Low--frequency tremors, intraslab and interplate earthquakes in

Southwest Japan—from a viewpoint of slab dehydration

Tetsuzo Seno1 and Tadashi Yamasaki2

Received 6 August 2003; revised 20 October 2003; accepted 28 October 2003; published 28 November 2003.

[1] Low-frequency tremors (LFT) recently found in theSouthwest Japan fore-arc likely occur due to hydro-fracturing from fluids leaving the dehydrating PhilippineSea slab. However, there are some places without such LFT;they are N. Izu-Kanto, E. Shikoku, and S. Kyushu. Theseare the places where island-arc type crust is subducting.We propose that dehydration of the subducted crust doesnot occur beneath these regions, because it is composedmainly of tonalite, lacking the quantity of hydrousminerals seen in normal subducting oceanic crust. Weshow that almost no earthquake occurs within the subductedcrust in such regions, consistent with dehydrationembrittlement hypothesis for intraslab seismicity. Thelack of dehydration from the crust would also affectthe mode of occurrence of interplate earthquakes, becausepore fluid pressure is more difficult to rise at the thrustzone. This might be reflected in the irregular occurrence ofgreat earthquakes in these regions. INDEX TERMS: 7209

Seismology: Earthquake dynamics and mechanics; 7220

Seismology: Oceanic crust; 7230 Seismology: Seismicity and

seismotectonics; 8120 Tectonophysics: Dynamics of lithosphere

and mantle—general; 8123 Tectonophysics: Dynamics,

seismotectonics. Citation: Seno, T., and T. Yamasaki, Low-

frequency tremors, intraslab and interplate earthquakes in

Southwest Japan—from a viewpoint of slab dehydration,Geophys. Res. Lett., 30(22), 2171, doi:10.1029/2003GL018349,

2003.

1. Introduction

[2] Because of large confining pressure in the slab, aweakening (embrittlement) mechanism for intraslab seismic-ity is required. At the intermediate-depth range, dehydrationembrittlement, a mechanical instability associated withdehydration of hydrated metamorphic minerals [Raleighand Paterson, 1965], would be the most likely mechanism[Kirby et al., 1996; Hacker et al., 2003; Yamasaki and Seno,2003].[3] A weakening mechanism is also required for inter-

plate earthquakes at the subduction zone thrust becauseconfining pressure in the thrust zone at 10–30 km depthreaches 0.3–1 GPa; on the other hand, tectonic stresses areon the order of 0.1 GPa or less [Fleitout, 1991]. Elevatedpore fluid pressure in fault zones would be one of the mostplausible factors to reduce rock strength by reducing effec-tive normal stresses [e.g., Hickman et al., 1995]. In asubduction zone thrust at depth, the most likely source of

the fluid would be water contained in the hydrous mineralsof the subducting oceanic crust, because dehydration of clayminerals in sediments is completed at depths shallower than10 km [Hyndman et al., 1997].[4] Therefore not only intraslab seismicity, but also

interplate seismic coupling would reflect dehy dration ofthe subducting slab. Recently low-frequency tremors (LFT)have been found around the Moho depth of the SouthwestJapan fore-arc, forming a narrow zone along the strike of thearc (Figure 1, Obara [2002]). These are small amplitudeevents, lacking sharp onsets and lasting a few minutes toa few days, with predominant frequency of 1 to 10 Hz,and the high sensitivity seismograph network over Japan(Hi-net) has made it possible to detect these subtle events[Obara, 2002]. They likely occur in relation with fluidsleaving the dehydrating slab, because they are located abovethe 35–45 km depth surface of the subducting PhilippineSea Plate (PHS). In this article, we explore the relationbetween LFT, intraslab seismicity and occurrence of greatinterplate earthquakes, in terms of dehydration of the PHSslab.

