Islands, which is different from

68

X. Liu et al. / Physics of the Earth and Planetary Interiors 216 (2013) 59–73

Fig. 8. Vertical cross-sections of P-wave velocity images and topography along the profiles as shown on the insert map. Red color represents low velocity, whereas blue color indicates high velocity. Red stars denote large earthquakes (M P 6.0; 416–2011) within a 10 km width along each profile (Utsu, 1982; Usami, 2003). Note that the focal depths of the historic inland large crustal earthquakes are set at 10 km, and the focal depths of suboceanic earthquakes in the forearc region are set above the upper boundary of the subducting Philippine Sea slab, because the accurate focal depths are unclear for most of them. The velocity perturbation scale and the earthquake magnitude scale are shown at the bottom. The red triangles denote active arc volcanoes within a 20 km width along each profile. The names of volcanoes are labeled above the topography. Big white dots denote the relocated earthquakes used in this study within a 5-km width along each profile. Small white dots denote all the earthquakes occurred during July 2002 to March 2012 within a 5-km width along each profile. The three curved lines show the Conrad and Moho discontinuities and the upper boundary of the subducting Philippine Sea slab, while the dashed lines denote the inferred lower boundary of the Philippine Sea slab. The red triangles on the inset map denote active arc volcanoes. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

the results of Wang and Zhao (2006a) but generally consistent with the tomographic results obtained with the OBS data (Tahara et al.,

2008), probably because we have used much more P, S and sP arrival times collected from the suboceanic earthquakes and the

X. Liu et al. / Physics of the Earth and Planetary Interiors 216 (2013) 59–73

69

Fig. 9. The same as Fig. 8, but for S-wave velocity images. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

updated results of the Conrad and Moho discontinuities and the upper boundary of the Philippine Sea slab. We consider that the forearc low-V zone probably reflects a highly serpentinized and hydrated forearc mantle resulting from the dehydration of the young and warm Philippine Sea slab (Zhao et al., 2000; Hyndman and Peacock, 2003). This serpentinized area may have a low strength and so control the down-dip limit of interplate large earthquakes, as shown in the Figs. 8 and 9. Repeating earthquake activity seems

to share the same down-dip limit under the Hyuga-nada area in the Kyushu forearc (Yamashita et al., 2012). This feature is different from that in Tohoku where the old and cold Pacific slab is subducting beneath the Okhotsk plate (Huang et al., 2011). However, a similar result has been revealed in the central Cascadia subduction zone where the young Juan de Fuca slab is subducting beneath the North American plate (Bostock et al., 2002). In addition, beneath the serpentinized forearc mantle wedge, many earthquakes

70

X. Liu et al. / Physics of the Earth and Planetary Interiors 216 (2013) 59–73

Fig. 10. (a) Vp and (b) Vs tomographic images of the interplate megathrust zone along the upper boundary of the subducting Philippine Sea slab in the Nankai subduction zone. Red and blue colors denote low and high velocities, respectively. Red stars denote large earthquakes (M P 6.0; 1900–2011) that occurred in the forearc region beneath the Philippine Sea (Utsu, 1982; Usami, 2003). The earthquake magnitude scale and the velocity perturbation scale are shown in (a and b), respectively. The areas shown in yellow dashed lines in (a0 and b0 ) denote the inferred subducted oceanic ridges and seamounts. From the west to east, they are the subducted parts of the Kyushu-Palau ridge (Nishizawa et al., 2009), the Kinan seamount chain (Kodaira et al., 2002) and the Izu-Bonin arc (Park et al., 2004). The unsubducted parts of these topographic highs on the seafloor of the Philippine Sea plate are also shown in (a0 and b0 ). The coseismic slips of the large earthquakes in 1944 (blue lines, Kikuchi et al., 2003), 1946 (white lines, Sagiya and Thatcher, 1999), 1968 (red lines, Yagi et al., 1998) and 1996 (red lines, Yagi et al., 1999) are also shown in the (a0 and b0 ). The inner contour lines denote larger coseismic slips (see the above references for details). The solid triangles denote active arc volcanoes. The sawtooth and dashed lines indicate Nankai trough and the depth contours of the upper boundary of the subducting Philippine Sea plate, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

