Author's personal copy. Available online at

Transcription

1 Available online at Journal of Asian Earth Sciences 31 (2008) Imaging of V p, V s, and Poisson s ratio anomalies beneath Kyushu, southwest Japan: Implications for volcanism and forearc mantle wedge serpentinization Mohamed K. Salah *, Tetsuzo Seno Earthquake Research Institute, University of Tokyo, Tokyo , Japan Received 16 November 2006; received in revised form 10 July 2007; accepted 23 July 2007 Abstract We determine detailed 3-D V p and V s structures of the crust and uppermost mantle beneath the Kyushu Island, southwest Japan, using a large number of arrival times from local earthquakes. From the obtained V p and V s models, we further calculate Poisson s ratio images beneath the study area. By using this large data set, we successfully image the 3-D seismic velocity and Poisson s ratio structures beneath Kyushu down to a depth of 150 km with a more reliable spatial resolution than previous studies. Our results show very clear low V p and low V s anomalies in the crust and uppermost mantle beneath the northern volcanoes, such as Abu, Kujyu and Unzen. Lowvelocity anomalies are seen in the mantle beneath most other volcanoes. In contrast, there are no significant low-velocity anomalies in the crust or in the upper mantle between Aso and Kirishima. The subducting Philippine Sea slab is imaged generally as a high-velocity anomaly down to a depth of 150 km with some patches of normal to low seismic wave velocities. The Poisson s ratio is almost normal beneath most volcanoes. The crustal seismicity is distributed in both the high- and low-velocity zones, but most distinctly in the low Poisson s ratio zone. A high Poisson s ratio region is found in the forearc crustal wedge above the slab in the junction area with Shikoku and Honshu; this high Poisson s ratio could be caused by fluid-filled cracks induced by dehydration from the Philippine Sea slab. The Poisson s ratio is normal to low in the forearc mantle in middle-south Kyushu. This is consistent with the absence of low-frequency tremors, and may indicate that dehydration from the subducting crust is not vigorous in this region. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: Kyushu; Southwest Japan; Shikoku; Philippine Sea; Seismic tomography; Poisson s ratio; Serpentinized forearc mantle; Low-frequency tremor; Volcano 1. Introduction At present, beneath the Japanese Islands, the Philippine Sea plate (PHS) is subducting beneath the Eurasian plate along the Nankai Trough and the Ryukyu Trench in a N50 W direction at a rate of 4 5 cm/yr, while the Pacific plate is subducting from the east beneath the PHS and the Okhotsk plate at a rate of 6 8 cm/yr (Fig. 1, Seno et al., 1993, 1996). Beneath Kyushu, the subducting PHS forms intraslab seismic activity down to a depth of about * Corresponding author. Present address: Geology Department, Faculty of Science, Tanta University, Tanta 31527, Egypt. address: nada6899@yahoo.com (M.K. Salah). 200 km (Fig. 2). Associated with this subduction, active volcanoes form a volcanic front in the central part of the island (Figs. 1 and 2). Quaternary volcanoes also exist along the coast of the Sea of Japan in Chugoku (Yokoyama et al., 1987). It was first believed that these Quaternary volcanoes are also associated with the subduction of the PHS, and that they form a continuous volcanic front from Kyushu (Sugimura, 1960). Later, petrological and geochemical studies indicated that the volcanoes in Chugoku and those in NW. Kyushu, including Unzen, have been formed from more fertile magma sources deep in the mantle (e.g., Nakamura et al., 1989; Iwamori, 1991; Nakada and Kamata, 1991; Sumino et al., 2000). On the other hand, Yoshida and Seno (1992) argued that the Aso /$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi: /j.jseaes

2 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Fig. 1. Distribution of active and Quaternary volcanoes on the Japan Islands. Curved lines show the trenches, which represent the major plate boundaries around the Japanese region. Black rectangle shows the present study area. Arrows and attached numerals show direction and rate of movement (mm/yr) of the Pacific (PA), and Philippine Sea (PHS) slabs. EU, Eurasian plate; OK, Okhotsk plate. volcano is distinct from other volcanoes at the volcanic front, such as Kirishima, because the seismic slab does not reach the depth beneath it, and, thus, might not be related with the PHS subduction, along with other volcanoes in north (N.) Kyushu Chugoku, such as Unzen, Kujyu, and Abu. Seismic tomography results provide useful information about the production, transfer, and distribution of fluids and/or melts in subduction zones in addition to other details of the subduction process. Thus, many authors have studied both the velocity and/or attenuation (Q) structures beneath Japan (e.g., Hirahara, 1981; Zhao et al., 1992a, 2000a,b; Tsumura et al., 2000; Nakajima et al., 2001a; Hasegawa et al., 2005). All these studies confirmed the existence of high-velocity/low-attenuation subducting slabs and low-velocity/high-attenuation anomalies landward, indicating an ascending flow from the deep mantle in the backarc side to the crust beneath active volcanoes. However, most of these studies were concentrated in the northeastern (NE) Japan subduction zone or in wide areas. Therefore, seismic tomography studies in Kyushu would provide useful information on the subduction process of the PHS, and on seismicity and volcanism. Previous seismic tomography studies, revealing mostly large-scale 3-D seismic structures, had a low resolution beneath Kyushu (Hirahara, 1981; Zhao et al., 1994, 2000a; Sadeghi et al., 2000; Honda and Nakanishi, 2002; Nakamura et al., 2002). Zhao et al. (2000a) showed a P- Fig. 2. Distribution of the 4224 events used in this study shown as circles. Event symbols vary in color according to focal depth and in size according to magnitude. Open and solid triangles denote Quaternary and active volcanoes, respectively. wave velocity structure in Kyushu, using earthquake data of the Japan University Network Earthquake Catalogue (JUNEC) during They showed low P-wave velocity anomalies in the mantle beneath the Kujyu volcano both in the backarc and forearc sides. The number of events used is, however, only 486, recorded by few seismic stations, which is barely adequate to reveal the detailed structure. Sadeghi et al. (2000) analyzed P-wave arrivals covering a wide area from southwest (SW.) Japan to eastern China. They found no significant low-velocity anomalies just beneath the volcanic front in Kyushu, and found them rather in the upper mantle west (W.) of Kyushu. The resolution of these tomographic studies, however, seems too low to satisfactorily discuss the volcanism in Kyushu. Recently, high Poisson s ratio regions have been found in the forearc mantle in Kanto SW. Japan (Kamiya and Kobayashi, 2000, 2006; Honda and Nakanishi, 2003). These high Poisson s ratio regions have been interpreted as serpentinized mantle hydrated by water released