2. Low––Frequency Tremor and Bathymetryof the Philippine Sea

[5] Although LFT are distributed above the slab along thearc, there are three places where LFT are not seen. They areN. Izu-Kanto, E. Shikoku, and S. Kyushu (Figure 1). TheHi-net covers these areas, confirming that they are not out ofrange. We note that these regions are the places where non-oceanic crust is subducting (Figure 1). In N. Izu-Kanto, theIzu-Bonin volcanic ridge and its fore-arc are subducting.The Izu-Bonin Ridge is an island-arc that has been formedin an intra-oceanic setting. Seismic refraction studies along32� N indicate that the arc crust has an intermediate-depthlayer with a velocity of 6.0–6.3 km/sec of 10 km thickness,which is interpreted as a tonalitic plutonic layer [Taira et al.,1998]. Above this layer, there is a layer with a velocity of4.5–5.5 km/sec of 5 km thickness, composed of volcani-clastic rocks, lavas and intrusives of boninitic composition[Taira et al., 1998]. In S. Kyushu, the Kyushu-Palau Ridge,a remnant arc separated from the Izu-Bonin arc in associ-ation with the Shikoku Basin spreading in the Oligo-Miocene, is subducting. This ridge has also a tonaliticplutonic layer of �5 km thickness [Li et al., 1997]. InE. Shikoku, the Kinan Seamount Chain is subducting. Thiswas produced by the volcanic activity during 10 m.y. afterthe cessation of the spreading of the Shikoku Basin [Sato etal., 2002]. It is possible that the Shikoku basin basalt wasre-melted by this volcanism and produced granitic rocks atdepth, similarly to the Izu-Bonin Ridge, although there is nodirect evidence for this at present. These indicate that the

GEOPHYSICAL RESEARCH LETTERS, VOL. 30, NO. 22, 2171, doi:10.1029/2003GL018349, 2003

1Earthquake Research Institute, University of Tokyo, Japan.2National Institute of Polar Research, Japan.

Copyright 2003 by the American Geophysical Union.0094-8276/03/2003GL018349$05.00

SDE 8 -- 1

regions without LFT are mostly the places where island-arctype crust with an intermediate-depth tonalitic layer issubducting.

3. Seismicity Within the SubductingPhilippine Sea Slab

[6] We discuss here regional variation of the intraslabseismicity along the arc. Figure 2 shows the contours ofthe upper surface of the intraslab seismicity, LFT, and therupture zones of great or large earthquakes along theNankai–Sagami Trough. Generally the intraslab seismicityis forming a single zone within the subducting crust[Peacock and Wang, 1999; Seno et al., 2001]. There arefive places that do not obey this rule; they are Kanto, N. Izu,Kii Peninsula, E. Shikoku and S. Kyushu.[7] Beneath Kii Peninsula, a double seismic zone exists;

the upper one in the oceanic crust, and the lower one in themantle [Seno et al., 2001]. Beneath Kanto, E. Shikoku, andprobably S. Kyushu, intraslab seismicity exists only in themantle (Figure 3). Beneath Kanto, there are thrust zoneactivities at the plate boundary down to 50–70 km depth,and beneath them a double seismic zone exists (Figures 3aand 3b). Because no crustal later phase is observed for theearthquakes in the double seismic zone [Hori, 1990], theyare occurring in the slab mantle. This peculiar doubleseismic zone beneath Kanto is likely to be produced bythe subduction of the serpentinized mantle wedge of the Izu-Bonin fore-arc [Seno et al., 2001]. Thus there is noseismicity within the subducting crust beneath Kanto, asshown in Figures 3a and 3b, except for a minor activity at50–70 km depth, shaded by yellow in Figure 3a. BeneathE. Shikoku, it has been known that intraslab seismicityoccurs in the uppermost mantle, not in the crust (Figure 3c).Beneath S. Kyushu, intermediate-depth earthquakes deeperthan 100 km are down-dip tensional, except for one down-dip compressional event located above them (Figure 3d),indicating that the down-dip tensional earthquakes are

corresponding to the lower plane activity of the doubleseismic zone and thus occurring within the slab mantle. At ashallower depth, there is a thrust zone down to 70 km depthwith some activities beneath it [Goto, 2001]. It has not yetbeen known whether they are occurring in the subductedcrust or not.

Figure 1. Morphology of the Philippine Sea [Okino et al.,1999] and distribution of LFT [pink, Obara, 2002]. Lines a,b, c, and d correspond to the sections in Figures 3a, 3b, 3c,and 3d, respectively. Line t corresponds to the section inFigure 4.