occurred within the subducting Philippine Sea slab (Figs. 8 and 9), which may be related to dehydration reactions that release water from the subducting slab to the forearc mantle wedge (Xia et al., 2008; Zhao et al., 2000, 2011a). The above-mentioned three groups of low-V zones are located above the subducting Philippine Sea slab, whereas the fourth group of low-V anomalies is located beneath the slab (Figs. 8 and 9). Although this subslab low-V zone is only visible in some cross-sections and its outline is also not very clear because of the lack of intermediate-depth and deep earthquakes in our data set, the results of the synthetic test (Figs. S6 and S7) indicate that it is a reliable feature. The results of regional tomographic studies (Huang and Zhao, 2006; Abdelwahed and Zhao, 2007; Zhao et al., 2012) show that this subslab low-V zone exists down to a depth of 500 km under SW Japan and it is associated with the deep dehydration of the Pacific slab, as well as the convective circulation process in the big mantle wedge above the stagnant Pacific slab. 4.2. Interplate megathrust zone and seismotectonics Fig. 10 shows that the Vp and Vs images in the interplate megathrust zone are generally similar to each other, and strong lateral

heterogeneities exist in the megathrust zone beneath the forearc. The results of the synthetic test (Fig. 11) indicate that these are reliable features. The high-V anomalies in the megathrust zone may be formed by the subducted seamounts or other topographic highs on the seafloor of the oceanic plate (Zhao et al., 2011b). There are three noticeable seamount chains on the Philippine Sea plate close to the Nankai trough. From the west to east, they are the Kyushu-Palau ridge, the Kinan seamount chain, and the Izu-Bonin arc (Figs. 1 and 10). The Kyushu-Palau ridge separates the young Shikoku Basin from the older West Philippine Basin (Okino et al., 1999; Deschamps and Lallemand, 2002). Seismic exploration results revealed an irregular Vp structure in the northwestern extension of the Kyushu-Palau ridge beneath the Hyuga-nada, which may be associated with the subducted part of this ridge (Nishizawa et al., 2009). In addition, they also found high-V anomalies north of this irregular Vp structure, where the 1968 Hyuga-nada earthquake (M 7.5) occurred (Fig. 10). Two-dimensional (2-D) seismic reflection results also revealed a subducting seamount under the forearc region off Shikoku Island (Kodaira et al., 2002; Fig. 10). This seamount probably belongs to the Kinan seamount chain which originated from igneous activity after back-arc spreading of the Shikoku Basin (Kobayashi et al., 1995; Okino et al., 1999). The

X. Liu et al. / Physics of the Earth and Planetary Interiors 216 (2013) 59–73

71

Fig. 11. Results of a synthetic resolution test for (a) Vp and (b) Vs tomographic images of the megathrust zone of the Nankai subduction zone along the upper boundary of the subducting Philippine Sea slab. The input models are shown in (a and b), while the inversion results are shown in (a0 and b0 ). The velocity perturbation scale is shown at the bottom. The other labeling is the same as that in Fig. 10.

Izu-Bonin arc marks the boundary between the Shikoku Basin and the Pacific plate. Zenisu ridge is a subbranch of this arc and subparallel to the Nankai trough (Figs. 1 and 10). Seismic reflection and refraction results revealed a subducting ridge subparallel to Zenisu ridge beneath the accretionary wedge in the Nankai subduction zone (Park et al., 2004). The estimated locations of these subducted ridges and seamounts are generally consistent with those of the high-V anomalies in the megathrust zone revealed by this study (Fig. 10). Our present results show that the interplate large earthquakes (M P 6.0) that occurred in the forearc region during 1900–2011 (Utsu, 1982; Usami, 2003; JMA Unified catalog) are mainly located in areas with high-V anomalies, or at the boundary between the high-V and low-V zones (Fig. 10). Wang and Zhao (2006a) also showed that the interplate large earthquakes in the Kyushu forearc region mainly occurred outside the low-V anomalies in the megathrust zone. This feature has been confirmed by Liu et al. (in review) with the updated geometry of the subducting Philippine Sea slab and much more high-quality arrival-time data. Such a relationship was also revealed in the Tohoku forearc region (Huang et al., 2011; Zhao et al., 2011b). We think that most of these high-V anomalies reflect the subducted oceanic ridges and seamounts as mentioned above, and they can form asperities in the megathrust zone, where the average quasi-static slip rate is generally low (e.g., Yamashita et al., 2012), suggesting that the subducting oceanic slab is strongly coupled with the overriding continental plate

