- Email: [email protected]
Unusually deep Bonin earthquake of 30 May 2015: A precursory signal to slab penetration?
JID:EPSL
AID:14183 /SCO
[m5G; v1.191; Prn:5/12/2016; 8:22] P.1 (1-6)
Earth and Planetary Science Letters ••• (••••) •••–•••
Contents lists available at ScienceDirect
Earth and Planetary Science Letters www.elsevier.com/locate/epsl
Unusually deep Bonin earthquake of 30 May 2015: A precursory signal to slab penetration? Masayuki Obayashi a,∗ , Yoshio Fukao b , Junko Yoshimitsu a a
Department of Deep Earth Structure and Dynamics Research, Japan Agency for Marine–Earth Science and Technology, Natsushima-cho 2-15, Yokosuka, Kanagawa, 237-0061, Japan b Research and Development Center for Earthquake and Tsunami, Japan Agency for Marine–Earth Science and Technology, Yokohama, Kanagawa, 236-0001, Japan
a r t i c l e
i n f o
Article history: Received 26 August 2016 Received in revised form 7 November 2016 Accepted 9 November 2016 Available online xxxx Editor: P. Shearer Keywords: deep earthquake stagnant slab slab penetration 660-km discontinuity
a b s t r a c t An M7.9 earthquake occurred on 30 May 2015 at an unusual depth of 680 km downward and away from the well-defined Wadati–Benioff (WB) zone of the southern Bonin arc. To the north (northern Bonin), the subducted slab is stagnant above the upper–lower mantle boundary at 660-km depth, where the WB zone bends forward to sub-horizontal. To the south (northern Mariana), it penetrates the boundary, where the WB zone extends near-vertically down to the boundary. Thus, the southern Bonin slab can be regarded as being in a transitional state from slab stagnation to penetration. The transition is shown to happen rapidly within the northern half of the southern Bonin slab where the heel part of the shoelike configured stagnant slab hits the significantly depressed 660-km discontinuity. The mainshock and aftershocks took place in this heel part where they are sub-vertically aligned in approximate parallel to their maximum compressional axes. Here, the dips of the compressional axes of WB zone earthquakes change rapidly across the thickness of the slab from the eastern to western side and along the strike of the slab from the northern to southern side, suggesting rapid switching of the downdip compression axis in the shoe-shaped slab. Elastic deformation associated with the WB zone seismicity is calculated by viewing it as an integral part of the slab deformation process. With this deformation, the heel part is deepened relative to the arch part and is compressed sub-vertically and stretched sub-horizontally, a tendency consistent with the idea of progressive decent of the heel part in which near-vertical compressional stress is progressively accumulated to generate isolated shocks like the 2015 event and eventually to initiate slab penetration. © 2016 Elsevier B.V. All rights reserved.
1. Introduction The earthquake occurred on 30 May 2015 at unusual depth of 680 km was the 5th largest of earthquakes below 300 km since 1960 (Houston, 2007). This event, referred to as the Bonin super deep (BSD) shock, is unique not only because of its considerable depth and magnitude (global CMT; http://www.globalcmt.org; Dziewonski et al., 1981; Ekström et al., 2012) but also because of its location in the subducting Pacific slab (Ye et al., 2016; Takemura et al., 2016). Fig. 1A shows the vertical cross-section of a three-dimensional P-wave tomography model, GAP_P4 (Obayashi et al., 2013; Fukao and Obayashi, 2013) along the east–west profile through the hypocenter (red dot in Fig. 1). The cold subducting Pacific slab is imaged as fast anomalies (in blue). From the surface down to the transition zone the slab sub-vertically dips and then
*
Corresponding author. E-mail address: [email protected] (M. Obayashi).
http://dx.doi.org/10.1016/j.epsl.2016.11.019 0012-821X/© 2016 Elsevier B.V. All rights reserved.
sharply bends to sub-horizontal near the 660-km discontinuity. The hypocentral distribution of deep shocks in the past (Engdahl et al., 1998) defines the WB zone, which delineates the cold core of the subducting slab. Note that the WB zone between 450 and 550 km depths also kinks where the tomographically imaged slab bends. An event isolated from the WB zone is the 6.7 magnitude, 547.5-km-depth event on 4 July 1982 (Lundgren and Giardini, 1994) (green dot in Fig. 1A). This event occurred several hundreds of kilometers west in front of the main WB zone; however, it was still located near the core of the horizontally deflected part of the slab. An even more isolated event was the BSD shock that was located downwards and away from the WB zone. Its location corresponds to the heel portion of the bent slab, the appearance of which is similar to a shoe in the cross-sectional view. Although the focal depth exceeds the nominal depth of the 660-km discontinuity, the actual depth of the 660-km discontinuity can vary significantly depending on the thermal and compositional environment because of its phase boundary nature (Ito and Takahashi, 1989; Bina and Helffrich, 1994). The boundary depth
JID:EPSL
2
AID:14183 /SCO
[m5G; v1.191; Prn:5/12/2016; 8:22] P.2 (1-6)
M. Obayashi et al. / Earth and Planetary Science Letters ••• (••••) •••–•••
2. Slab morphology from Bonin to Mariana Fig. 2 shows a morphological change of the slab from northern Bonin (A–D), southern Bonin (E–H) to northern Mariana (I–L) based on the GAP_P4 model. In northern Bonin, the slab bends horizontally along the 660-km discontinuity, and its bending becomes increasingly sharp towards the south (A→D) by a progressive deepening of the bending portion. In northern Mariana, on the other hand, the slab is penetrating and the penetrated mass increases towards the south (I→L) (Fukao and Obayashi, 2013). The images in southern Bonin are understood to be those at the transitional stage from slab stagnation to penetration (E→H). Figs. 3A, 3B and 3C show slab images between the two adjacent sections C and D, E and F, and G and H, respectively, where the seismicity and focal mechanisms in the northern and southern sections are superposed in different colors. The resolution tests indicate that each of the targeted blocks near the 660-km discontinuity is reasonably resolved (Appendix A in Supplementary material). At the C→D stage (Fig. 3A) the heel part of the stagnant slab is rapidly deepening as it goes southward, although it does not yet reach the 660-km discontinuity. At the E→F stage (Fig. 3B) the heel part of the stagnant slab hits the 660-km discontinuity and depresses the 660-km phase boundary (Porritt and Yoshioka, 2016). This depression generates near-vertical compressional stresses in the heel portion