Figure 2. Fault planes with recurrence intervals (yrs) ofgreat earthquakes, distribution of LFT and contours of theupper surface of intraslab seismicity along the Nankai–Sagami Troughs. The fault planes are those of the Taishoand Genroku Kanto earthquakes [Kasahara et al., 1973],the Tokai earthquake [Ishibashi, 1981] and the 1944Tonankai and 1946 Nankai earthquakes [Ando, 1975]. Atthe time of the Genroku earthquake, the Taisho fault planealso ruptured. Rupture zones of large earthquakes inHyuganada are from Utsu [1974]. Contours of the uppersurface of the intraslab seismicity (blue lines) are fromNoguchi and Sekiguchi [2001], Nakamura et al. [1997], andGoto [2001]. LFT (light brown) is from Obara [2002]. Thered triangles indicate active volcanoes.

Figure 3. Seismicity associated with the subduction of thePhilippine Sea plate in cross-sections along lines a, b, c, andd in Figure 1. Types of focal mechanisms are shown bydifferent colors except for section c: Green, thrust; blue,down-dip compression; red, down-dip tension; yellow,normal fault. (a), (b) Two sections in Kanto [Hori, 1997],(c) section in E. Shikoku [Kurashimo et al., 2002] and(d) section in S. Kyushu [Hayashimoto et al., 2001].

SDE 8 - 2 SENO AND YAMASAKI: TREMORS AND EARTHQUAKES IN SOUTHWEST JAPAN

[8] Beneath north of the Izu collision zone, there is nointraslab seismicity (Figure 2). The vacancy in the intraslabseismicity north of Izu does not necessarily mean that thereis no slab. If there is no slab, the difference in slab lengthbetween the Izu collision zone and the Tokai or Kantoregion amounts to 200 km (Figure 2) and is difficult to becompensated by deformation of PHS south of the collisionzone. Seismic tomography studies also revealed a slab-likefeature with a high P-wave velocity north of Izu at depth[Sekiguchi, 2001]. Therefore we believe that a slab is likelyto exist north of Izu, but is aseismic.[9] After all, almost no earthquake is occurring in the

subducted crust where LFT are not seen. However, earth-quakes are occurring in the mantle part of the slab in theseplaces, except for the Izu collision zone.

4. Interplate Earthquakes Along theNankai––Sagami Trough

[10] Along most part of the Nankai Trough, greatinterplate earthquakes have recurred every 100–200 yearshistorically (Figure 2). We note that the regions withoutLFT are also the places where great interplate earthquakeshave not occurred regularly. In fact, no historical great orlarge interplate earthquakes are known in the Izu collisionzone. Southeast off Kyushu, i.e., in Hyuganada, no greatearthquake has been known but M�7 earthquakes haverecurred intermittently [Utsu, 1974]. At the SagamiTrough, the recurrence intervals of great earthquakes are400�1500 years [Matsuda et al., 1978; Kayane andYoshikawa, 1986], much larger than those along the NankaiTrough.

5. Dehydration Embrittlement Model

[11] Accepting the dehydration embritllement hypothesisfor intraslab seismicity shallower than 250 km, it is groupedinto those representing dehydration of hydrated basalticoceanic crust, and those representing dehydration ofserpentinized mantle. Spatial distribution of intraslabseismicity can be predicted by identifying dehydration lociusing phase diagrams of hydrous minerals in associationwith a temperature calculation of the slab [Hacker et al.,2003; Yamasaki and Seno, 2003].[12] Mostly along the Nankai Trough, normal oceanic

crust of the Shikoku Basin is subducting (Figure 1); intra-slab seismicity associated with this subduction generallyforms a single zone, representing dehydration of thehydrated oceanic crust at 20–50 km depth [Peacock andWang, 1999; Seno et al., 2001; Hacker et al., 2003;Yamasaki and Seno, 2003]. Figure 4 shows an example ofsuch seismicity [Matsumura, 1997] and dehydration loci ofhydrated basalts in the cross-section at the eastern end of theNankai Trough, i.e., the Suruga Trough-Tokai region (SeeFigure 1 for location), where a great earthquake is expectedin the near future [Ishibashi, 1981].[13] The temperatures of the slab and mantle wedge are