at these asperities. In contrast, the low-V anomalies in the megathrust zone may contain subducted sediments and fluids released from the slab dehydration (Zhao et al., 2011b). As shown in Fig. 10, obvious low-V anomalies exist adjacent to the areas (asperities) where many interplate large earthquakes occurred. We consider that the fluids may control the pore fluid pressure and so play an important role in the nucleation of interplate large earthquakes at the asperities. The asperities formed by the subducted oceanic ridges and seamounts may increase the shear stress/strain in the megathrust zone and promote the nucleation of interplate large earthquakes (Cloos, 1992; Scholz and Small, 1997; Baba et al., 2001). However, the areas where megathrust events reoccurred are not always consistent with their high coseismic slips. As shown in Fig. 10, the coseismic slip distributions of the 1944 Tonankai earthquake (M 7.9) (Kikuchi et al., 2003) and the 1946 Nankai earthquake (M 8.0) (Sagiya and Thatcher, 1999) are not limited in the high-V patches (asperities) where the initial ruptures occurred. In contrast to the high-V patches (strong coupled areas) in the megathrust zone, the low-V patches probably represent weakly coupled or even decoupled areas (Zhao et al., 2011b). Three-dimensional seismic reflection results also revealed a low seismic impedance layer (low-V zone) in the megathrust zone southeast off the Kii Peninsula, which represents a weakly coupled area (Bangs et al., 2009). We think that due to the weak interplate coupling in the low-V patches, the ruptures of some megathrust earthquakes could

72

X. Liu et al. / Physics of the Earth and Planetary Interiors 216 (2013) 59–73

unimpededly pass through the low-V anomalies that are adjacent to high-V patches (asperities) where megathrust events reoccurred. The asperities may also act as barriers hindering or even ceasing the rupture processes of some megathrust earthquakes triggered by other asperities (Nakanishi et al., 2002; Park et al., 2004). This scenario may explain the feature that many large interplate earthquakes (M 6–7) occurred beneath the Hyuga-nada, where many high-V pitches (asperities) and a few low-V anomalies exist and so their rupture extents (coseismic slips) are limited (Nishizawa et al., 2009; Fig. 10). In contrast, many connected low-V anomalies exist south off Shikoku and Honshu and so the rupture processes are unimpeded. Therefore, M 8 class interplate earthquakes, such as the 1946 Nankai earthquake (M 8.0), are generated. This hypothesis seems to be also applicable to the Tohoku megathrust zone where the huge earthquake (Mw 9.0) nucleated at an isolated large high-V patch that is surrounded by low-V anomalies, thus the rupture was not limited due to the lack of other strongly-coupled patches (high-V zones) adjacent to the isolated high-V patch (see Fig. 3 in Zhao et al., 2011b).