just above the depressed boundary (Yoshioka et al., 1997; Bina, 1997; Bina et al., 2001). We interpret the BSD shock as a consequence of this stress environment. The horizontally deflected part of the slab becomes progressively decoupled from the rest. The rest behaves as a coherent body of near-vertical, slightly buckled slab (Alpert et al., 2010), which begins to penetrate the 660-km discontinuity at the G→H stage (Fig. 3C). Deep shocks at this stage are consistently down-dip compressional. Fig. 1. Location of the BSD shock. (A) The east–west cross-section (solid red line in (B)) of the GAP_P4 model. White dots indicate the hypocenters (Engdahl et al., 1998) within a band 50 km wide on both sides of the section plane. The focal mechanisms of the BSD shock and a subsequent earthquake on 23 June 2015 are projected onto the section plane. Black and white dots on the focal mechanism indicate the maximum and minimum compression directions, respectively. (B) Geographical location of the BSD shock (red dot). The hypocenters below 400 km are shown by colors varying with depth. The three broken red lines and a thick purple line are referred to in Figs. 3 and 5, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
in this region has been studied using S-to-P converted waves (Vidale and Benz, 1992; Wicks and Richards, 1993; Collier and Helffrich, 1997; Castle and Creager, 1998). The 660-km discontinuity is depressed by 35 ± 17 km (Castle and Creager, 1998); therefore, the hypocenter is in close proximity to the boundary, although it is difficult to judge whether the hypocenter lies above or below the depressed boundary owing to the lack of nearby conversion points and uncertainties in the hypocentral and conversion depths. A recent study using P-to-S converted waves reported that the 660-km discontinuity is at 690 km depth with up to ∼15 km variability, immediately below the BSD shock (Porritt and Yoshioka, 2016). Previous seismic tomography models have consistently shown morphological changes of subducting slabs from Izu–Bonin where the slab is stagnant at the base of the upper mantle to Mariana where the slab is penetrating vertically into the lower mantle (e.g. van der Hilst et al., 1991; Fukao et al., 1992; van der Hilst and Seno, 1993; Miller et al., 2005). The BSD shock was located in the transitional region from slab stagnation to penetration. In this paper we consider the implications for the 2015 event by integrating our recent tomographic images, hypocentral distribution and focal mechanisms of the mainshock and aftershocks and the strain field around the hypocenter induced by WB zone seismicity.
3. Aftershocks The slab at the E→F stage (Fig. 3B) is therefore the mature stagnant slab in a transitional state to penetrating slab. The BSD shock and its aftershocks may be regarded as the forerunners of WB zone earthquakes at greatest depths in the near-vertically downgoing slab continuing all the way from the surface, as will be discussed below. Only five aftershocks were reported by the USGS National Earthquake Information Center (USGS-NEIC) (Japan Meteorological Agency (JMA) reported four aftershocks except for the first one). The first three aftershocks occurred within 2 h of the BSD shock and the remaining two, including the largest event with Mb 4.9 (2 June 2015), occurred 3 to 5 days after the BSD shock. We simultaneously relocated the mainshock and aftershocks using the first arrival data collected by USGS-NEIC. The relocation method incorporates absolute travel-time residuals and differential traveltime residuals between different events at the same stations, referred to as double differences (Waldhauser and Ellsworth, 2000), to constrain their relative locations. The travel time residuals were calculated with respect to the GAP_P4 model. The result is shown with estimated errors in Fig. 4A compared to the USGS-NEIC and JMA hypocenters. The depth of the BSD shock was relocated to 683.6 km that is closer to the JMA determination. The relocated aftershocks became closer to the mainshock and to each other than the original USGS-NEIC determinations. Unlike the subevents of the mainshock (Ye et al., 2016), the aftershocks do not lie on either of the nodal planes of the mainshock but rather extend in deeper directions. Fig. 4B shows the first motion polarities of the largest aftershock. The up-or-down initial motions of this aftershock were visually evaluated on the vertical P-wave records of the IRIS broadband network and the Japanese high-sensitivity seismograph net-
JID:EPSL
AID:14183 /SCO
[m5G; v1.191; Prn:5/12/2016; 8:22] P.3 (1-6)
M. Obayashi et al. / Earth and Planetary Science Letters ••• (••••) •••–•••
3
Fig. 2. Successive cross-sections of the GAP_P4 model across the northern Bonin (A–D), southern Bonin (E–H) and northern Mariana (I–L) arcs. The profiles of the crosssections are shown in the bottom map. Section (E) is the same as in Fig. 1A. The three depth lines indicate 410 km, 660 km and 1000 km. The hypocenters (Engdahl et al., 1998) within a band 50 km wide on both sides of the section plane are shown by white dots. In Fig. 3A, the hypocenters in sections C and D will be distinguished by white and red dots across the tomographic cross-section at the middle of these two sections. Similarly, in Fig. 3B (3C), the hypocenters in sections E and F (G and H) will be distinguished by white and red dots across the tomographic cross-section at the middle of these two sections. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
work (Hi-Net). Fig. 4B shows the P-wave polarity distribution of the largest aftershock projected onto the lower focal hemisphere. The polarity data in the upper hemisphere are plotted on the lower hemisphere, assuming an anti-podal symmetry of the polarity distribution. Superposed on these plots is the beach ball pattern of the best double couple solution of the BSD shock. Although most of the polarity data are located around the periphery of the focal hemisphere, their azimuthal coverage is sufficient to confirm their consistency with the pattern expected from the mechanism of the mainshock. This consistency indicates a similarity in the mechanism between the mainshock and the largest aftershock. The aftershock distribution extends downwards nearvertically but slightly eastwards, approximately in parallel with the principal compression axes of the mainshock and largest aftershock (Fig. 4A). The near-vertical alignment of the mainshock–aftershock sequence with the near-vertical compression axes suggests that the mainshock–aftershock distribution can be viewed as the incipient WB zone, which will develop into the primary WB zone at the G→H stage (Fig. 3C).