3 406 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) from the subducting PHS slab (Kamiya and Kobayashi, 2000; Seno, 2005). On the other hand, low-frequency tremors have been found to occur just above the slab upper surface at km depths in the SW. Japan forearc (Obara, 2002; see also Katsumata and Kamaya, 2003). These lowfrequency tremors are also believed to have been caused by dehydration of the PHS slab (Obara, 2002; Seno and Yamasaki, 2003). Seno and Yamasaki (2003) pointed out that the regions lacking the low-frequency tremors, east (E.) Shikoku and middle (M.)-south (S.) Kyushu in SW. Japan, and Kanto N. Izu in central Japan, might be the places where dehydration from the subducting PHS crust does not occur. If this is true, it is expected that serpentinization of the forearc mantle might not be conspicuous in these regions. Consistently, the results of Honda and Nakanishi (2003) show that the Poisson s ratio is low in E. Shikoku and modest in M.-S. Kyushu. The extent of the dehydration from the subducting crust may also be inferred from the occurrence of intraslab earthquakes in the crustal part (Seno and Yamasaki, 2003), if the dehydration embrittlement is accepted as a mechanism for intraslab earthquakes (Kirby et al., 1996; Seno and Yamanaka, 1996; Peacock and Wang, 1999; Yamasaki and Seno, 2003). Recently, Okamoto et al. (2005) examined whether intraslab earthquakes in Kyushu are occurring in the crust or in the mantle by examining later phases traveling within the crust. They found that in M. Kyushu, there are several intracrustal earthquakes, and in contrast, in S. Kyushu, there are no crustal earthquakes and all events are occurring in the peridotite part of the slab. Therefore the presence of fluids in the forearc crust-mantle related to the occurrence or non-occurrence of dehydration from the subducting PHS slab should be further investigated in this region. In this study, we select a large number of arrival time data generated by local earthquakes recorded by dense seismic networks, and use them to determine detailed seismic velocity and Poisson s ratio structures beneath Kyushu and the junction area with Honshu and Shikoku, SW. Japan. The obtained images down to a depth of 150 km, with their reliable resolution enable us to understand the magma generation processes, shallow crustal seismicity, and serpentinization state in the forearc mantle wedge, which are associated with the subduction of the PHS slab. 2. Data and method 2.1. Data We use P- and S-wave arrival time data from shallow and intermediate-depth earthquakes that occurred in and around Kyushu Island, SW. Japan, between latitudes N, and longitudes E (Fig. 1) in the depth range of km. We have carefully selected 4224 local earthquakes recorded by the High Sensitivity (Hi-net) and the JUNEC seismic networks. A total of Fig. 3. Distribution of Hi-net (solid squares), and JUNEC (solid triangles) seismic stations used in the present study events were selected from JUNEC data from the period from January, 1990 to December, 1998, and 512 events were selected from Hi-net data from the period from December, 1999 to October, 2000 (Fig. 2). The JUNEC network has 46 seismic stations; while the Hinet has 106 seismic stations falling within the study area (Fig. 3). The JUNEC stations are operated by eight national universities in Japan and are equipped with short-period and broadband seismographs (Tsuboi et al., 1989). The data of this seismic network are collected and published by the Earthquake Research Institute (ERI), University of Tokyo. The Hi-net seismograph network consists of more than 600 seismic stations covering the whole of Japan (Obara, 2002), and are operated by the National Research Institute for Earth Sciences and Disaster Prevention (NIED). Combining these two different data sets together significantly improves the station density (Fig. 3) and the seismicity coverage compared with many previous studies. We try to select events that are very well located and uniformly distributed in the study area. For shallow earthquakes (depth 6 20 km), the focal depth error does not exceed 2.5 km and all events are recorded by more than 10 seismic stations. For deeper events (20 km <

