abstract - usgs...depths of 20{40km in the japan islands. the jma has distinguished those deep...

Proceeding of the UJNR 5th Earthquake Research Panel Meeting, 2004 1

Low-frequency tremor and slow slip around the probable source region of the Tokai Earthquake– A new indicator for the Tokai Earthquake prediction provided

by unified seismic networks in Japan –

Noriko Kamaya (Seismological and Volcanological Department, JMA),Akio Katsumata (Meteorological College, JMA)

andYuzo Ishigaki ( Seismological and Volcanological Department, JMA)

Abstract

Enhancement of seismic networks in the Japan Islands of these years revealed occurrence of deeplow-frequency continuous tremors of a belt distribution in the southwest Japan, where the subductingPhilippine Sea plate reaches depths of 25–40 km. The source region of the tremor is assumed tocorrespond to boundary region between oceanic and island crusts close to mantle wedge. The eastend of the belt tremor distribution is adjacent to the probable source region of the Tokai Earthquake.A slow-slip event has continued in the west of the source region of the Tokai Earthquake since 2000.Synchronous activation of the tremor and slow-slip speed was observed in 2003. This suggests somerelationship between the slow-slip and the tremor activity. We expect that the tremor activity couldbe a new valuable indicator for the Tokai Earthquake prediction. The long duration of the tremorindicates existence of fluid relating to its generation. The most probable fluid is considered to be waterproduced by dehydration of chlorite and formation of clinopyroxene in the oceanic crust on the basisof data from high pressure and temperature experiments. The northern border of the belt distributionis possibly rimmed by the edge of the mantle wedge because serpentine formation absorbs fluid water.

1 Introduction

Since Oct. 1997, short-period seismic records ob-tained at most of stations in the Japan Islandshave been transmitted to the Japan Meteorologi-cal Agency (JMA). The JMA has processed thoserecords to make a comprehensive seismic catalogin Japan. The short-period instruments includethose installed by universities, the National Re-search Institute for Earth Science and DisasterPrevention (NIED), the JMA, other national in-stitutes, and local governments. This integratedprocessing lowered the detectable magnitude ofearthquakes. Moreover, deployment of the Hi-net[e.g., Okada et al., 2000], which is a dense net-work of highly sensitive short-period instrumentsinstalled by the NIED, improved the detectivityremarkably. Hi-net data has been included to pro-duce the seismic catalog since Oct. 2000.

Fig. 1 shows stations which contribute the cur-rent unified seismic catalog. Seven cable-typeocean-bottom seismometry systems are includedin the network.

Fig. 2 shows magnitude-time distribution since

1965. The magnitudes are determined from maxi-mum displacement or velocity amplitudes [Tsuboi

, 1954; Katsumata, 2004; Funasaki et al., 2004]. Itis recognized that the lowermost magnitude hascontinued to decrease year by year since 1970s.More than 130,000 hypocenters are located in andaround the Japanese Islands in 2003. Fig. 3 showsnumbers of hypocenter determinations in a year inthe JMA (1926–Sep. 1997) and the unified (Oct.1997–) seismic catalogs. A level of one thousand ayear had continued until mid of 1970s. Since then,introductions of high-sensitive seismometers, au-tomated digital processing systems and denselydistributed seismic networks have increased thenumber of located hypocenters.

The enhanced detection ability revealed oc-currence of low-frequency microearthquakes atdepths of 20–40km in the Japan Islands. TheJMA has distinguished those deep low-frequencyearthquakes in the seismic catalog since Oct.1999. The deep low-frequency earthquakes are in-dicated by red symbols in Fig. 2. Magnitude ofmost of those events are less than magnitude 2.0,and many of those are less magnitude 1.0. Occur-


2 5 N

3 0 N

3 5 N

4 0 N

4 5 N

1 2 5 E 1 3 0 E 1 3 5 E 1 4 0 E 1 4 5 E

0 5 0 0 1 0 0 0 km

JMA N I ED Un i v e r s i t i e sGS J JAMS TECO t h e r s

Figure 1: Stations which contribute the current unified seismic catalog.


rence of low-frequency events around the Mohodiscontinuity in the areas away from volcanoesbecame recognized based on the seismic catalog[Nishide et al., 2000]. Although seismicity of low-frequency events beneath volcanoes has been well-known [e.g., Hasegawa et al., 1991; Hasegawa and

Yamamoto, 1994; Ukawa and Obara, 1993], seis-mic activities away from volcanic areas had notbeen noticed.