4. Cumulative deformation field due to the WB zone earthquakes Transition from the sub-horizontally deflected WB zone (Fig. 3A) to the sub-vertically aligned WB zone (Fig. 3C) takes place sharply through the intermediate stage (Fig. 3B) where the BSD shock happened. As elastic deformation associated with this WB zone seismicity can be viewed as an integral part of the slab deformation, its calculation should give some information about the ongoing process of the latter which is likely to involve large aseismic deformation outside the cold slab core. We calculated the deformational field produced by deep earthquakes using a simple dislocation model (Okada, 1992). We used a cross-section through the BSD shock normal to the strike of the WB zone in a depth range between 400 km and 500 km (thick purple line in Fig. 1B). We selected earthquakes at depths greater than 200 km within a band 25 km wide on both sides of the section plane from the global CMT catalogue in the period from 1976 to 2014 and obtained a total of 24 earthquakes. We rotated each focal mechanism into a local slab reference frame that consists of the slab strike-
JID:EPSL
4
AID:14183 /SCO
[m5G; v1.191; Prn:5/12/2016; 8:22] P.4 (1-6)
M. Obayashi et al. / Earth and Planetary Science Letters ••• (••••) •••–•••
Fig. 3. Cross-sections of the GAP_P4 model across the southern section of the northern Bonin (A) and the northern and southern section of the southern Bonin (B and C, respectively). The profiles of the cross-sections are shown by the broken red lines in Fig. 1B. The hypocenters (Engdahl et al., 1998) and focal mechanisms (global CMT) within a band 100 km wide on the northern and southern sides of the section plane are shown in black and red, respectively. See the caption of Fig. 2 for more detail. The mechanisms are projected onto the corresponding section planes. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
normal (along the purple line in Fig. 1B), slab strike-parallel and vertical coordinate axes. We calculated the deformation in a homogeneous isotropic elastic mantle due to a point source for each earthquake and summed up the deformation for all the events. As the earthquakes are mostly down-dip compression type, such as the 23 June 2015 earthquake (yellow dot in Fig. 1A), the displacement component along the cross-section plane is dominant (larger by more than ten times than the component orthogonal to the plane); therefore, only this component of deformation is shown in Fig. 5. Fig. 5A indicates how a square frame of slab thickness size on the section plane is deformed by the seismicity. Large deformation is concentrated near the bent portion of the WB zone where the slab is shortened in the down-dip direction and stretched in the orthogonal direction. The lower frame that was originally at a constant depth of 680 km is warped to cause relative deepening of the heel part of the stagnant slab against the arch part in qualitative agreement with the slab bottom image (Fig. 3B). Fig. 5B shows a close up of the deformation of the rect-
Fig. 4. (A) Locations of the BSD shock and its aftershocks. The events are numbered in temporal order. Events 1 and 5 are the main shock and largest aftershock. The locations from USGS-NEIC, JMA and this study are shown in black, blue and red, respectively. Our result is shown with error bars. Open circles indicate the location of Event 2 (not reported by JMA) calculated using less than 20 stations. Closed circles indicate the locations of all the other events calculated using more than 60 stations. The CMT solution of the mainshock (Event 1) and the inferred focal mechanism of the largest aftershock (Event 5) are shown at the relocated hypocenters. Top: Projection onto the horizontal plane. Bottom: Projection onto the section plane in Fig. 1A. (B) Polarity distribution of the initial motions for Event 5. The up and down motions, shown by the red and blue dots, respectively, are plotted on the lower half of the focal sphere. The beach-ball pattern represents the best double couple solution of the BSD shock. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
angular frame with square meshes in a depth range from 620 km to 700 km. The square frame surrounding the hypocenter of the BSD shock (marked by the thick lines) is compressed in the subvertical direction and stretched in the sub-horizontal direction in qualitative agreement with the focal mechanisms of the BSD shock and the largest aftershock. This mode of deformation is qualitatively consistent with the idea that the heel part of the slab is progressively descending relative to the arch part and that nearvertical compressional stress is progressively accumulating in this heel part.
JID:EPSL
AID:14183 /SCO
[m5G; v1.191; Prn:5/12/2016; 8:22] P.5 (1-6)
M. Obayashi et al. / Earth and Planetary Science Letters ••• (••••) •••–•••
Fig. 5. Slab deformation due to deep WB zone earthquakes within a band 25 km wide on both sides of the section plane along the thick purple line in Fig. 1B. Panel (A) shows deformation of a square frame of slab size on the section plane. The red focal mechanisms are used in the calculations. (B) Close up of the deformation of the rectangular frame with square meshes. Deformation of the 4 × 4 square meshes surrounding the hypocenter of the BSD shock is shown by the thick lines. Displacements at the calculated points are shown by white arrows and the displacement of 1 mm is indicated by horizontal arrow near the bottom of each panel. Background colors represent the GAP_P4 model. The location of the BSD shock is shown by its focal mechanism in each panel. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
5. Discussions We have shown that 1) The southern Bonin slab is in a transitional state from the stagnant slab in northern Bonin to the penetrated slab in northern Mariana and this transition occurs rapidly over a short distance of 200 km along the southern Bonin arc where the stagnant slab is shoe-shaped. 2) The 2015 event occurred in the heel part of this shoe-shaped stagnant slab, where the main shock and aftershocks are aligned near-vertically under near-vertical compression axes in close proximity to the depressed 660-km discontinuity. 3) The deformational field due to the WB zone seismicity is qualitatively consistent with the idea of progressive development of the heel part in which vertical compressional stress is progressively accumulating to generate the BSD shock. To discuss more about these results, Fig. 6 shows the distributions of the principal compressional axes of deep shocks near the bent portion of the stagnant slab. In Fig. 6A, the dip of the compressional axis systematically changes along a constant depth line in the slab, gentler on the Philippine Sea (western) side and steeper on the Pacific (eastern) side. To this end, the compressional axes in the easternmost part of the slab dip almost vertically and seem to continue to the near-vertical compressional axis of the BSD shock. Progressively thickening of the heel part enhances negative buoyancy and acts as a driving force for slab penetration (Goes et al., 2008; Nakakuki et al., 2010). On the other hand, the resultant lowering of the environmental temperature depresses
5
Fig. 6. Same as Fig. 3, but compressional axes are shown instead of focal mechanisms. (A) and (B) correspond to Fig. 3B and 3C, respectively. Contours of 0.5% and 0.6% fast anomalies are superimposed on (A) to show how the heel part develops from the more central part of the slab.
the 660-km phase boundary, which generates positive buoyancy, thereby causes near-vertical compressional stresses in the heel portion (Yoshioka et al., 1997; Bina, 1997; Bina et al., 2001). The stresses become large enough to cause high stress-drop events, such as the BSD shock (38 MPa) (Ye et al., 2016), and to allow the initiation of slab penetration. Fig. 6A depicts such a stress state just prior to slab penetration. Further to the south (Fig. 6B), the slab begins to penetrate into the lower mantle. The slab right above the depressed 660-km discontinuity is buckled to the Pacific side in response to the accumulated near-vertical compressive stresses (Myhill, 2013). The horizontal part of the stagnant slab is mechanically decoupled from the near-vertically downgoing part and is left over the 660-km discontinuity. Acknowledgement We thank anonymous reviewers for their reviews of our manuscript and comments for improving the manuscript. The seismic waveform data relevant to this letter are distributed by IRISDMC and National Research Institute for Earth Science and Disaster Prevention (Hi-Net). We used PTGMT (http://yoshida-geophys. jp/ptgmt.html) by Masaki Yoshida to plot compressional axes in Fig. 6. MO and YF had supports from JSPS KAKENHI Grant Number JP 25287116 and JSPS KAKENHI Grant Number JP 25247074, respectively. Appendix A. Supplementary material Supplementary material related to this article can be found online at http://dx.doi.org/10.1016/j.epsl.2016.11.019. References Alpert, L.A., Becker, T.W., Bailey, I.W., 2010. Global slab deformation and centroid moment tensor constraints on viscosity. Geochem. Geophys. Geosyst. 11, Q12006. http://dx.doi.org/10.1029/2010GC003301.