calculated by the finite element method solving the New-tonian fluid motion-energy equations (See Yamasaki andSeno [2003] for details of the method). The slab geometry isfixed such that the slab earthquakes, of which focal mech-anism solutions are characterized by lateral T-axes along thearc [Matsumura, 1997], are placed just below the slab

surface. The age of the plate subducting at the trench ischanged from 10 Ma at the initiation of subduction to 25 Maat present. We indicate the part of the subducting crustwhere dehydration is expected by the green line.[14] The water dehydrated from the crust probably causes

serpentinization of the mantle wedge as indicated by thehigh Poisson’s ratio. LFT are located at the landwardupper edge of the serpentinized mantle, suggesting thathydro-fracturing is a possible mechanism for LFT andserpentinization. The rupture zone of the expected Tokaiearthquake is extending roughly to the 350�C isotherm atthe plate interface, and dehydration in the subducting crustbeneath the thrust zone would play an important role to raisethe pore fluid pressure at the thrust zone for possibleoccurrence of a future shock.[15] Beneath N. Izu, Kanto, E. Shikoku, and S. Kyushu,

island-arc type crust is subducting, and LFT and intraslabseismicity within the subducted crust are lacking. Based onthis, we propose that subducted island-arc type crust doesnot involve dehydration much since its intermediate layer istonalitic, lacking the quantity of hydrous minerals seen innormal subducting oceanic crust.[16] We infer that no or less dehydration in the subducted

slab does not favor raising the pore fluid pressure at thethrust zone, resulting in no or infrequent occurrence of greatinterplate earthquakes. In N. Izu, where the Izu-Bonin Ridgeis subducting, there is no intraslab seismicity and then nodehydration in the slab is expected. Consistently, there is nogreat or large interplate earthquake.[17] In Kanto and S. Kyushu, although there is almost

no seismicity in the subducted crust, intraslab seismicityin the mantle part exists. This indicates that dehydrationof the serpentinized mantle occurs beneath these regions(Figures 3a, 3b, and 3d; See Seno et al. [2001] forhydration mechanism of the mantle in these regions). Thiswould help, to some extent, to raise the pore fluid pressureat the thrust zone as seen in the intermittent occurrence ofgreat or large earthquakes in these regions.[18] In E. Shikoku, similarly, intraslab seismicity exists

only in the mantle. This place is, however, located in the

Figure 4. Seismicity and temperatures in a cross-sectionperpendicular to the trough axis in the Tokai region at theeastern end of the Nankai Trough (Suruga Trough). Thesection is along line t in Figure 1. The dotted line indicates350�C isotherm. The seismicity is from Matsumura [1997].The dehydration loci in the subducted crust are marked bythe green line. The distribution of LFT [Obara, 2002] isshaded by pink. The high Poisson’s ratio region [Kamiyaand Kobayashi, 2003] is shaded by blue. The Moho depth(30 km) is from Iidaka et al. [2003].

SENO AND YAMASAKI: TREMORS AND EARTHQUAKES IN SOUTHWEST JAPAN SDE 8 - 3

rupture zone of great earthquakes repeating every 100–200 years (Figure 2). Seismic tomography studies show alow velocity zone in the mantle wedge just above the slab inthis region [Zhao et al., 2000], which suggests significantdehydration from the slab. The large amount of dehydrationof the slab mantle might result in the regular occurrence ofgreat earthquakes in this region.

6. Conclusions

[19] LFT have been recently found in the SouthwestJapan fore-arc beneath which the PHS subducts. However,there are some places without such LFT; they are N. Izu-Kanto, E. Shikoku, and S. Kyushu, where island-arc typecrust is subducting. We propose that dehydration of thesubducted crust does not occur beneath these regions,because it is composed mainly of the intermediate-depthtonalitic layer, lacking the quantity of hydrous minerals seenin normal subducting oceanic crust. As expected from thedehydration embrittlement hypothesis, almost no intraslabearthquake occurs within the subducted crust in theseregions. The lesser extent of dehydration of the slab wouldalso reduce the number of occurrence of large interplateearthquakes, because pore fluid pressure is more difficult torise at the thrust zone. This could be reflected in theirregular or no occurrence of great interplate earthquakesin these regions.