Nishide et al. (2000) recognized low-frequencyseismic activities in the east of Aichi Prefecture(34.8◦ ∼ 35.3◦N, 137◦ ∼ 138◦E. See Fig. 8),which was recognized as a part of belt distributionof deep low-frequency tremor along the strike ofthe subducting Philippine Sea plate [Obara, 2002;Katsumata and Kamaya, 2003] after deploymentof the Hi-net.

Figure 2: Magnitude-time distribution since 1965in the JMA and the unified seismic catalogs. Redsmall circles denote the events labeled as the deep low-frequency earthquakes.

The deep low-frequency tremor (LFT in this ar-ticle) is the most notable phenomena which wasfound by the recently deployed dense seismic net-

1 0 0

1 0 1

1 0 2

1 0 3

1 0 4

1 0 5

1 0 6

Nu

mb

er

of

Ea

rt

hq

ua

ke

s

1 9 3 0

1 9 4 0

1 9 5 0

1 9 6 0

1 9 7 0

1 9 8 0

1 9 9 0

2 0 0 0

Y e a r

Figure 3: Number of earthquakes listed in the JMA(1926–Sep. 1997) and the unified (Oct. 1997–) seismiccatalogs.

work. We describe the activity of the deep tremor,and discuss its source characteristics in this arti-cle. In addition to the deep tremor, low-frequencyearthquakes around the Moho discontinuity werefound not only in volcanic regions but also innon-volcanic areas. We use a term of deep low-frequency earthquake (LFE in this article) to de-scribe such activities.

2 Seismicity of Deep Low-

Frequency Tremors and

Earthquakes

In this section, we describe spacial and temporalcharacteristics of the LFT in the southwest Japan.Activities of LFE in volcanic and non-volcanic re-gions are also explained.

2.1 Source Locations of Deep Low-

Frequency Tremors in the South-

west Japan

The JMA estimates the source locations of LFTby an ordinary hypocenter determination methodbased on the onset times of enlarging parts ofthe tremors. This is different from the method ofObara (2002) who estimated the locations of LFTby applying S-wave velocity model to envelopesof tremor waveforms. Fig. 4 shows examples ofthe observed tremor waveforms. Broken curves inthe figure denote calculated arrival times basedon locations estimated by the JMA. Onsets of thetremor are close to the calculated S travel times.Corresponding onsets of P-waves are seldom seeneven on the vertical component. Lack of P-wavearrival data makes accuracy of source depths poor.

Although most of LFT do not show onsets of P-wave, a small number of LFT segments do showonsets of P-wave. Depth of those events are ratherprecise. Fig. 5 shows an epicenter map and a crosssection of the deep low-frequency events with rel-atively small estimation errors in the southwestJapan. In the figure, orange circles show eventsoccurring within 10 km (epicentral distance) fromthe centers of volcanoes which have been active inthe Quaternary [Committee for Catalog of Quater-

nary Volcanoes in Japan, 1999], and blue circlesshow events away from volcanoes. The estimateddepths of the most of non-volcanic events were dis-tributed from 25 to 40 km. and the average depthis about 30 km.


N / S

0

1 0

2 0

3 0

4 0

5 0

Epic

entr

al D

ista

nce

(km

)0 8 : 4 0 : 0 0 0 8 : 4 0 : 3 0 0 8 : 4 1 : 0 0 0 8 : 4 1 : 3 0 ( JS T )

T i me

P SP SP S

LG . HRN

ND . ACH

ND . HTNND . KGW

ND . KSH

ND .MSK

ND . SMY

ND . TNR

ND . TOE

NU . S TN

TAK I SA

YASUOK

Figure 4: Example of LFT observed on June 3, 2001. Waveforms are arranged in epicentral distance orderfrom an estimated source location. Broken lines show calculated P and S travel times for enlarging parts of thetremor. Station code are denoted on the right side. ND, NU, and LG signify the NIED, Nagoya University, andlocal government, respectively. YASUOK and TAKASA are stations of the JMA.

Fig. 6 show source locations of LFT (blue cir-cles) from September, 1999 to September, 2001beneath eastern Shikoku (133.6◦ – 134.5◦E) pro-jected onto the vertical structural cross-sectionfrom Kurashimo et al. (2002), which is based onseismic velocities. The LFT are located in bound-ary region between oceanic and island crusts closeto the mantle wedge. It is notable that theLFT are not distributed boundary regions be-tween oceanic crust and mantle wedge.