JID:EPSL
6
AID:14183 /SCO
[m5G; v1.191; Prn:5/12/2016; 8:22] P.6 (1-6)
M. Obayashi et al. / Earth and Planetary Science Letters ••• (••••) •••–•••
Bina, C.R., 1997. Patterns of deep seismicity reflect buoyancy stresses due to phase transitions. Geophys. Res. Lett. 24, 3301–3304. Bina, C.R., Helffrich, G.R., 1994. Phase transition Clapeyron slopes and transition zone seismic discontinuity topography. J. Geophys. Res. 99, 15853–15860. Bina, C.R., Stein, S., Marton, F.C., Van Ark, E.M., 2001. Implications of slab mineralogy for subduction dynamics. Phys. Earth Planet. Inter. 127, 51–66. Castle, J.C., Creager, K.C., 1998. Topography of the 660-km seismic discontinuity beneath Izu–Bonin: implications for tectonic history and slab deformation. J. Geophys. Res. 103, 12511–12527. Collier, J.D., Helffrich, G.R., 1997. Topography of the “410” and “660” km seismic discontinuities in the Izu–Bonin Subduction Zone. Geophys. Res. Lett. 24, 1535–1538. Dziewonski, A.M., Chou, T.A., Woodhouse, J.H., 1981. Determination of earthquake source parameters from waveform data for studies of global and regional seismicity. J. Geophys. Res. 86, 2825–2852. Ekström, G., Nettles, M., Dziewonski, A.M., 2012. The global CMT project 2004–2010: centroid-moment tensors for 13,017 earthquakes. Phys. Earth Planet. Inter. 200–201, 1–9. Engdahl, E.R., van der Hilst, R., Buland, R., 1998. Global teleseismic earthquake relocation with improved travel times and procedures for depth determination. Bull. Seismol. Soc. Am. 88, 722–743. Fukao, Y., Obayashi, M., 2013. Subducted slabs stagnant above, penetrating through and trapped below the 660-km discontinuity. J. Geophys. Res. 118, 5920–5938. Fukao, Y., Obayashi, M., Inoue, H., Nenbai, M., 1992. Subducting slabs stagnant in the mantle transition zone. J. Geophys. Res. 97, 4809–4822. Goes, S., Capitanio, F.A., Morra, G., 2008. Evidence of lower-mantle slab penetration phases in plate motions. Nature 451, 981–984. Houston, H., 2007. Deep Earthquakes. In: Schubert, G. (Ed.), Treatise on Geophysics, Vol. 4, 1st ed. Elsevier, Amsterdam, pp. 321–350. Ito, E., Takahashi, E., 1989. Post-spinel transformations in the system Mg2 SiO4 – Fe2 SiO4 and some geophysical implications. J. Geophys. Res. 94, 10637–10646. Lundgren, P., Giardini, D., 1994. Isolated deep earthquakes and the fate of subduction in the mantle. J. Geophys. Res. 99, 15833–15842. Miller, M.S., Gorbatov, A., Kennett, B.L.N., 2005. Heterogeneity within the subducting Pacific slab beneath the Izu–Bonin–Mariana arc: evidence from tomography using 3D ray tracing inversion techniques. Earth Planet. Sci. Lett. 235, 331–342.
Myhill, R., 2013. Slab buckling and its effect on the distributions and focal mechanisms of deep focus earthquakes. Geophys. J. Int. 192, 837–853. Nakakuki, T., Tagawa, M., Iwase, Y., 2010. Dynamical mechanisms controlling formation and avalanche of a stagnant slab. Phys. Earth Planet. Inter. 183, 309–320. Obayashi, M., Yoshimitsu, J., Nolet, G., Fukao, Y., Shiobara, H., Sugioka, H., Miyamachi, H., Gao, Y., 2013. Finite frequency whole mantle P-wave tomography: improvement of subducted slab. Geophys. Res. Lett. 40, 5652–5657. Okada, Y., 1992. Internal deformation due to shear and tensile faults in a half-space. Bull. Seismol. Soc. Am. 82, 1018–1040. Porritt, R.W., Yoshioka, S., 2016. Slab pileup in the mantle transition zone and the 30 May 2015 Chichi-jima earthquake. Geophys. Res. Lett. 43, 4905–4912. Takemura, S., Maeda, T., Furumura, T., Obara, K., 2016. Constraining the source location of the 30 May 2015 (Mw 7.9) Bonin deep-focus earthquake using seismogram envelopes of high-frequency P waveforms: occurrence of deepfocus earthquake at the bottom of a subducting slab. Geophys. Res. Lett. 43, 4297–4302. van der Hilst, R.D., Engdahl, R., Spakman, W., Nolet, G., 1991. Tomographic imaging of subducted lithosphere below northwest Pacific island arcs. Nature 353, 37–43. van der Hilst, R.D., Seno, T., 1993. Effects of relative plate motion on the deep structure and penetration depth of slabs below the Izu–Bonin and Mariana island arcs. Earth Planet. Sci. Lett. 120, 395–407. Vidale, J.E., Benz, H.M., 1992. Upper-mantle seismic discontinuities and the thermal structure of subduction zones. Nature 356, 678–683. Waldhauser, F., Ellsworth, W.L., 2000. A double-difference earthquake location algorithm: method and application to the Northern Hayward Fault. Bull. Seismol. Soc. Am. 90 (6), 1353–1368. Wicks, C.W., Richards, M.A., 1993. A detailed map of the 660-kilometer discontinuity beneath the Izu–Bonin subduction zone. Science 261, 1424–1427. Ye, L., Lay, T., Zhan, Z., Kanamori, H., Hao, J., 2016. The isolated ∼680 km deep 30 May 2015 MW 7.9 Ogasawara (Bonin) Islands earthquake. Earth Planet. Sci. Lett. 433, 169–179. Yoshioka, S., Daessler, R., Yuen, D.A., 1997. Stress fields associated with metastable phase transitions in descending slabs and deep-focus earthquakes. Phys. Earth Planet. Inter. 104, 345–561.