[20] Acknowledgments. We thank Kelin Wang for critical review ofthe earlier version of the manuscript, A. Ferris and an anonymous reviewerfor their critical review of the manuscript, and Kazu Obara, Kyoko Okino,Shin Kamiya and Kazuhiko Goto for sending unpublished materials.

ReferencesAndo, M., Source mechanisms and tectonic significance of historical earth-quakes along the Nankai Trough, Japan, Tectonophysics, 27, 119–140,1975.

Fleitout, L., The sources of lithospheric tectonic stresses, Phil. Trans.R. Soc. Lond. A, 337, 73–81, 1991.

Goto, K., Intermediate-depth earthquakes and seismotectonics, Abstractsof Research Project, Grant-in-Aid for Scientific Research, Ministryof Education, Culture, Sports, Science and Technology, 1–139, 2001(in Japanese).

Hacker, B. R., S. M. Peacock, G. A. Abers, and S. D. Holloway, Subductionfactory-2. Are intermediate-depth earthquakes in subducting slabs linkedto metamorphic dehydration reactions?, J. Geophys. Res., 108(B1),doi:10.1029/2001JB001129, 2003.

Hayashimoto, N., H. Yakiwara, T. Kakuta, and K. Goto, Relocation ofintermediate-depth earthquakes beneath southern Kyushu, Japan, Abstr.Seism. Soc. Jpn., B68, 2001 (in Japanese).

Hickman, S., R. Sibson, and R. Bruhn, Introduction to special section:Mechanical involvement of fluids in faulting, J. Geophys. Res., 100,12,831–12,840, 1995.

Hori, S., Seismic waves guided by untransformed oceanic crust subductinginto the mantle: The case of the Kanto district, central Japan, Tectono-physics, 176, 355–376, 1990.

Hori, S., Earthquake mechanism within the Philippine Sea plate subductingbeneath the Kanto district, central Japan, Jisin, 50, 203–213, 1997 (inJapanese).

Hyndman, R. D., M. Yamano, and D. A. Oleskevich, The seismogenic zoneof subduction thrust faults, The Island Arc, 6, 244–260, 1997.

Iidaka, T., T. Iwasaki, T. Takeda, T. Moriya, I. Kumakawa, E. Kurashimo,T. Kawamura, F. Yamasaki, K. Koike, and G. Aoki, Configuration ofsubducting Philippine Sea plate and crustal structure in the central Japanregion, Geophys. Res. Lett., 30(5), 1219, doi:10.1029/2002GL016517,2003.

Ishibashi, K., Specification of a soon-to-occur seismic faulting in the Tokaidistrict, central Japan, based upon seismotectonics, Earthquake Predici-ton -An International Review, Maurice Ewing Series, 4, 297–332, 1981.

Kamiya, S., and Y. Kobayashi, Seismological evidence for the existenceof serpentinized wedge mantle in the Tokai district, to be submitted,2003.

Kasahara, K., J. Yamada, and M. Ando, Crustal movements in the southernKanto district, and a related working hypothesis, Publications for the50th anniversary of the Great Kanto earthquake, 1923, EarthquakeResearch Institute, University of Tokyo, 103–116, 1973 (in Japanese).

Kayane, H., and T. Yoshikawa, Comparative study between present andemergent erosional landforms on the southeast coast of Boso peninsula,central Japan, Geogr. Review Jpn, 59, 18–36, 1986 (in Japanese).

Kirby, S., E. R. Engdhal, and R. Denlinger, Intermediate-depth intraslabearthquakes and arc volcanism as physical expressions of crustal anduppermost mantle metamorphism in subducting slabs (Overview), inSubduction Top to Bottom, edited by G. E. Bebout, D. W. Scholl, S. H.Kirby, and J. P. Platt, Geophys. Monogr., AGU, Washington D.C., 96,195–214, 1996.