4 depth km), we select those whose maximum focal depth error does not exceed 4 km and are recorded by at least 8 seismic stations. This is because we need a larger number of deeper events to accurately image the upper mantle velocity structure. The shallow earthquakes are uniformly distributed in the study area; while the deeper events are associated with the subduction of the PHS slab. However, we notice that there is a cluster of shallow seismicity in W. Kyushu around Unzen volcano (Fig. 2). The selected events generate a total number of P- and S-wave arrival times of 71,100 and 44,990 respectively. The uncertainty of P-wave first arrival times is estimated to be about 0.1 s; while the uncertainty for S-wave first arrivals is slightly greater, being about s. To check the horizontal ray path coverage for P- and S-wave data sets, we plot each path between an epicenter and a station as one straight line in Fig. 4. It is clear that both P- and S-wave data have generally similar and comparable ray coverage in most parts of the study area Method M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) The seismic tomography method was first developed by Aki and Lee (1976). Subsequently, many researchers have improved this technique and applied it to various regions (e.g., Hirahara, 1977; Thurber, 1983; Spakman and Nolet, 1988; Zhou and Clayton, 1990; Zhao et al., 1992a, 1994; Nakajima et al., 2001a; Mishra et al., 2003). In this study, we use the tomographic method of Zhao et al. (1992a). This method is preferred for subduction zone regions, where many seismic discontinuities, such as the Conrad, Moho, or the subducting slab boundary, exist. It uses an efficient 3-D ray tracing scheme to compute travel times and ray paths. We adopt a grid spacing of 0.2 horizontally and km in depth (Fig. 5). In our tomographic inversion, we take two discontinuities (the Conrad and the Moho) into account in the velocity model. It is well known that these discontinuities are not simple flat planes, but have complicated geometries (Horiuchi et al., 1982a,b; Hasegawa et al., 1983; Zhao et al., 1992b; Nakajima et al., 2002; Salah and Zhao, 2004). For this reason, it is necessary to consider these discontinuities in the initial velocity model. The Conrad and Moho depths determined by Zhao et al. (1992b) are adopted in the present inversion. Table 1 shows the initial velocity model used in the tomographic inversion. It is generally similar to the models used previously by other researchers in SW. Japan (e.g., Zhao and Negishi, 1998; Zhao et al., 2000a; Salah and Zhao, 2003a,b, 2004; Salah et al., 2005). Velocities at grid nodes are taken as unknown parameters and the velocity at any point in the model is calculated by linearly interpolating the velocities at the eight grid nodes surrounding that point. For more details about the method, see Zhao et al. (1992a, 1994). The interpretation of tomographic images is usually non-unique, because any given velocity anomaly can be Fig. 4. Ray path coverage for P-wave data (a) and S-wave data (b). Every path between an epicenter and a station is drawn as one straight line.

5 408 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Fig. 5. Configuration of the grid net adopted for the present study in horizontal directions (a), and in depth direction (b). Straight lines in (a) show the locations of cross sections AA 0,BB 0,CC 0,DD 0,EE 0, and FF 0 shown in Figs Numbers (1) (10) represent the names of volcanoes as follow: (1) Abu; (2) Tsurumi; (3) Kujyu; (4) Aso; (5) Unzen; (6) Kirishima; (7) Sakura-jima; (8) Kaimon; (9) Kikai; (10) Kuchinoerabu-jima. Table 1 Initial velocity model adopted for the present study Layer No. Depth (km) V p (km/s) V s (km/s) attributed to either a thermal or a chemical variation. For this reason, it is useful to consider some other physical parameters for a more reliable interpretation. Compared to the seismic velocities themselves, the Poisson s ratio (or V p /V s ratio) is a better indicator for the content of fluids and/or magma (Zhao and Negishi, 1998; Kayal et al., 2002; Takei, 2002; Salah and Zhao, 2003a; Nakajima et al., 2005; Sadeghi et al., 2006) or serpentinization (Christensen, 1996; Kamiya and Kobayashi, 2000). Therefore, after V p and V s models are calculated from travel time inversions, we compute the Poisson s ratio (r) from V p /V s. To ensure having reliable Poisson s ratio anomalies, we consider results only in areas having wellconstrained V p and V s structures with nearly equal resolution (Widiyantoro et al., 1999; Gorbatov and Kennett, 2003).

6 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Resolution and results 3.1. Resolution Before describing main features of the results, we show the results of the checkerboard resolution test (CRT) (Inoue et al., 1990; Zhao et al., 1992a, 1994) to see the reliability of the obtained tomographic images both in map views and in cross sections. The values of assumed checkerboard-type perturbations assigned to grid nodes are ±3%. Synthetic arrival times are then calculated for the checkerboard model. Numbers of stations and events with their exact locations in the synthetic data are taken as the same as those in the real data Fig. 6. Results of a checkerboard resolution test (CRT) of P-wave velocity (see text for details) for seven representative depth layers. The grid separation is 20 km. Solid and open circles denote high- and low-velocities, respectively. The depth of each layer is shown at the lower right. The perturbation scale is shown at the bottom.

7 410 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Fig. 6 (continued ) set. The Gaussian noise of 0.1 s is assigned to the arrival times. The CRT results for the P- and S-wave velocity structures at some representative depth slices are shown in Figs. 6 and 7, respectively. It is clear that the resolution is good where many ray paths are included. At the crustal and uppermost mantle layers, most parts of the study area have a reliable resolution, as most of the input synthetic anomalies are very well recovered, except for the northwestern part of the area at 40 and 55 km depths. For deeper layers (depths 90, 120, and 150 km), the resolution is good where the intermediate-depth events that occur in the subducting PHS slab are located (Figs. 6 and 7). However, some parts of the backarc side and sea-side portions, especially in the deep layers have no resolution because of the insufficient lengths and directions of the rays traveling in these layers. Figs. 8 and 9 show the results of the CRT along six cross sections (see Fig. 5 for the location of the cross

8 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) sections) for which P- and S-wave velocity, and Poisson s ratio structures are shown. A close inspection of these results reveals also that most of the obtained anomalies have a good resolution except for the deep zones in the backarc side and the shallow and deep zones in the forearc. In general, both P- and S-wave velocity structures have almost identical spatial resolutions. We plot the P- and S-wave velocity structures both in the map views and in the cross sections only where more than 25% of the synthetic anomalies are recovered. We then plot the Poisson s ratio only in areas having a same resolution for both P- and S-waves. Therefore, we believe that the estimated Poisson s ratio anomalies are reliable features and their estimation errors are on the same level of P- and S-wave velocity perturbations. Fig. 7. The same as Fig. 6, but for S-wave velocity.