AID:14183 /SCO
[m5G; v1.191; Prn:5/12/2016; 8:22] P.1 (1-6)
Earth and Planetary Science Letters ••• (••••) •••–•••
Contents lists available at ScienceDirect
Earth and Planetary Science Letters www.elsevier.com/locate/epsl
Unusually deep Bonin earthquake of 30 May 2015: A precursory signal to slab penetration? Masayuki Obayashi a,∗ , Yoshio Fukao b , Junko Yoshimitsu a a
Department of Deep Earth Structure and Dynamics Research, Japan Agency for Marine–Earth Science and Technology, Natsushima-cho 2-15, Yokosuka, Kanagawa, 237-0061, Japan b Research and Development Center for Earthquake and Tsunami, Japan Agency for Marine–Earth Science and Technology, Yokohama, Kanagawa, 236-0001, Japan
a r t i c l e
i n f o
Article history: Received 26 August 2016 Received in revised form 7 November 2016 Accepted 9 November 2016 Available online xxxx Editor: P. Shearer Keywords: deep earthquake stagnant slab slab penetration 660-km discontinuity
a b s t r a c t An M7.9 earthquake occurred on 30 May 2015 at an unusual depth of 680 km downward and away from the well-defined Wadati–Benioff (WB) zone of the southern Bonin arc. To the north (northern Bonin), the subducted slab is stagnant above the upper–lower mantle boundary at 660-km depth, where the WB zone bends forward to sub-horizontal. To the south (northern Mariana), it penetrates the boundary, where the WB zone extends near-vertically down to the boundary. Thus, the southern Bonin slab can be regarded as being in a transitional state from slab stagnation to penetration. The transition is shown to happen rapidly within the northern half of the southern Bonin slab where the heel part of the shoelike configured stagnant slab hits the significantly depressed 660-km discontinuity. The mainshock and aftershocks took place in this heel part where they are sub-vertically aligned in approximate parallel to their maximum compressional axes. Here, the dips of the compressional axes of WB zone earthquakes change rapidly across the thickness of the slab from the eastern to western side and along the strike of the slab from the northern to southern side, suggesting rapid switching of the downdip compression axis in the shoe-shaped slab. Elastic deformation associated with the WB zone seismicity is calculated by viewing it as an integral part of the slab deformation process. With this deformation, the heel part is deepened relative to the arch part and is compressed sub-vertically and stretched sub-horizontally, a tendency consistent with the idea of progressive decent of the heel part in which near-vertical compressional stress is progressively accumulated to generate isolated shocks like the 2015 event and eventually to initiate slab penetration. © 2016 Elsevier B.V. All rights reserved.
1. Introduction The earthquake occurred on 30 May 2015 at unusual depth of 680 km was the 5th largest of earthquakes below 300 km since 1960 (Houston, 2007). This event, referred to as the Bonin super deep (BSD) shock, is unique not only because of its considerable depth and magnitude (global CMT; http://www.globalcmt.org; Dziewonski et al., 1981; Ekström et al., 2012) but also because of its location in the subducting Pacific slab (Ye et al., 2016; Takemura et al., 2016). Fig. 1A shows the vertical cross-section of a three-dimensional P-wave tomography model, GAP_P4 (Obayashi et al., 2013; Fukao and Obayashi, 2013) along the east–west profile through the hypocenter (red dot in Fig. 1). The cold subducting Pacific slab is imaged as fast anomalies (in blue). From the surface down to the transition zone the slab sub-vertically dips and then
*
Corresponding author. E-mail address: [email protected] (M. Obayashi).
http://dx.doi.org/10.1016/j.epsl.2016.11.019 0012-821X/© 2016 Elsevier B.V. All rights reserved.
sharply bends to sub-horizontal near the 660-km discontinuity. The hypocentral distribution of deep shocks in the past (Engdahl et al., 1998) defines the WB zone, which delineates the cold core of the subducting slab. Note that the WB zone between 450 and 550 km depths also kinks where the tomographically imaged slab bends. An event isolated from the WB zone is the 6.7 magnitude, 547.5-km-depth event on 4 July 1982 (Lundgren and Giardini, 1994) (green dot in Fig. 1A). This event occurred several hundreds of kilometers west in front of the main WB zone; however, it was still located near the core of the horizontally deflected part of the slab. An even more isolated event was the BSD shock that was located downwards and away from the WB zone. Its location corresponds to the heel portion of the bent slab, the appearance of which is similar to a shoe in the cross-sectional view. Although the focal depth exceeds the nominal depth of the 660-km discontinuity, the actual depth of the 660-km discontinuity can vary significantly depending on the thermal and compositional environment because of its phase boundary nature (Ito and Takahashi, 1989; Bina and Helffrich, 1994). The boundary depth
JID:EPSL
2
AID:14183 /SCO
[m5G; v1.191; Prn:5/12/2016; 8:22] P.2 (1-6)
M. Obayashi et al. / Earth and Planetary Science Letters ••• (••••) •••–•••
2. Slab morphology from Bonin to Mariana Fig. 2 shows a morphological change of the slab from northern Bonin (A–D), southern Bonin (E–H) to northern Mariana (I–L) based on the GAP_P4 model. In northern Bonin, the slab bends horizontally along the 660-km discontinuity, and its bending becomes increasingly sharp towards the south (A→D) by a progressive deepening of the bending portion. In northern Mariana, on the other hand, the slab is penetrating and the penetrated mass increases towards the south (I→L) (Fukao and Obayashi, 2013). The images in southern Bonin are understood to be those at the transitional stage from slab stagnation to penetration (E→H). Figs. 3A, 3B and 3C show slab images between the two adjacent sections C and D, E and F, and G and H, respectively, where the seismicity and focal mechanisms in the northern and southern sections are superposed in different colors. The resolution tests indicate that each of the targeted blocks near the 660-km discontinuity is reasonably resolved (Appendix A in Supplementary material). At the C→D stage (Fig. 3A) the heel part of the stagnant slab is rapidly deepening as it goes southward, although it does not yet reach the 660-km discontinuity. At the E→F stage (Fig. 3B) the heel part of the stagnant slab hits the 660-km discontinuity and depresses the 660-km phase boundary (Porritt and Yoshioka, 2016). This depression generates near-vertical compressional stresses in the heel portion just above the depressed boundary (Yoshioka et al., 1997; Bina, 