Acknowledgments We thank the data center of the Kiban seismic network and the JMA unified catalog for providing the high-quality waveform and arrival-time data used in this study. Some of the arrival-time data were measured from the original seismograms by the staffs of Research Center for Prediction of Earthquakes and Volcanic Eruptions, Tohoku University. We are very grateful to Dr. A. Katsumata who kindly provided his data about the Conrad and Moho depth variations beneath the Japan Islands. Most of the figures were made by using GMT (Wessel and Smith, 1998). Miss S. Zhao helped us in computer graphics of Fig. 10. We appreciate the helpful discussions with Drs. Z. Huang, Y. Nishizono and H. Inakura. We are very grateful to Prof. G. Helffrich (editor), and two anonymous reviewers who provided thoughtful review comments on the manuscript, which improved the present work. This work was supported by a grant (Kiban-S 11050123) to D. Zhao from Japan Society for the Promotion of Science, a Chinese S863 key project (2009AA093401), NSFC projects (Grant Nos. 41190072 and 41190070) and a scholarship award for excellent doctoral student to X. Liu granted by Ministry of Education, China.

5. Conclusions To better understand arc magmatism, seismogenesis and dynamics of the Nankai subduction zone, we determined high-resolution Vp and Vs images of the crust and upper mantle of the entire Nankai subduction zone, from the Nankai trough to the Japan Sea, using a large number of high-quality P- and S-wave arrivaltime data from local shallow and intermediate-depth earthquakes. To determine the 3-D seismic structure under the Philippine Sea and the Japan Sea, we used data from many suboceanic earthquakes that are precisely relocated with sP depth phases collected from three-component seismograms recorded by the dense seismic network on the Japan Islands. Our results reveal strong lateral heterogeneities in the crust and upper mantle beneath the entire Nankai subduction zone. The main findings of this work are summarized as follows. (1) The subducting Philippine Sea slab is imaged clearly as highV zones. Four groups of low-V anomalies are revealed in different parts of the Nankai subduction zone. The first group is located under the volcanic front. The second group is visible in the lower crust and uppermost mantle under the back-arc region. The third group is visible beneath the forearc region adjacent to the megathrust zone. The three groups of low-V zones are caused by the slab dehydration and corner flow associated with the subduction of the Philippine Sea slab. The fourth group of low-V zones is located beneath the subducting Philippine Sea slab, which are associated with the deep dehydration of the Pacific slab, as well as the convective circulation process in the big mantle wedge above the Pacific slab. (2) In the Nankai interplate megathrust zone, large interplate earthquakes mainly occurred in or around the high-V patches that may represent asperities (strongly-coupled areas) formed by the subducted oceanic ridges and seamounts. These high-V patches are generally surrounded by significant low-V anomalies which may contain subducted sediments and fluids associated with slab dehydration. (3) The coseismic slip distributions of large megathrust earthquakes seem to be not limited in the high-V patches (asperities) where the ruptures initiated. Because of the weak interplate coupling in the low-V areas, the rupture of an interplate earthquake could unimpededly pass through the low-V anomalies and so lead to a great megathrust earthquake.