2.2 Temporal Characteristics of the

Deep Low-Frequency Tremor Ac-

tivities

Fig. 7 (b) shows time-space distribution of LFT.Most event show episodic activity. Active periodcontinues for several hours – several days.

Fig. 7 (c) and (d) show accumulated numbersof events and quiescent periods of the tremor ac-tivities in partitioned areas shown in (b). Thequiescent period in the figure is defined as a timeinterval without any detected tremor events morethan two days.

It is recognized from Fig. 7 (b) and (c) that thetremor activities are not stationary. Accumulatednumbers and quiescent periods changes with sometrends.

In the region (2) (Fig. 7 (b)), shortening ofquiescent periods started late in 2002. Short qui-

escent periods have been kept for about half yearfrom the 4th quarter of 2003 to the 1st quarter of2004. Then quiescent periods were prolonged in2004, but they are still short compared with theaverage in 2002. The period of 2003–2004 cor-responds to periods of relatively fast slow-slip inTokai area, which is discussed in a latter section.

In the region (4), short quiescent periods havebeen kept since middle of 2004. It seems thatthere has been some activity in the region.

In the regions (8), (9) and (10), shortening ofquiescent periods appear relatively regularly. Ittakes about a few months from starting of short-ening to returning to regular quiescent period ofabout several tens of days.

In the regions (11) and (12), state of shortquiescent periods was kept for a few months in2003. The shortening started from the beginningof 2003, and prolonging had continued until be-ginning of 2004. Slow-slip events were observed inthe period by Hirose and Obara (2003) in the re-gion. The graph of accumulated number of eventsalso show a unusual rate in the period.

The activity of LFT should indicate some stateof plate coupling. But its mechanism is to be in-vestigated.


3 6 N

1 3 2 E0 5 0 1 0 0 km W

E

2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3 2 0 0 4

1

2

3

4

5

6

7

8

9

1 0

1 1

1 2

1 9 9 9 / 0 9 / 0 1 - 2 0 0 4 / 0 8 / 2 3

1 0 81

1 1 2 22

3 13

3 8 74

2 2 15

2 8 16

77

3 4 38

5 6 19

8 0 91 0

6 7 21 1

3 0 41 2

2 0 0 0

2 0 0 1

2 0 0 2

2 0 0 3

2 0 0 4

Ac

cum

ul

at

ed

Num

be

r

Qu

ie

sc

en

t P

er

iod

(d

ay

)

1 0 01 0 11 0 2

1

1 0 01 0 11 0 2

2

1 0 01 0 11 0 2

3

1 0 01 0 11 0 2

4

1 0 01 0 11 0 2

5

1 0 01 0 11 0 2

6

1 0 01 0 11 0 2

7

1 0 01 0 11 0 2

8

1 0 01 0 11 0 2

9

1 0 01 0 11 0 2

1 0

1 0 01 0 11 0 2

1 1

1 0 01 0 11 0 2

1 2

2 0 0 0

2 0 0 1

2 0 0 2

2 0 0 3

2 0 0 4

(a) (b)

(c) (d)

Figure 7: Temporal variation of the deep low-frequency tremor activities. (a) Epicenter map; (b) time-spacedistribution; (c) accumulated number; (d) quiescent period. Accumulated number of events and quiescent periodare shown for regions of which numbers are denoted on the right of (b).


3 0 N

1 3 5 E

0 2 0 0 km

1 9 9 9 / 0 1 / 0 1 - 2 0 0 4 / 0 8 / 2 3

0

1 0

2 0

3 0

4 0

5 0( km)

Kii

Chan.

Kii

Penin.A

B

ShikokuKyushu

Honshu

Figure 5: Distribution of LFT and LFE in the south-west Japan (Jan. 1999 – Aug. 2004). Blue circles,orange ones and red triangles show source locations ofnon-volcanic LFE and LFT, LFE close to volcanoes,and active volcanoes in the Quaternary, respectively.The line segments A and B indicate Tonankai andNankaido profiles for which Hyndman et al. (1995)estimated geotherms, respectively.