1997; Bina et al., 2001). We interpret the BSD shock as a consequence of this stress environment. The horizontally deflected part of the slab becomes progressively decoupled from the rest. The rest behaves as a coherent body of near-vertical, slightly buckled slab (Alpert et al., 2010), which begins to penetrate the 660-km discontinuity at the G→H stage (Fig. 3C). Deep shocks at this stage are consistently down-dip compressional. Fig. 1. Location of the BSD shock. (A) The east–west cross-section (solid red line in (B)) of the GAP_P4 model. White dots indicate the hypocenters (Engdahl et al., 1998) within a band 50 km wide on both sides of the section plane. The focal mechanisms of the BSD shock and a subsequent earthquake on 23 June 2015 are projected onto the section plane. Black and white dots on the focal mechanism indicate the maximum and minimum compression directions, respectively. (B) Geographical location of the BSD shock (red dot). The hypocenters below 400 km are shown by colors varying with depth. The three broken red lines and a thick purple line are referred to in Figs. 3 and 5, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
in this region has been studied using S-to-P converted waves (Vidale and Benz, 1992; Wicks and Richards, 1993; Collier and Helffrich, 1997; Castle and Creager, 1998). The 660-km discontinuity is depressed by 35 ± 17 km (Castle and Creager, 1998); therefore, the hypocenter is in close proximity to the boundary, although it is difficult to judge whether the hypocenter lies above or below the depressed boundary owing to the lack of nearby conversion points and uncertainties in the hypocentral and conversion depths. A recent study using P-to-S converted waves reported that the 660-km discontinuity is at 690 km depth with up to ∼15 km variability, immediately below the BSD shock (Porritt and Yoshioka, 2016). Previous seismic tomography models have consistently shown morphological changes of subducting slabs from Izu–Bonin where the slab is stagnant at the base of the upper mantle to Mariana where the slab is penetrating vertically into the lower mantle (e.g. van der Hilst et al., 1991; Fukao et al., 1992; van der Hilst and Seno, 1993; Miller et al., 2005). The BSD shock was located in the transitional region from slab stagnation to penetration. In this paper we consider the implications for the 2015 event by integrating our recent tomographic images, hypocentral distribution and focal mechanisms of the mainshock and aftershocks and the strain field around the hypocenter induced by WB zone seismicity.
3. Aftershocks The slab at the E→F stage (Fig. 3B) is therefore the mature stagnant slab in a transitional state to penetrating slab. The BSD shock and its aftershocks may be regarded as the forerunners of WB zone earthquakes at greatest depths in the near-vertically downgoing slab continuing all the way from the surface, as will be discussed below. Only five aftershocks were reported by the USGS National Earthquake Information Center (USGS-NEIC) (Japan Meteorological Agency (JMA) reported four aftershocks except for the first one). The first three aftershocks occurred within 2 h of the BSD shock and the remaining two, including the largest event with Mb 4.9 (2 June 2015), occurred 3 to 5 days after the BSD shock. We simultaneously relocated the mainshock and aftershocks using the first arrival data collected by USGS-NEIC. The relocation method incorporates absolute travel-time residuals and differential traveltime residuals between different events at the same stations, referred to as double differences (Waldhauser and Ellsworth, 2000), to constrain their relative locations. The travel time residuals were calculated with respect to the GAP_P4 model. The result is shown with estimated errors in Fig. 4A compared to the USGS-NEIC and JMA hypocenters. The depth of the BSD shock was relocated to 683.6 km that is closer to the JMA determination. The relocated aftershocks became closer to the mainshock and to each other than the original USGS-NEIC determinations. Unlike the subevents of the mainshock (Ye et al., 2016), the aftershocks do not lie on either of the nodal planes of the mainshock but rather extend in deeper directions. Fig. 4B shows the first motion polarities of the largest aftershock. The up-or-down initial motions of this aftershock were visually evaluated on the vertical P-wave records of the IRIS broadband network and the Japanese high-sensitivity seismograph net-
JID:EPSL
AID:14183 /SCO
[m5G; v1.191; Prn:5/12/2016; 8:22] P.3 (1-6)
M. Obayashi et al. / Earth and Planetary Science Letters ••• (••••) •••–•••
3
Fig. 2. Successive cross-sections of the GAP_P4 model across the northern Bonin (A–D), southern Bonin (E–H) and northern Mariana (I–L) arcs. The profiles of the crosssections are shown in the bottom map. Section (E) is the same as in Fig. 1A. The three depth lines indicate 410 km, 660 km and 1000 km. The hypocenters (Engdahl et al., 1998) within a band 50 km wide on both sides of the section plane are shown by white dots. In Fig. 3A, the hypocenters in sections C and D will be distinguished by white and red dots across the tomographic cross-section at the middle of these two sections. Similarly, in Fig. 3B (3C), the hypocenters in sections E and F (G and H) will be distinguished by white and red dots across the tomographic cross-section at the middle of these two sections. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
work (Hi-Net). Fig. 4B shows the P-wave polarity distribution of the largest aftershock projected onto the lower focal hemisphere. The polarity data in the upper hemisphere are plotted on the lower hemisphere, assuming an anti-podal symmetry of the polarity distribution. Superposed on these plots is the beach ball pattern of the best double couple solution of the BSD shock. Although most of the polarity data are located around the periphery of the focal hemisphere, their azimuthal coverage is sufficient to confirm their consistency with the pattern expected from the mechanism of the mainshock. This consistency indicates a similarity in the mechanism between the mainshock and the largest aftershock. The aftershock distribution extends downwards nearvertically but slightly eastwards, approximately in parallel with the principal compression axes of the mainshock and largest aftershock (Fig. 4A). The near-vertical alignment of the mainshock–aftershock sequence with the near-vertical compression axes suggests that the mainshock–aftershock distribution can be viewed as the incipient WB zone, which will develop into the primary WB zone at the G→H stage (Fig. 3C).