Kurashimo, E., M. Tokunaga, N. Hirata, T. Iwasaki, S. Kodaira, Y. Kaneda,K. Ito, R. Nishida, S. Kimura, and T. Ikawa, Geometry of the subductingPhilippine Sea plate and the crustal and upper mantle structure beneathEastern Shikoku Island revealed by seismic refraction/wide-angle reflec-tion profiling, Jisin., 54, 489–505, 2002 (in Japanese).

Li, K., M. Shinohara, K. Suyehiro, E. Kurashimo, S. Miura, andH. Nishisaka, Crustal structure of north Kyushu-Palau Ridge by oceanbottom seismographic observation, Abstr. Seism. Soc. Jpn., 2, 38, 1997 (inJapanese).

Matsuda, T., Y. Ota, M. Ando, and N. Yonekura, Fault mechanism andrecurrence time of major earthquakes in southern Kanto district, Japan,as deduced from coastal terrace data, Geol. Soc. Am. Bull., 89, 1610–1618, 1978.

Matsumura, S., Focal zone of a future Tokai earthquake inferred from theseismicity pattern around the plate interface, Tectonophysics, 273, 271–291, 1997.

Nakamura, M., H. Watanabe, T. Konomi, S. Kimura, and K. Miura,Characteristic activities of subcrustal earthquakes along the outer zoneof southwestern Japan, Ann. Disast. Prev. Res. Inst., Kyoto Univ., 40,1–20, 1997 (in Japanese).

Noguchi, S., and S. Sekiguchi, Subduction and deformation of the Philip-pine Sea and Pacific plates beneath the Kanto region, Earth Month., 23,733–741, 2001 (in Japanese).

Obara, K., Nonvolcanic deep tremor associated with subdution in southwestJapan, Science, 296, 1679–1681, 2002.

Okino, K., Y. Ohara, S. Kasuga, and Y. Kato, The Philippine Sea: Newsurvey results reveal the structure and the history of the marginal basins,Geophys. Res. Lett., 26, 2287–2290, 1999.

Peacock, S. M., and K. Wang, Seismic consequences of warm versus coolsubduction metamorphism: Examples from southwest and northeastJapan, Science, 286, 937–939, 1999.

Raleigh, C. B., and M. S. Paterson, Experimental deformation of serpen-tinite and its tectonic implications, J. Geophys. Res., 70, 3965–3985,1965.

Sato, H., S. Machida, S. Kanayama, H. Taniguchi, and T. Ishii, Ceochem-ical and isotopic characteristics of the Kinan Seamount Chain in theShikoku Basin, Geochem. J., 36, 519–526, 2002.

Sekiguchi, S., A new configuration and an aseismic slab of the descendingPhilippine Sea plate revealed by seismic tomography, Tectonophysics,341, 19–32, 2001.

Seno, T., D. Zhao, Y. Kobayashi, and M. Nakamura, Dehydration inserpentinized slab mantle: Seismic evidence from southwest Japan, EarthPlanets Space, 53, 861–871, 2001.

Taira, A., S. Saito, K. Aoike, S. Morita, H. Tokuyama, K. Suyehiro,N. Takahashi, M. Shinohara, S. Kiyokawa, J. Naka, and A. Klaus, Natureand growth rate of the Northern Izu-Bonin (Ogasawara) arc crust andtheir implications for continental crust formation, The Island Arc, 7,395–407, 1998.

Utsu, T., Space-time pattern of large earthquakes occurring off the Pacificcoast of the Japanese islands, J. Phys. Earth, 22, 325–342, 1974.

Yamasaki, T., and T. Seno, Double seismic zones and dehydrationembrittlement, J. Geophys. Res., 108(B4), 2212, doi:10.1029/2002JB001918, 2003.

Zhao, D., F. Ochi, A. Hasegawa, and A. Yamamoto, Evidence for thelocation and cause of large crustal earthquakes in Japan, J. Geophys.Res., 105, 13,579–13,594, 2000.

�����������������������T. Seno, Earthquake Research Institute, University of Tokyo, Japan.

(seno@eri.u-tokyo.ac.jp)T. Yamasaki, National Institute of Polar Research, Japan.

SDE 8 - 4 SENO AND YAMASAKI: TREMORS AND EARTHQUAKES IN SOUTHWEST JAPAN