2.3 Deep Low-Frequency Earthquakes

in the Northeast Japan

Deep low-frequency earthquakes have been foundin wide area in the Japan Islands [Kamaya and

Katsumata, 2004]. Fig. 8 shows epicenters of theLFE. Orange circles show LFE occurring within10 km (epicentral distance) from the centers ofvolcanoes which have been active in the Quater-nary [Committee for Catalog of Quaternary Vol-

canoes in Japan, 1999], and blue circles show LFEaway from volcanoes which include LFT. The es-timated depths of the most of non-volcanic eventswere distributed from 25 to 40 km, and the aver-age depth is about 30 km. Active volcanoes andthe Quaternary volcanoes are shown by red tri-angles. In the figure, contours show the tops ofthe Pacific and the Philippine Sea plates whichwere estimated as upper limits of seismicity in theplate.

LFE in the northeast Japan are distributed onlyin the back-arc side of the volcanic front which isalmost coincides with a 110 km depth contour ofthe Pacific plate. There is no activities in thefore-arc side of the volcanic front. The distribu-tion indicates the LFE in the northeast Japan are

Figure 6: Source locations of LFT (blue circles) fromSeptember, 1999 to September, 2001 beneath easternShikoku (133.6◦ – 134.5◦E) projected onto the verticalstructural cross-section from Kurashimo et al. (2002),which is based on seismic velocities. The horizontalaxis represents the distance from shot J2 of Kurashimo,

et al. (2002) which was located near the south coastof Shikoku Island. Black dots denote ordinary earth-quakes.

related to volcanic activity even in the cases awayfrom volcanoes. If we include regions with igneousrocks in the Tertiaries, all LFE-active regions inthe northwest Japan are covered.

2.4 Deep Low-Frequency Earthquakes

in the Southwest Japan

In addition to deep low-frequency tremor activi-ties, there are some spot distributions of LFE inthe southwest Japan (gray-filled circles in Fig. 7(a)). Some of those clusters are distributed alongthe volcanic front of the Philippine Sea plate. Butsome clusters are distributed in fore-arc side of thevolcanic front (clusters in 133◦–136◦E and 35.1◦–35.7◦N).

Fig. 9 (a) and (d) show waveform and spec-trum of a horizontal (NS) component from a low-frequency earthquake obtained at a station inTannan (DP.TNJ, 35.0313◦N, 135.2137◦E) for anevent which occurred in Kyoto Pref. (hypocenter:35.1185◦N, 135.5775◦E, 34.7km) at 01:41:41.1,July 30 2000 JST. Waveforms of events in thisregion usually show a clear onsets and long du-ration. This is similar to waveforms of LFE involcanic regions.

The bold spectrum curve corresponds to the 30swaveform segment indicated by the bold line inthe waveform. The thin spectrum curve corre-sponds to the noise level as calculated for the 30s


3 5 N

4 0 N

4 5 N

1 3 0 E 1 3 5 E 1 4 0 E 1 4 5 E

0 5 0 0 km

1 9 9 9 / 0 1 / 0 1 - 2 0 0 4 / 0 8 / 2 3

5 01 0 0

1 5 02 0 0

2 5 0

2 5 0

3 0 0

2 0

2 0

2 0

3 0

3 0

4 0

4 0

6 0

6 0

Figure 8: Distribution of epicenters of low-frequency events near the Mohorovicic discontinuity under theJapan Islands as determined by the Japan Meteorological Agency (JMA) from September, 1999 to August 23,2004. Orange circles show events within 10km of active volcanoes and volcanoes which have been active in theQuaternary [Committee for catalog of Quaternary volcanoes in Japan, 1999], and blue circles show events awayfrom those volcanoes. Active volcanoes and the Quaternary volcanoes are shown by red triangles. Also shownare parts of depth contours for the upper surfaces of the Pacific Plate and the Philippine Sea Plate. The contoursare labeled with depth in km.


(a)

(d)2 0 0 0 / 0 7 / 3 0 0 1 : 4 1 H : 3 4 km MI D KYOTO PREF

DP . TN J NS DLT= 3 4 . 1 km

1 0 - 1 1 0 0 1 0 1 1 0 2

F r e q u e n c y ( Hz )

1 0 - 9

1 0 - 8

Sp

ec

tr

al

Am

pl

it

ud

e

1 0 - 1 1 0 0 1 0 1 1 0 2


1 0 - 9

1 0 - 8

Sp

ec

tr

al

Am

pl

it

ud

e

2 0 0 0 / 0 7 / 3 0 0 1 : 4 1 H : 3 4 km MI D KYOTO PREF

- 3 0 - 2 0 - 1 0 0 1 0 2 0 3 0 4 0 ( s )