4. Cumulative deformation field due to the WB zone earthquakes Transition from the sub-horizontally deflected WB zone (Fig. 3A) to the sub-vertically aligned WB zone (Fig. 3C) takes place sharply through the intermediate stage (Fig. 3B) where the BSD shock happened. As elastic deformation associated with this WB zone seismicity can be viewed as an integral part of the slab deformation, its calculation should give some information about the ongoing process of the latter which is likely to involve large aseismic deformation outside the cold slab core. We calculated the deformational field produced by deep earthquakes using a simple dislocation model (Okada, 1992). We used a cross-section through the BSD shock normal to the strike of the WB zone in a depth range between 400 km and 500 km (thick purple line in Fig. 1B). We selected earthquakes at depths greater than 200 km within a band 25 km wide on both sides of the section plane from the global CMT catalogue in the period from 1976 to 2014 and obtained a total of 24 earthquakes. We rotated each focal mechanism into a local slab reference frame that consists of the slab strike-
JID:EPSL
4
AID:14183 /SCO
[m5G; v1.191; Prn:5/12/2016; 8:22] P.4 (1-6)
M. Obayashi et al. / Earth and Planetary Science Letters ••• (••••) •••–•••
Fig. 3. Cross-sections of the GAP_P4 model across the southern section of the northern Bonin (A) and the northern and southern section of the southern Bonin (B and C, respectively). The profiles of the cross-sections are shown by the broken red lines in Fig. 1B. The hypocenters (Engdahl et al., 1998) and focal mechanisms (global CMT) within a band 100 km wide on the northern and southern sides of the section plane are shown in black and red, respectively. See the caption of Fig. 2 for more detail. The mechanisms are projected onto the corresponding section planes. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
normal (along the purple line in Fig. 1B), slab strike-parallel and vertical coordinate axes. We calculated the deformation in a homogeneous isotropic elastic mantle due to a point source for each earthquake and summed up the deformation for all the events. As the earthquakes are mostly down-dip compression type, such as the 23 June 2015 earthquake (yellow dot in Fig. 1A), the displacement component along the cross-section plane is dominant (larger by more than ten times than the component orthogonal to the plane); therefore, only this component of deformation is shown in Fig. 5. Fig. 5A indicates how a square frame of slab thickness size on the section plane is deformed by the seismicity. Large deformation is concentrated near the bent portion of the WB zone where the slab is shortened in the down-dip direction and stretched in the orthogonal direction. The lower frame that was originally at a constant depth of 680 km is warped to cause relative deepening of the heel part of the stagnant slab against the arch part in qualitative agreement with the slab bottom image (Fig. 3B). Fig. 5B shows a close up of the deformation of the rect-
Fig. 4. (A) Locations of the BSD shock and its aftershocks. The events are numbered in temporal order. Events 1 and 5 are the main shock and largest aftershock. The locations from USGS-NEIC, JMA and this study are shown in black, blue and red, respectively. Our result is shown with error bars. Open circles indicate the location of Event 2 (not reported by JMA) calculated using less than 20 stations. Closed circles indicate the locations of all the other events calculated using more than 60 stations. The CMT solution of the mainshock (Event 1) and the inferred focal mechanism of the largest aftershock (Event 5) are shown at the relocated hypocenters. Top: Projection onto the horizontal plane. Bottom: Projection onto the section plane in Fig. 1A. (B) Polarity distribution of the initial motions for Event 5. The up and down motions, shown by the red and blue dots, respectively, are plotted on the lower half of the focal sphere. The beach-ball pattern represents the best double couple solution of the BSD shock. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
angular frame with square meshes in a depth range from 620 km to 700 km. The square frame surrounding the hypocenter of the BSD shock (marked by the thick lines) is compressed in the subvertical direction and stretched in the sub-horizontal direction in qualitative agreement with the focal mechanisms of the BSD shock and the largest aftershock. This mode of deformation is qualitatively consistent with the idea that the heel part of the slab is progressively descending relative to the arch part and that nearvertical compressional stress is progressively accumulating in this heel part.
JID:EPSL
AID:14183 /SCO
[m5G; v1.191; Prn:5/12/2016; 8:22] P.5 (1-6)
M. Obayashi et al. / Earth and Planetary Science Letters ••• (••••) •••–•••
Fig. 5. Slab deformation due to deep WB zone earthquakes within a band 25 km wide on both sides of the section plane along the thick purple line in Fig. 1B. Panel (A) shows deformation of a square frame of slab size on the section plane. The red focal mechanisms are used in the calculations. (B) Close up of the deformation of the rectangular frame with square meshes. Deformation of the 4 × 4 square meshes surrounding the hypocenter of the BSD shock is shown by the thick lines. Displacements at the calculated points are shown by white arrows and the displacement of 1 mm is indicated by horizontal arrow near the bottom of each panel. Background colors represent the GAP_P4 model. The location of the BSD shock is shown by its focal mechanism in each panel. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
5. Discussions We have shown that 1) The southern Bonin slab is in a transitional state from the stagnant slab in northern Bonin to the penetrated slab in northern Mariana and this transition occurs rapidly over a short distance of 200 km along the southern Bonin arc where the stagnant slab is shoe-shaped. 2) The 2015 event occurred in the heel part of this shoe-shaped stagnant slab, where the main shock and aftershocks are aligned near-vertically under near-vertical compression axes in close proximity to the depressed 660-km discontinuity. 3) The deformational field due to the WB zone seismicity is qualitatively consistent with the idea of progressive development of the heel part in which vertical compressional stress is progressively accumulating to generate the BSD shock. To discuss more about these results, Fig. 6 shows the distributions of the principal compressional axes of deep shocks near the bent portion of the stagnant slab. In Fig. 6A, the dip of the compressional axis systematically changes along a constant depth line in the slab, gentler on the Philippine Sea (western) side and steeper on the Pacific (eastern) side. To this end, the compressional axes in the easternmost part of the slab dip almost vertically and seem to continue to the near-vertical compressional axis of the BSD shock. Progressively thickening of the heel part enhances negative buoyancy and acts as a driving force for slab penetration (Goes et al., 2008; Nakakuki et al., 2010). On the other hand, the resultant lowering of the environmental temperature depresses
5
Fig. 6. Same as Fig. 3, but compressional axes are shown instead of focal mechanisms. (A) and (B) correspond to Fig. 3B and 3C, respectively. Contours of 0.5% and 0.6% fast anomalies are superimposed on (A) to show how the heel part develops from the more central part of the slab.
the 660-km phase boundary, which generates positive buoyancy, thereby causes near-vertical compressional stresses in the heel portion (Yoshioka et al., 1997; Bina, 1997; Bina et al., 2001). The stresses become large enough to cause high stress-drop events, such as the BSD shock (38 MPa) (Ye et al., 2016), and to allow the initiation of slab penetration. Fig. 6A depicts such a stress state just prior to slab penetration. Further to the south (Fig. 6B), the slab begins to penetrate into the lower mantle. The slab right above the depressed 660-km discontinuity is buckled to the Pacific side in response to the accumulated near-vertical compressive stresses (Myhill, 2013). The horizontal part of the stagnant slab is mechanically decoupled from the near-vertically downgoing part and is left over the 660-km discontinuity. Acknowledgement We thank anonymous reviewers for their reviews of our manuscript and comments for improving the manuscript. The seismic waveform data relevant to this letter are distributed by IRISDMC and National Research Institute for Earth Science and Disaster Prevention (Hi-Net). We used PTGMT (http://yoshida-geophys. jp/ptgmt.html) by Masaki Yoshida to plot compressional axes in Fig. 6. MO and YF had supports from JSPS KAKENHI Grant Number JP 25287116 and JSPS KAKENHI Grant Number JP 25247074, respectively. Appendix A. Supplementary material Supplementary material related to this article can be found online at http://dx.doi.org/10.1016/j.epsl.2016.11.019. References Alpert, L.A., Becker, T.W., Bailey, I.W., 2010. Global slab deformation and centroid moment tensor constraints on viscosity. Geochem. Geophys. Geosyst. 11, Q12006. http://dx.doi.org/10.1029/2010GC003301.