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.pepi.2013.01.003. References Abdelwahed, M., Zhao, D., 2007. Deep structure of the Japan subduction zone. Physics of the Earth and Planetary Interiors 162, 32–52. Baba, T., Hori, T., Hirano, S., Cummins, P., Park, J., Kameyama, M., Kaneda, Y., 2001. Deformation of a seamount subducting beneath an accretionary prism: constraints from numerical simulation. Geophysical Research Letters 28, 1827–1830. Bai, L., Kawasaki, I., Zhang, T., Ishikawa, Y., 2006. An improved double-difference earthquake location algorithm using sP phases: application to the foreshock and aftershock sequences of the 2004 earthquake offshore of the Kii peninsula, Japan (Mw = 7.5). Earth Planets Space 58, 823–830. Bangs, N., Moore, G., Gulick, S., Pangborn, E., Tobin, H., Kuramoto, S., Taira, A., 2009. Broad, weak regions of the Nankai Megathrust and implications for shallow coseismic slip. Earth and Planetary Science Letters 284, 44–49. Bostock, M., Hyndman, R., Rondenay, S., Peacock, S., 2002. An inverted continental Moho and serpentinization of the forearc mantle. Nature 417, 536–538. Cloos, M., 1992. Thrust-type subduction-zone earthquakes and seamount asperities: a physical model for seismic rupture. Geology 20, 601–604. Deschamps, A., Lallemand, S., 2002. The West Philippine Basin: an Eocene to early Oligocene back arc basin opened between two opposed subduction zones. Journal of Geophysical Research 107, 2322. Gamage, S., Umino, N., Hasegawa, A., Kirby, S., 2009. Offshore double-planed shallow seismic zone in the NE Japan forearc region revealed by sP depth phases recorded by regional networks. Geophysical Journal International 178, 195– 214. Hall, R., 2002. Cenozoic geological and plate tectonic evolution of SE Asia and the SW Pacific: computer-based reconstructions, model and animations. Journal of Asian Earth Sciences 20, 353–431. Hata, M., Oshiman, N., Yoshimura, R., Tanaka, Y., Uyeshima, M., 2012. Fluid upwelling beneath arc volcanoes above the subducting Philippine Sea Plate: evidence from regional electrical resistivity structure. Journal of Geophysical Research 117, B07203. Huang, J., Zhao, D., 2006. High-resolution mantle tomography of China and surrounding regions. Journal of Geophysical Research 111, B09305. Huang, Z., Zhao, D., Umino, N., Wang, L., Matsuzawa, T., Hasegawa, A., Yoshida, T., 2010. P-wave tomography, anisotropy and seismotectonics in the eastern margin of Japan Sea. Tectonophysics 489, 177–188. Huang, Z., Zhao, D., Wang, L., 2011. Seismic heterogeneity and anisotropy of the Honshu arc from the Japan Trench to the Japan Sea. Geophysical Journal International 184, 1428–1444. Hyndman, R., Peacock, S., 2003. Serpentinization of the forearc mantle. Earth and Planetary Science Letters 876, 417–432. Jeffreys, H., Bullen, K., 1940. Seismological Tables. British Association for the Advancement of Science, London. Katsumata, A., 2010. Depth of the Moho discontinuity beneath the Japanese islands estimated by traveltime analysis. Journal of Geophysical Research 115, B04303. Kikuchi, M., Nakamura, M., Yoshikawa, K., 2003. Source rupture processes of the 1944 Tonankai earthquake and the 1945 Mikawa earthquake derived from lowgain seismograms. Earth, Planets and Space 55, 159–172.