0

2 . 8

m / s

DP . TN J NS DLT= 3 4 . 1 km

P S

(b)

(e)

(f)

1 0 - 1 1 0 0 1 0 1 1 0 2


1 0 - 1 0

1 0 - 9

1 0 - 8

Sp

ec

tr

al

Am

pl

it

ud

e

1 0 - 1 1 0 0 1 0 1 1 0 2


1 0 - 1 0

1 0 - 9

1 0 - 8

Sp

ec

tr

al

Am

pl

it

ud

e2 0 0 1 / 0 6 / 0 3 0 8 : 4 0 H : 3 2 km NE A I CH I PREF

NU . S TN NS DLT= 4 . 6 km

2 0 0 1 / 0 6 / 0 3 0 8 : 4 0 H : 3 2 km NE A I CH I PREF

- 3 0 - 2 0 - 1 0 0 1 0 2 0 3 0 4 0 ( s )

0

5 8 0 . 3

nm / s

NU . S TN NS DLT= 4 . 6 km

P S

1 9 9 9 / 1 0 / 1 0 1 3 : 4 6 H : 3 0 km NORTHERN AK I TA PREF

- 3 0 - 2 0 - 1 0 0 1 0 2 0 3 0 4 0 ( s )

0

1 . 5

m / s

TU . GJM NS DLT= 2 1 . 6 km

P S

1 9 9 9 / 1 0 / 1 0 1 3 : 4 6 H : 3 0 km NORTHERN AK I TA PREF

TU . GJM NS DLT= 2 1 . 6 km

1 0 - 1 1 0 0 1 0 1 1 0 2


1 0 - 9

1 0 - 8

1 0 - 7

Sp

ec

tr

al

Am

pl

it

ud

e

1 0 - 1 1 0 0 1 0 1 1 0 2


1 0 - 9

1 0 - 8

1 0 - 7

Sp

ec

tr

al

Am

pl

it

ud

e

(c)

Figure 9: Spectra of deep low-frequency events in the Japan Islands. a) and d) A waveform and its spectrum ofhorizontal (NS) component from a low-frequency earthquake obtained at a station in Tannan (DP.TNJ, 35.0313N,135.2137E) for an event which occurred in Kyoto Pref. at 01:41:41.1, July 30 2000. The bold spectrum curvecorresponds to the 30s waveform segment indicated by the bold line in the waveform. The thin spectrum curvecorresponds to the noise level as calculated for the 30s waveform segment indicated by the thin line in thewaveform. H and DLT denote the focal depth and epicentral distance. b), c), e) and f) show waveforms andspectra for a low-frequency tremor in southwestern Japan and a low-frequency earthquake away from volcanoesin northeastern Japan, respectively. The records were obtained at Shin-Toyone (NU.STN, 35.1355N, 137.7437E)and Gojome (TU.GJM, 39.9520N, 140.1160E).


waveform segment indicated by the thin line in thewaveform. The seismic wave from the event haspower in the frequency rage of 1–10 Hz. (b), (c),(e) and (f) in Fig. 9 show waveforms and spectrafor a low-frequency tremor in southwestern Japanand a low-frequency earthquake away from vol-canoes in northeastern Japan, respectively. Thespectra of these events do not show no effectivespectral contents in a frequency range lower than1 Hz. It is interesting that the deep tremor andLFE has common characteristics in spectrum.

3 Tokai Slow-Slip and the

Tremor

A great earthquake is expected to occur alongthe Suruga Trough where the Philippine Sea platesubducts beneath the Eurasian plate. The earth-quake is named Tokai Earthquake prior to its oc-currence. Fig. 10 shows the probable source re-gion of the Tokai Earthquake.

Probable Source Region of the Tokai Earthquake

Su

rug

a Tr

ou

gh

Figure 10: Probable source region of the Tokai Earth-quake and epicenters of deep low-frequency events.

In the west of the probable source area, slow-slip events has been observed since 2000. Fig.11 shows the vertical displacement measured byGEONET, GPS Earth Observation Network ofthe Geographical Survey Institute of Japan. Fig.12 shows temporal variation of crustal deforma-tion observed at Hamakita GEONET station.

Relatively high-speed deformation is seen in2003–2004 in Fig. 12. Active state of the deeptremor has been continuing in the northwest re-gion of the slow-slip area since 2003 (the region 2in Fig. 7). The fast crustal deformation in 2003–2004 is simultaneous with the abnormal tremor ac-tivity since 2003. We suspect that there would besome interactions between the slow-slip and deep

tremor activity. We consider that the deep tremoractivity would be a indicator of some status of thecoupling between the plates.