9 412 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Fig. 7 (continued ) 3.2. Results The final inversion results were obtained after three iterations. The P- and S-wave root-mean-square (RMS) travel time residuals calculated after the three iterations were and s, respectively. This is equivalent to a larger than 30% reduction from the initial residuals. In the following paragraphs, we describe the results of V p, V s, and Poisson s ratio. Figs. 10 and 11 show velocity perturbations of the P- and S-waves at depths of 5, 15, 40, 55, 90, 120, and 150 km, respectively, together with the distribution of seismicity within a km thick slice around the studied depth. Fig. 12 shows perturbations of the Poisson s ratio at the same depth slices. In the upper crust (5 and 15 km depths), the velocity structure is generally heterogeneous having strong lateral variations amounting to ±5%. Prominent low-velocity

10 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Fig. 8. Results of CRT of P-wave velocity structures along the six cross sections shown in Fig. 5. Solid and open circles denote high- and low-velocities, respectively. The perturbation scale is shown at the bottom. anomalies are detected beneath the Abu, Unzen, Tsurumi, and Kujyu volcanoes, but not beneath the Aso, Kirishima, and Sakura-jima volcanoes (see Fig. 5 for the location of the volcanoes). The Poisson s ratio is high in some parts of the forearc and near some volcanoes but low to moderate in other areas. Shallow crustal seismic events occur in both low- and high-velocity zones. However, their frequency seems to decrease very near and/or directly below the volcanic areas. In contrast, the shallow seismicity tends to occur in the low Poisson s ratio region. At depths of 40 and 55 km in the uppermost mantle, seismicity is concentrated in the subducting PHS slab along the southeastern parts of the study area. A low-velocity region is seen beneath the forearc mantle at a depth of 55 km in the junction of Shikoku and Honshu and at a depth of 40 km in the forearc of M.-S. Kyushu. Moderate to high Poisson s ratio anomalies are mapped near volcanic areas at a depth of 40 km. A high Poisson s ratio zone is also detected in the junction of Shikoku and Honshu at a depth of 40 km (Fig. 12). From the results in cross sections shown later, we see that this high Poisson s ratio is mostly in the crustal part of the wedge above the slab. We will discuss the origin of this high Poisson s ratio anomaly in a later section. The subducting PHS slab is clearly seen as a 4% higher-than average velocity zone at depths of 90, 120, and 150 km. These high-velocity anomalies reach the location beneath the volcanic front at depths of 90 and 120 km in the south, but not in the north. This is concordant with the conclusion of Yoshida and Seno (1992) based on the slab seismicity. Intermediate-depth seismicity occurs within these highvelocity anomalies. The subducting PHS slab has mostly normal to high Poisson s ratio (Fig. 12). Line AA 0 (Fig. 13) passes through the Quaternary volcano (Abu), and lines CC 0,DD 0,EE 0, and FF 0 (Figs ) run through or near four active volcanoes (Kujyu, Aso, Unzen, and Kirishima, respectively). Line BB 0, however, crosses between volcanic areas (Fig. 14). A line through Sakura-jima is not shown because the resolution in this region is low. Low-velocity zones in the crust are seen beneath active volcanoes along lines AA 0,CC 0, and EE 0. No prominent low-velocity zone is seen in the crust beneath DD 0 and FF 0 (Aso and Kirishima volcanoes, respectively). The Poisson s ratio is slightly above the average beneath these volcanoes. In lines BB 0 and EE 0, volcanoes are absent at the location of the volcanic front and there is no low-velocity zone in the crust, but there is a slight indication of low-velocities in the uppermost mantle.

11 414 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Fig. 9. The same as Fig. 8, but for S-wave velocity. The PHS slab has generally normal to higher than average velocity at most depths. However, in cross sections CC 0 and FF 0, some parts of it have moderate- to low-velocities. This unusual feature has also been detected by other studies (e.g., Yamane et al., 2000; Zhao et al., 2000a,b; Seno et al., 2001; Honda and Nakanishi, 2003). The forearc wedge above the subducting slab has a high Poisson s ratio beneath lines BB 0 and DD 0 in the junction area (Figs. 14 and 16). It can be seen that this high Poisson s ratio region is mostly in the crustal part of the wedge (depth <30 km). The forearc mantle in other sections has a normal to low Poisson s ratio. 4. Discussion 4.1. Subduction of the PHS slab The angle of the subducting PHS slab beneath Kyushu changes from the Shikoku Basin to the Kyushu-Palau Ridge around 32 N (Fig. 1). The Shikoku Basin was formed in the Miocene (Okino et al., 1994) and the Kyushu-Palau Ridge dates back to the Eocene (Seno, 1988). The seismic PHS slab subducts with a small angle (30 ) to a depth of about km beneath Shikoku, and with a large angle (more than 60 ) to depths of about km beneath Kyushu (Fig. 2, see also Uyehira et al., 2001; Honda and Nakanishi, 2003), which may reflect the above age difference. We detect the high-velocity anomalies indicating the PHS slab, down to a depth of about 150 km beneath Kyushu, with a reasonable resolution. Some patches of low-velocity anomalies are seen within the seismic PHS slab, and this may partly be due to low resolution or real heterogeneous structures within the slab itself. There is no significant distinction in the magnitude of the high-velocity between the junction area (lines AA 0 and BB 0 ) and Kyushu (lines CC 0 FF 0 ) Seismological evidence on magmatism in Kyushu The low-velocity anomalies beneath some volcanoes amounting to 5% cannot be explained only by high temperatures (e.g., Nakajima et al., 2005). For this reason, in northern Honshu, such anomalies have been explained by the existence of melt-filled pores and cracks (Nakajima et al., 2001a,b, 2005). The similar magnitude of the lowvelocity anomalies in the crust and upper mantle associated with the northern volcanoes, Abu, Kujyu, and Unzen (Figs. 13, 15, and 17) suggests that melting beneath these volcanoes occurs similarly. The high-velocity seismic slab, however, does not reach the depth below these volcanoes.