X. Liu et al. / Physics of the Earth and Planetary Interiors 216 (2013) 59–73 Kobayashi, K., Kasuga, S., Okino, K., 1995. Shikoku Basin and its Margins, in Backarc Basins. In: Taylor, B. (Ed.), Tectonics and Magmatism. Plenum Press, New York, pp. 381–405. Kodaira, S., Takahashi, N., Park, J., Mochizuki, K., Shinohara, M., Kimura, S., 2000. Western Nankai Trough seismogenic zone: result from a wide-angle ocean bottom seismic survey. Journal Geophysical Research 105, 5887–5905. Kodaira, S., Kurashimo, E., Park, J., Takahashi, N., Nakanishi, A., Miura, S., Iwasaki, T., Hirata, N., Ito, K., Kaneda, Y., 2002. Structural factors controlling the rupture process of a megathrust earthquake at the Nankai trough seismogenic zone. Geophysical Journal International 149, 815–835. Laske, G., Masters, G., Reif, C., 2003. CRUST2.0: a new global crustal model at 2  2 degrees. . Liu, X., Zhao, D., Li, S., Huang, Z., Nishizono, Y., Inakura, H., in review. Seismic imaging of the entire Kyushu subduction zone: insight into forearc interplate coupling and back-arc magmatism and dynamics. Tectonophysics. Nakanishi, A., Kodaira, S., Park, J., Kaneda, Y., 2002. Deformable backstop as seaward end of coseismic slip in the Nankai Trough seismogenic zone. Earth and Planetary Science Letters 203, 255–263. Nishizawa, A., Kaneda, K., Oikawa, M., 2009. Seismic structure of the northern end of the Ryukyu Trench subduction zone, southeast of Kyushu, Japan. Earth Planets Space 61, e37–e40. Obana, K., Kodaira, S., Kaneda, Y., Mochizuki, K., Shinohara, M., Suyehiro, K., 2003. Microseismicity at the seaward updip limit of the western Nankai Trough seismogenic zone. Journal of Geophysical Research 108, 2459. http://dx.doi.org/ 10.1029/2002JB002370. Obana, K., Kodaira, S., Kaneda, Y., 2004. Microseismicity around rupture area of the 1944 Tonankai earthquake from ocean bottom seismograph observations. Earth and Planetary Science Letters 222, 561–572. Obana, K., Kodaira, S., Kaneda, Y., 2005. Seismicity in the incoming/subducting Philippine Sea plate off the Kii Peninsula, central Nankai trough. Journal of Geophysical Research 110, B11311. Obana, K., Kodaira, S., Kaneda, Y., 2009. Seismicity at the eastern end of the 1944 Tonankai earthquake rupture area. Bulletin of the Seismological Society of America 99, 110–122. Okada, Y., Kasahara, K., Hori, S., Obara, K., Sekiguchi, S., Fujiwara, H., Yamamoto, A., 2004. Recent progress of seismic observation networks in Japan: Hi-net, F-net, K-net and KiK-net. Earth Planets Space 56, 15–28. Okino, K., Ohara, Y., Kasuga, S., Kato, Y., 1999. The Philippine Sea: new survey results reveal the structure and the history of the marginal basins. Geophysical Research Letters 26, 2287–2290. Park, J., Moore, G., Tsuru, T., Kodaira, S., Kaneda, Y., 2004. A subducted oceanic ridge influencing the Nankai megathrust earthquake rupture. Earth and Planetary Science Letters 217, 77–84. Sagiya, T., Thatcher, W., 1999. Coseismic slip resolution along a plate boundary megathrust: the Nankai Trough, southwest Japan. Journal of Geophysical Research 104, 1111–1129. Sakai, S., Yamada, T., Shinohara, M., Hagiwara, H., Kanazawa, T., Obana, K., Kodaira, S., Kaneda, Y., 2005. Urgent aftershock observation of the 2004 off the Kii Peninsula earthquake using ocean bottom seismometers. Earth Planets Space 57, 363–368. Scholz, C., Small, C., 1997. The effect of seamount subduction on seismic coupling. Geology 25, 487–490. Sdrolias, M., Roest, W., Müller, R., 2004. An expression of Philippine Sea plate rotation: the Parece Vela and Shikoku Basins. Tectonophysics 394, 69–86. Tahara, M., Shimizu, H., Nakada, M., Ito, Y., 2006. Focal depth distribution using sP depth phase and implications for plate coupling in the Hyuganada region, Japan. Physics of the Earth and Planetary Interiors 155, 219–235. Tahara, M., Uehira, K., Shimizu, H., Nakada, M., Yamada, T., Mochizuki, K., Shinohara, M., Nishino, M., Hino, R., Yakiwara, H., Miyamachi, H., Umakoshi, K., Goda, M., Matsuwo, N., Kanazawa, T., 2008. Seismic velocity structure around the Hyuganada region, Southwest Japan, derived from seismic tomography using land and OBS data and its implications for interplate coupling and vertical crustal uplift. Physics of the Earth and Planetary Interiors 167, 19–33.