JID:EPSL
6
AID:14183 /SCO
[m5G; v1.191; Prn:5/12/2016; 8:22] P.6 (1-6)
M. Obayashi et al. / Earth and Planetary Science Letters ••• (••••) •••–•••
Bina, C.R., 1997. Patterns of deep seismicity reflect buoyancy stresses due to phase transitions. Geophys. Res. Lett. 24, 3301–3304. Bina, C.R., Helffrich, G.R., 1994. Phase transition Clapeyron slopes and transition zone seismic discontinuity topography. J. Geophys. Res. 99, 15853–15860. Bina, C.R., Stein, S., Marton, F.C., Van Ark, E.M., 2001. Implications of slab mineralogy for subduction dynamics. Phys. Earth Planet. Inter. 127, 51–66. Castle, J.C., Creager, K.C., 1998. Topography of the 660-km seismic discontinuity beneath Izu–Bonin: implications for tectonic history and slab deformation. J. Geophys. Res. 103, 12511–12527. Collier, J.D., Helffrich, G.R., 1997. Topography of the “410” and “660” km seismic discontinuities in the Izu–Bonin Subduction Zone. Geophys. Res. Lett. 24, 1535–1538. Dziewonski, A.M., Chou, T.A., Woodhouse, J.H., 1981. Determination of earthquake source parameters from waveform data for studies of global and regional seismicity. J. Geophys. Res. 86, 2825–2852. Ekström, G., Nettles, M., Dziewonski, A.M., 2012. The global CMT project 2004–2010: centroid-moment tensors for 13,017 earthquakes. Phys. Earth Planet. Inter. 200–201, 1–9. Engdahl, E.R., van der Hilst, R., Buland, R., 1998. Global teleseismic earthquake relocation with improved travel times and procedures for depth determination. Bull. Seismol. Soc. Am. 88, 722–743. Fukao, Y., Obayashi, M., 2013. Subducted slabs stagnant above, penetrating through and trapped below the 660-km discontinuity. J. Geophys. Res. 118, 5920–5938. Fukao, Y., Obayashi, M., Inoue, H., Nenbai, M., 1992. Subducting slabs stagnant in the mantle transition zone. J. Geophys. Res. 97, 4809–4822. Goes, S., Capitanio, F.A., Morra, G., 2008. Evidence of lower-mantle slab penetration phases in plate motions. Nature 451, 981–984. Houston, H., 2007. Deep Earthquakes. In: Schubert, G. (Ed.), Treatise on Geophysics, Vol. 4, 1st ed. Elsevier, Amsterdam, pp. 321–350. Ito, E., Takahashi, E., 1989. Post-spinel transformations in the system Mg2 SiO4 – Fe2 SiO4 and some geophysical implications. J. Geophys. Res. 94, 10637–10646. Lundgren, P., Giardini, D., 1994. Isolated deep earthquakes and the fate of subduction in the mantle. J. Geophys. Res. 99, 15833–15842. Miller, M.S., Gorbatov, A., Kennett, B.L.N., 2005. Heterogeneity within the subducting Pacific slab beneath the Izu–Bonin–Mariana arc: evidence from tomography using 3D ray tracing inversion techniques. Earth Planet. Sci. Lett. 235, 331–342.
Myhill, R., 2013. Slab buckling and its effect on the distributions and focal mechanisms of deep focus earthquakes. Geophys. J. Int. 192, 837–853. Nakakuki, T., Tagawa, M., Iwase, Y., 2010. Dynamical mechanisms controlling formation and avalanche of a stagnant slab. Phys. Earth Planet. Inter. 183, 309–320. Obayashi, M., Yoshimitsu, J., Nolet, G., Fukao, Y., Shiobara, H., Sugioka, H., Miyamachi, H., Gao, Y., 2013. Finite frequency whole mantle P-wave tomography: improvement of subducted slab. Geophys. Res. Lett. 40, 5652–5657. Okada, Y., 1992. Internal deformation due to shear and tensile faults in a half-space. Bull. Seismol. Soc. Am. 82, 1018–1040. Porritt, R.W., Yoshioka, S., 2016. Slab pileup in the mantle transition zone and the 30 May 2015 Chichi-jima earthquake. Geophys. Res. Lett. 43, 4905–4912. Takemura, S., Maeda, T., Furumura, T., Obara, K., 2016. Constraining the source location of the 30 May 2015 (Mw 7.9) Bonin deep-focus earthquake using seismogram envelopes of high-frequency P waveforms: occurrence of deepfocus earthquake at the bottom of a subducting slab. Geophys. Res. Lett. 43, 4297–4302. van der Hilst, R.D., Engdahl, R., Spakman, W., Nolet, G., 1991. Tomographic imaging of subducted lithosphere below northwest Pacific island arcs. Nature 353, 37–43. van der Hilst, R.D., Seno, T., 1993. Effects of relative plate motion on the deep structure and penetration depth of slabs below the Izu–Bonin and Mariana island arcs. Earth Planet. Sci. Lett. 120, 395–407. Vidale, J.E., Benz, H.M., 1992. Upper-mantle seismic discontinuities and the thermal structure of subduction zones. Nature 356, 678–683. Waldhauser, F., Ellsworth, W.L., 2000. A double-difference earthquake location algorithm: method and application to the Northern Hayward Fault. Bull. Seismol. Soc. Am. 90 (6), 1353–1368. Wicks, C.W., Richards, M.A., 1993. A detailed map of the 660-kilometer discontinuity beneath the Izu–Bonin subduction zone. Science 261, 1424–1427. Ye, L., Lay, T., Zhan, Z., Kanamori, H., Hao, J., 2016. The isolated ∼680 km deep 30 May 2015 MW 7.9 Ogasawara (Bonin) Islands earthquake. Earth Planet. Sci. Lett. 433, 169–179. Yoshioka, S., Daessler, R., Yuen, D.A., 1997. Stress fields associated with metastable phase transitions in descending slabs and deep-focus earthquakes. Phys. Earth Planet. Inter. 104, 345–561.