12 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Because the melting of the wedge is related to the flow in the mantle and the porous flow of the aqueous fluids induced by the subduction (e.g., Iwamori and Zhao, 2000; Hasegawa and Nakajima, 2004), the absence of the slab directly beneath the volcanoes does not preclude a possibility of the volcanism to be induced by the PHS subduction. The low-velocity anomalies in the mantle beneath some volcanoes inclined landward above the slab would provide a more useful discriminator for the slab-induced volcanism (Hasegawa and Nakajima, 2004). The lowvelocity anomaly beneath Kujyu is inclined landward (line CC 0, Fig. 15); almost parallel to the subducting PHS slab. On the other hand, the low-velocity anomaly in the forearc mantle wedge pointed out by Zhao et al. (2000a) is not seen Fig. 10. P-wave velocity structures (in %) at seven depth slices. Red and blue colors indicate low- and high-velocities, respectively. Open and solid triangles show Quaternary and active volcanoes, respectively. The depth to each layer is shown at the lower right. Numbers between brackets show the depth range of seismicity (crosses) plotted along with the velocity image. The color scale at the bottom varies from 5% to 5% at depths of 5 and 40 km; while it varies from 4% to 4% at depths of 55, 90, 120, and 150 km. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

13 416 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Fig. 10 (continued ) in this study. The low-velocity anomalies associated with Abu and Unzen are not inclined landward. Although these volcanoes have been argued to have a plume origin from the deep mantle (Iwamori, 1991; Nakada and Kamata, 1991; Sadeghi et al., 2000; see also references in Seno, 1999), it is difficult to discriminate their origin from the results in this study, because of the lack of the resolution in the backarc side and deeper portion of the study area. Beneath Aso, Kirishima and Sakura-jima (Figs. 10, 11, 16, and 18), and the area where volcanoes are absent between Aso and Kirishima, the crust shows high-velocities while the upper mantle has an indication of low- to moderate-velocities. Previously, Sadeghi et al. (2000) pointed out that there is no low-velocity anomaly beneath the volcanic front in Kyushu. This does not seem correct except for limited areas.

14 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Dehydration of the subducting slab and serpentinized forearc mantle The serpentinization of the forearc mantle would have important implications for tectonics of arcs and proto-arcs (Seno, 2005; Kirby et al., 2006). There are four ways to infer the existence of the serpentinized mantle associated with the subduction of the altered basaltic crust of the subducting slab (Seno, 2005); they are the downdip limit of interplate thrust earthquakes, intraslab earthquakes within the crust, low-frequency tremors in the forearc, and seismic tomography. In Kyushu, interplate earthquakes are mostly occurring at the contact of the oceanic plate with the crust of the upper plate (Yagi and Kikuchi, 2003). Because the temperature structure of the plate interface is unknown in this region, the usefulness of the downdip limit of interplate earthquakes to infer the serpentinization (Hyndman Fig. 11. The same as Fig. 10, but for S-wave velocity structures.

15 418 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Fig. 11 (continued ) et al., 1997; Hyndman and Peacock, 2003; Seno, 2005) is limited in this region. The low-frequency tremors along the SW. Japan forearc (Obara, 2002) are not seen in Kanto, N. Izu, E. Shikoku, and M.-S. Kyushu. Seno and Yamasaki (2003) inferred from this that the dehydration from the crust of the subducting PHS is not occurring in these places, due to the subduction of non-typical oceanic crust or island-arc crust. The serpentinization of the forearc mantle was expected to be weak in these regions, although a limited high Poisson s ratio region was found in the mantle wedge beneath Kanto around depths of km (Kamiya and Kobayashi, 2000; Matsubara et al., 2005). In other regions without low-frequency tremors, the Poisson s ratio in the forearc mantle is low in E. Shikoku or modest in M.-S. Kyushu (Honda and Nakanishi, 2003).

16 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Our tomography results show that the Poisson s ratio is high in the forearc crustal wedge in the northeastern part of the study area at the junction of Shikoku and Honshu (Figs. 14 and 16). Low-frequency tremors were detected in this area (Obara, 2002). The high Poisson s ratio region in the crustal wedge may represent the effects of fluid-filled cracks induced by the dehydration of the PHS slab (see Takei, 2002 for such effects). Previously, Ohkura (2000) detected intraslab earthquakes that occurred within the subducting crust beneath W. Shikoku and N. Kyushu from the later phases traveling within the crust. If we accept that intraslab earthquakes occur due to dehydration embrittlement, these intracrustal earthquakes represent dehydration in the crust beneath W. Shikoku and N. Kyushu, which is consistent with the above inference on the dehydration of the PHS slab from the low-frequency tremors and the seismic tomography. It is, however, difficult to recognize the existence of the high Poisson s ratio zone in the forearc mantle (e.g., Fig. 14). The dehydration of the subducting slab thus might not be so persistent to produce serpentinized mantle in this region. Our tomography results show that the Poisson s ratio is low in the forearc mantle in M.-S. Kyushu (Fig. 12). With the absence of low-frequency tremors in this region (Obara, 2002), this may imply that the dehydration from the crust of the subducting slab is not significant. The recent study of intraslab earthquakes beneath Kyushu (Okamoto et al., 2005) made the situation more complex. They found several intracrustal earthquakes in the slab in M. Kyushu, but not in S. Kyushu. One possible interpretation of this Fig. 12. Distribution of Poisson s ratio structures at seven depth slices. Red and blue colors denote high and low Poisson s ratio, respectively. The range of the color scale to the right is in the first two layers and in the upper mantle layers. Other details are similar to those of Fig. 10. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