73

Umino, N., Hasegawa, A., 1994. Aftershock focal depths of the 1993 HokkaidoNansei-Oki earthquake estimated from sP depth phase at small epicentral distances. Journal of Physics of the Earth 42, 321–329. Umino, N., Hasegawa, A., Matsuzawa, T., 1995. SP depth phase at small epicentral distances and estimated subducting plate boundary. Geophysical Journal International 120, 356–366. Usami, T., 2003. Materials for Comprehensive List of Destructive Earthquakes in Japan. University of Tokyo Press, Tokyo. Utsu, T., 1982. Catalog of large earthquakes in the region of Japan from 1885 through 1980. Bulletin of the Earthquake Research Institute, University of Tokyo 57, 401–463. Wang, J., Zhao, D., 2012. P wave anisotropic tomography of the Nankai subduction zone in Southwest Japan. Geochemistry Geophysics Geosystems 13, Q05017. Wang, Z., Zhao, D., 2005. Seismic imaging of the entire arc of Tohoku and Hokkaido in Japan using P-wave, S-wave and sP depth-phase data. Physics of the Earth and Planetary Interiors 152, 144–162. Wang, Z., Zhao, D., 2006a. Vp and Vs tomography of Kyushu, Japan: new insight into arc magmatism and forearc seismotectonics. Physics of the Earth and Planetary Interiors 157, 269–285. Wang, Z., Zhao, D., 2006b. Seismic evidence for the influence of fluids on the 2005 west off Fukuoka prefecture earthquake in southwest Japan. Physics of the Earth and Planetary Interiors 155, 313–324. Wessel, P., Smith, W., 1998. New, improved version of the Generic Mapping Tools released. EOS Transactions American Geophysical Union 79, 579. Xia, S., Zhao, D., Qiu, X., 2008. Tomographic evidence for the subducting oceanic crust and forearc mantle serpentinization under Kyushu, Japan. Tectonophysics 449, 85–96. Yagi, Y., Kikuchi, M., Yoshida, S., Yamanaka, Y., 1998. Source Process of the Hyuganada earthquake of April 1, 1968 (Mjma 7.5), and relationship to the subsequent seismicity. Zisin 51, 139–148 (in Japanese with English abstract). Yagi, Y., Kikuchi, M., Yoshida, S., Sagiya, T., 1999. Comparison of the coseismic rupture with the aftershock distribution in the Hyuga-nada earthquakes of 1996. Geophysical Research Letters 26, 3161–3164. Yamashita, Y., Shimizu, H., Goto, K., 2012. Small repeating earthquake activity, interplate quasi-static slip, and interplate coupling in the Hyuga-nada, southwestern Japan subduction zone. Geophysical Research Letters 39, L08304. Yoshizawa, K., Miyake, K., Yomogida, K., 2010. 3D upper mantle structure beneath Japan and its surrounding region from inter-station dispersion measurements of surface waves. Physics of the Earth and Planetary Interiors 183, 4–19. Zhao, D., Hasegawa, A., Horiuchi, S., 1992. Tomographic imaging of P and S wave velocity structure beneath northeastern Japan. Journal of Geophysical Research 97, 19909–19928. Zhao, D., Asamori, K., Iwamori, H., 2000. Seismic structure and magmatism of the Young Kyushu Subduction Zone. Geophysical Research Letters 27, 2057–2060. Zhao, D., Mishra, O., Sanda, R., 2002. Influence of fluids and magma on earthquakes: seismological evidence. Physics of the Earth and Planetary Interiors 132, 249– 267. Zhao, D., Wang, Z., Umino, N., Hasegawa, A., 2007. Tomographic imaging outside a seismic network: application to the Northeast Japan arc. Bulletin of the Seismological Society of America 97, 1121–1132. Zhao, D., Wei, W., Nishizono, Y., Inakura, H., 2011a. Low-frequency earthquakes and tomography in western Japan: insight into fluid and magmatic activity. Journal of Asian Earth Sciences 42, 1381–1393. Zhao, D., Huang, Z., Umino, N., Hasegawa, A., Kanamori, H., 2011b. Structural heterogeneity in the megathrust zone and mechanism of the 2011 Tohoku-oki earthquake (Mw 9.0). Geophysical Research Letters 38, L17308. Zhao, D., Yanada, T., Hasegawa, A., Umino, N., Wei, W., 2012. Imaging the subducting slabs and mantle upwelling under the Japan Islands. Geophysical Journal International 190, 816–828. Zheng, Y., Shen, W., Zhou, L., Yang, Y., Xie, Z., Ritzwoller, M., 2011. Crust and uppermost mantle beneath the North China Craton, northeastern China, and the Sea of Japan from ambient noise tomography. Journal of Geophysical Research 116, B12312.