17 420 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Fig. 12 (continued ) fact is to hypothesize that in M. Kyushu, dehydration is strong enough to cause intracrustal earthquakes, but is not yet in the stage to induce low-frequency tremors or serpentinization of the forearc mantle. A closer examination of the dehydration of the subducting slab and the state of serpentinization of the forearc mantle will be necessary in this region. 5. Conclusions We selected a large number of high-quality arrival time data from JUNEC and Hi-net seismic networks and used them to study P- and S-wave velocity structures down to a depth of 150 km beneath Kyushu, SW. Japan. From the obtained velocity models, we further determined Poisson s ratio structures. Low-velocity anomalies are detected beneath the active and Quaternary volcanoes. However, there are no significant low-velocities in the crust beneath southern volcanoes. The subducting Philippine Sea slab is imaged clearly as a high-velocity zone down to a depth of 150 km, but with some portions of low- to intermediate-velocity. It does not reach the depth beneath the volcanic front in the northern region, but reaches it in the southern region. The seismicity in the crust of the upper

18 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Fig. 13. Vertical cross sections of V p, V s, and Poisson s ratio along line AA 0 (see Fig. 5 for location). Low-velocities and high Poisson s ratio are shown in red; high-velocities and low Poisson s ratio are shown in blue. Crosses show seismicity within 30 km wide-zone around the profile. Open triangle is Abu volcano. The perturbation scale is shown at the bottom and for Poisson s ratio; it varies from 0.21 to 0.35 in the top 20 km and from 0.26 to 0.40 at greater depths. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) plate occurs in the low Poisson s ratio region. The forearc crustal wedge in N. Kyushu at the junction of Shikoku and Honshu has a high Poisson s ratio, suggesting that it is affected by fluid-filled cracks due to dehydration of the slab. The forearc mantle wedge in M.-S. Kyushu has a low Poisson s ratio, suggesting that dehydration of the slab is not significant. This is consistent with the inference of the non-existence of the low-frequency tremors in the latter

19 422 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Fig. 14. The same as Fig. 13, but along cross section BB 0.

20 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Fig. 15. The same as Fig. 13, but along cross section CC 0 passing through Kujyu volcano (solid triangle).

21 424 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Fig. 16. The same as Fig. 13, but along cross section DD 0 passing through Aso volcano (solid triangle). region (Obara, 2002), although the occurrence of intraslab earthquakes in the crust in M. Kyushu (Okamoto et al., 2005) makes the situation more complex. Future work is needed to get more fine seismic velocity structures with higher spatial resolutions, because the backarc side and the off-shore regions have not been accurately imaged owing to the paucity of rays passing beneath them. Installation of Ocean Bottom Seismographs (OBS) off the coasts, and employment of the depth phases of offshore earthquakes (Wang and Zhao, 2005) will help to

22 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Fig. 17. The same as Fig. 13, but along cross section EE 0 passing near Unzen volcano (solid triangle).

23 426 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Fig. 18. The same as Fig. 13, but along cross section FF 0 passing through Kirishima volcano (solid triangle).

24 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) solve this matter, and will enable us to study this area in more detail. Acknowledgements We thank the Hi-net data center and JUNEC for preparing the arrival time data for academic research purposes. We also thank Dapeng Zhao for letting us use his tomography source code. Comments by Masao Nakada and the two anonymous reviewers greatly improved the manuscript. Most figures in this paper were produced using Generic Mapping Tools (GMT) software written by Wessel and Smith (1998). M.K. Salah thank JSPS for a Postdoctoral Fellowship (2005/2007) to conduct research at the Earthquake Research Institute, University of Tokyo, where this work was done. References Aki, K., Lee, W., Determination of three dimensional velocity anomalies under a seismic array using first P arrival times from local earthquakes: 1. A homogeneous initial model. J. Geophys. Res. 81, Christensen, N.I., Poisson s ratio and crustal seismology. J. Geophys. Res. 101, Gorbatov, A., Kennett, B.L.N., Joint bulk-sound and shear tomography for Western Pacific subduction zones. Earth Planet. Sci. Lett. 210, Hasegawa, A., Nakajima, J., Geophysical constraints on slab subduction and arc magmatism. The state of the planet: frontiers and challenges in geophysics. In: Geophysical Monograph, vol IUGG and AGU, Washington D.C., pp Hasegawa, A., Nakajima, J., Umino, N., Miura, S., Deep structure of the northeastern Japan arc and its implications for crustal deformation and shallow seismic activity. Tectonophysics 403, Hasegawa, A., Umino, N., Takagi, A., Suzuki, S., Motoya, Y., Kameya, S., Tanaka, K., Sawada, Y., Spatial distribution of earthquakes beneath Hokkaido and northern Honshu, Japan. J. Seismol. Soc. Jpn. 36, (in Japanese with English abstract). Hirahara, K., A large scale three-dimensional seismic structure under Japan Island and the Sea of Japan. J. Phys. Earth 25, Hirahara, K., Three-dimensional seismic structure beneath southwest Japan: the subducting Philippine Sea plate. Tectonophysics 79, Honda, S., Nakanishi, I., Shape of the Philippine Sea slab in the mantle and volcanism in the Chubu region, Central Japan. Geophys. Bull. Hokkaido Univ. 65, (in Japanese with English abstract). Honda, S., Nakanishi, I., Seismic tomography of the uppermost mantle beneath southwestern Japan: seismological constraints on modeling subduction and magmatism for the Philippine Sea slab. Earth Planets Space 55, Horiuchi, S., Ishii, H., Takagi, A., 1982a. Two-dimensional depth structure of the crust beneath the Tohoku district, the northeastern Japan arc. I. Method and Conrad discontinuity. J. Phys. Earth 30, Horiuchi, S., Yamamoto, A., Ueki, A., Tachibana, K., Kono, T., Takagi, A., 1982b. Two-dimensional depth structure of the crust beneath the Tohoku district, the northeastern Japan arc. II. Moho discontinuity and P-wave velocity. J. Phys. Earth 30, Hyndman, R.D., Peacock, S.M., Serpentinization of the forearc mantle. Earth Planet. Sci. Lett. 212, Hyndman, R.D., Yamano, M., Oleskevich, D.A., The seismogenic zone of subduction thrust faults. The Island Arc. 6, Inoue, H., Fukao, Y., Tanabe, K., Ogata, Y., Whole mantle P wave travel time tomography. Phys. Earth Planet. Interiors 59, Iwamori, H., Zonal structure of Cenozoic basalts related to mantle upwelling in southwest Japan. J. Geophys. Res. 96, Iwamori, H., Zhao, D., Melting and seismic structure beneath the northeast Japan arc. Geophys. Res. Lett. 27, Kamiya, S., Kobayashi, Y., Seismological evidence for the existence of serpentinized wedge mantle. Geophys. Res. Lett. 27, Kamiya, S., Kobayashi, Y., in preparation. Three dimensional seismic velocity structures beneath Kanto-Tokai district, central Japan. Katsumata, A., Kamaya, N., Low-frequency continuous tremor around the Moho discontinuity away from volcanoes in the southwest Japan. Geophys. Res. Lett. 30 (1). doi: /2002gl Kayal, J.R., Zhao, D., Mishra, O.P., De, R., Singh, O.P., The 2001 Bhuj earthquake: tomographic evidence for fluids at the hypocenter and its implications for rupture nucleation. Geophys. Res. Lett. 29 (24), doi: /2002gl Kirby, S., Engdhal, E.R., Denlinger, R., Intermediate-depth intraslab earthquakes and arc volcanism as physical expressions of crustal and uppermost mantle metamorphism in subducting slabs (Overview). In: Bebout, G.E., Scholl, D.W., Kirby, S.H., Platt, J.P. (Eds.), Subduction Top to Bottom. Geophys. Monogr. AGU, Washington, DC, 96, Kirby, S.H., Wang, K., Brocher, T.M., submitted for publication. A large mantle water source for the northern San Andreas fault system. Nature. Matsubara, M., Hayashi, H., Obara, K., Kasahara, K., Lowvelocity oceanic crust at the top of the Philippine Sea and Pacific plates beneath the Kanto region, central Japan, imaged by seismic tomography. J. Geophys. Res doi: /2005jb Mishra, O.P., Zhao, D., Umino, N., Hasegawa, A., Tomography of northeast Japan forearc and its implications for interplate seismic coupling. Geophys. Res. Lett. 30 (16), doi: / 2003GL Nakada, S., Kamata, H., Temporal change in chemistry of magma source under Central Kyushu, Southwest Japan: progressive contamination of mantle wedge. Bull. Volcanol. 53, Nakajima, J., Matsuzawa, T., Hasegawa, A., Moho depth variation in the central part of northeastern Japan estimated from reflected and converted waves. Phys. Earth Planet. Inter. 130, Nakajima, J., Matsuzawa, T., Hasegawa, A., Zhao, D., 2001a. Threedimensional structure of V p, V s, and V p /V s beneath northeastern Japan: implications for arc magmatism and fluids. J. Geophys. Res. 106, Nakajima, J., Matsuzawa, T., Hasegawa, A., Zhao, D., 2001b. Seismic imaging of arc magma and fluids under the central part of northeastern Japan. Tectonophysics 341, Nakajima, J., Takei, Y., Hasegawa, A., Quantitative analysis of the inclined low-velocity zone in the mantle wedge of northeastern Japan: a systematic change of melt-filled pore shapes with depth and its implications for melt migration. Earth Planet. Sci. Lett. 234, Nakamura, E., Campbell, I.H., McCulloch, M.T., Sun, S., Chemical geodynamics in a back arc region around the Sea of Japan: implications for the genesis of alkaline basalts in Japan, Korea and China. J. Geophys. Res. 94, Nakamura, M., Yoshida, Y., Zhao, D., Yoshikawa, K., Takayama, H., Aoki, G., Kuroki, H., Yamazaki, T., Kasahara, J., Kanazawa, T., Sato, T., Shiobara, H., Shimamura, H., Nakanishi, A., Threedimensional P and S wave velocity structure beneath ventral Japan. Pap. Meteorol. Geophys. 53, 1 28 (in Japanese with English abstract). Obara, K., Nonvolcaic deep tremor associated with subduction in southwest Japan. Science 296, Ohkura, T., Structure of the upper part of Philippine Sea plate estimated by later phases of upper mantle earthquakes in and around Shikoku, Japan. Tectonophysics 321, Okamoto, K., Ohkura, T., Seno, T., Seismicity within the Philippine Sea slab in the central and south Kyushu, Japan. Abstr. Seism. Soc. Jpn., 122.

25 428 M.K. Salah, T. Seno / Journal of Asian Earth Sciences 31 (2008) Okino, K., Shimakawa, Y., Nagano, S., Evolution of the Shikoku Basin. J. Geomag. Geoelectr. 46, Peacock, S.M., Wang, K., Seismic consequences of warm versus cool subduction metamorphism: examples from southwest and northeast Japan. Science 286, Sadeghi, H., Fatemi Aghda, S.M., Suzuki, S., Nakamura, T., D velocity structure of the 2003 Bam earthquake area (SE Iran): existence of a low Poisson s ratio layer and its relation to heavy damage. Tectonophysics 417, Sadeghi, H., Suzuki, S., Takenaka, H., Tomographic low-velocity anomalies in the uppermost mantle around the northeastern edge of Okinawa Trough, the backarc of Kyushu. Geophys. Res. Lett. 27, Salah, M.K., Zhao, D., 2003a. 3-D seismic structure of Kii Peninsula in southwest Japan: evidence for slab dehydration in the forearc. Tectonophysics 364, Salah, M.K., Zhao, D., 2003b. Three-dimensional attenuation structure beneath southwest Japan estimated from spectra of microearthquakes. Phys. Earth Planet. Inter. 136, Salah, M.K., Zhao, D., Mapping the crustal thickness in southwest Japan using Moho-reflected waves. Phys. Earth Planet. Inter. 141, Salah, M.K., Zhao, D., Lei, J., Abdelwahed, M.F., Crustal heterogeneity beneath southwest Japan estimated from direct and Moho-reflected waves. Tectonophysics 395, Seno, T., Tectonic evolution of the West Philippine Basin. Mod. Geol. 12, Seno, T., Syntheses of the regional stress fields of the Japanese Islands. Isl. Arc. 8, Seno, T., Variation of downdip limit of the seismogenic zone near the Japanese islands: implications for the serpentinization mechanism of the forearc mantle wedge. Earth Planet. Sci. Lett. 231, Seno, T., Sakurai, T., Stein, S., Can the Okhotsk plate be discriminated from the North American plate? J. Geophys. Res. 101, Seno, T., Stein, S., Gripp, A., A model for the motion of the Philippine Sea plate consistent with NUVEL-1 and geological data. J. Geophys. Res. 98, Seno, T., Yamanaka, Y., Double seismic zones, deep compressional trench outer rise events and superplumes. In: Bebout, G.E., Scholl, D.W., Kirby, S.H., Platt, J.P. (Eds.), Subduction Top to Bottom, Geophys. Monogr. AGU, Washington, DC, 96, Seno, T., Yamasaki, T., Low-frequency tremors, intraslab and interplate earthquakes in southwest Japan-from a viewpoint of slab dehydration. Geophys. Res. Lett. 30 (22), doi: / 2003GL Seno, T., Zhao, D., Kobayashi, Y., Nakamura, M., Dehydration of serpentinized slab mantle: seismic evidence from southwest Japan. Earth Planets Space 53, Spakman, W., Nolet, G., Imaging algorithms, accuracy and resolution in delay time tomography. In: Vlaar, N.J. et al. (Eds.), Math. Geophys.. D. Reidel, Norwell, MA, pp Sugimura, A., Zonal arrangement of some geophysical and petrological features in Japan and its environments. J. Fac. Sci. Univ. Tokyo, Ser. II XII, Sumino, H., Nakai, S., Nagao, K., Notsu, K., High 3 He/ 4 He ratio in zenoliths from Takashima: evidence for plume type volcanism in southwestern Japan. Geophys. Res. Lett. 27, Takei, Y., Effects of pore geometry on V p /V s : from equilibrium geometry to crack. J. Geophys. Res doi: /2001jb Thurber, C.H., Earthquakes locations and three-dimensional crustal structure in the Coyote Lake area, central California. J. Geophys. Res. 88, Tsuboi, S., Koketsu, K., Takano, K., Miyatake, T., Abe, K., Hagiwara, Y., Hypocenter determination procedure of Japan University Network Earthquake Catalog. J. Seismol. Soc. Jpn. 42, Tsumura, N., Matsumoto, S., Horiuchi, S., Hasegawa, A., Threedimensional attenuation structure beneath the northeastern Japan arc estimated from spectra of small earthquakes. Tectonophysics 319, Uyehira, K., Shimizu, H., Matsuo, N., Goto, K., Geometry and focal mechanisms of intermediate-depth earthquakes from the western end of Shikoku and SW Honshu to Kyushu. The Earth Monthly 23, Wang, Z., Zhao, D., Seismic imaging of the entire arc of Tohoku and Hokkaido in Japan using P-wave, S-wave and sp depth-phase data. Phys. Earth Planet. Inter. 152, Wessel, P., Smith, W.H.F., New improved version of Generic Mapping Tools released. EOS Trans. Am. Geophys. Un. 79, 579. Widiyantoro, S., Kennett, B.L.N., van der Hilst, R.D., Seismic tomography with P and S data reveals lateral variations in the rigidity of deep slabs. Earth Planet. Sci. Lett. 173, Yagi, Y., Kikuchi, M., Partitioning between seismogenic and aseismic slip as highlighted from slow slip events in Hyuga-nada, Japan. Geophys. Res. Lett. 30. doi: /2003gl Yamane, T., Nakanishi, I., Kamiya, S., P-wave velocity structure of the Philippine Sea plate subducting from the Nankai trough and its relation with seismicity and volcanism in southwestern and central Japan. Bull. Earthq. Res. Inst. Univ. Tokyo 75, Yamasaki, T., Seno, T., Double seismic zones and dehydration reactions in the subducting slab. J. Geophys. Res doi: / 2002JB Yokoyama, I., Aramaki, S., Nakamura, K., Volcanoes. Iwanami Press, Tokyo, pp Yoshida, A., Seno, T., Tectonic origin of the Aso volcano. Kazan 37, (in Japanese). Zhao, D., Asamori, K., Iwamori, H., 2000a. Seismic structure and magmatism of the young Kyushu subduction zone. Geophys. Res. Lett. 27, Zhao, D., Hasegawa, A., Horiuchi, S., 1992a. Tomographic imaging of P and S wave velocity structure beneath northeastern Japan. J. Geophys. Res. 97, Zhao, D., Hasegawa, A., Kanamori, H., Deep structure of Japan subduction zone as derived from local, regional and teleseismic events. J. Geophys. Res. 99, Zhao, D., Horiuchi, S., Hasegawa, A., 1992b. Seismic velocity structure of the crust beneath the Japan Islands. Tectonophysics 212, Zhao, D., Negishi, H., The 1995 Kobe earthquake: seismic image of the source zone and its implications for the rupture nucleation. J. Geophys. Res. 103, Zhao, D., Ochi, F., Hasegawa, A., Yamamoto, A., 2000b. Evidence for the location and cause of large crustal earthquakes in Japan. J. Geophys. Res. 105, Zhou, H., Clayton, R.W., P and S wave travel time inversions for subducting slab under the island arcs of northwest Pacific. J. Geophys. Res. 95,