Figure 11: Vertical component of crustal deformationdue to the Tokai slow-slip observed by the GEONET,GPS Earth Observation Network of the GeographicalSurvey Institute.

Figure 12: Temporal variation of location of theHamakita GEONET station.

4 Source Characteristics of LFT

4.1 Source Force Direction of LFT

It is important to know the source mechanism ofthe tremor to understand this phenomena. S-waveis dominant on the tremor waveforms. It is con-sidered that some kind of shear force causes thetremor.

We estimate source-force direction from particlemotions of S-waves with a method by Ukawa and

Ohtake (1987). We get S-wave polarization direc-tions from particle motions of every two secondswith overlapping of one second as shown in Fig.13. Lower panel in Fig. 13 show particle motionson radial-transverse (R-T), vertical-radial (Z-R)


and vertical-transverse (Z-T) planes. Polarizationdata indicated by circles on the upper-right cor-ners of the chart of particle motions are used toestimate source mechanism.

We estimate source force directions for time-intervals with enough number of polarizationdata. Source force directions were estimated forsingle-force and double-couples models. Fig. 14shows an example of a solution. Red line seg-ments show observed polarization directions, andcolor map show residual distribution. Region ofsmall residual is rather concentrated. This indi-cates that this kind of analysis can bring out in-formations on the tremor source mechanism.

S F T : 1 N_ s t : 1 3

NU . S TNND . TOE

ND . HOU

YASUOK

ND . ASH

ND . ACH

TAK I SA

ND . SMYNU . TYD

ND . KGWND .MKB

ND . HTN

ND . NUK

LOW

A I C : 2 9 . 1 4

3 6 4 9 6 3Re s ( d e g )

DC T : 1 N_ s t : 1 3

NU . S TNND . TOE

ND . HOU

YASUOK

ND . ASH

ND . ACH

TAK I SA

ND . SMYNU . TYD

ND . KGWND .MKB

ND . HTN

ND . NUK

LOW

P

T

A I C : 2 8 . 1 7

3 2 4 4 5 6Re s ( d e g )

Figure 14: Source force directions inferred from par-ticle motions for single-force (SF) and double-couples(DC) models projected on lower hemispheres.

Inferred P/T axes or force directions are shownin Figs. 15 (double-couple mode) and 16 (singleforce model), respectively. In this analysis, P/Taxes are not distinguished because initial motionpolarities are not used. Solutions of single force(Fig. 16) show relatively clear foci than those fordouble couple model. Source force directions ofthe tremor are distributed along NW-SE directionwhich is parallel to the direction of the PhilippineSea plate motion. This indicates that the motionof the Philippine Sea plate related to the tremoractivities.

On the other hand, many of source-force direc-tions of events around Mt. Fuji (138.7◦E, 35.3◦N)are nearly vertical. Those of events in the westernTottori area (around 133.3◦E, 35.3◦N) show simi-lar tendency. Source-force direction of the tremoris clearly different from that of LFE in volcanicareas.

It is also notable that non-volcanic LFE insouthwest Japan around 135◦E, 35◦N show similar

0 2 0 0 km

3 3 N

3 4 N

3 5 N

3 6 N

3 7 N

1 3 2 E 1 3 3 E 1 3 4 E 1 3 5 E 1 3 6 E 1 3 7 E 1 3 8 E 1 3 9 E 1 4 0 E 1 4 1 E

Do u b l e Co u p l e

( Low e r h em i s p h e r e )

Figure 15: Inferred P- or T-axes for a double-couplemodel projected on a lower hemisphere of each region.Regions are shown with yellow rectangles.

0 2 0 0 km

3 3 N

3 4 N

3 5 N

3 6 N

3 7 N

1 3 2 E 1 3 3 E 1 3 4 E 1 3 5 E 1 3 6 E 1 3 7 E 1 3 8 E 1 3 9 E 1 4 0 E 1 4 1 E

S i n g l e F o r c e

( Low e r h em i s p h e r e )

Figure 16: Inferred force direction for a single-forcemodel projected on a lower hemisphere of each region.Regions are shown with yellow rectangles.

distribution as those in volcanic LFE. Althoughthe region is in the fore-arc side of the volcanicfront and in non-volcanic areas, the mechanismof LFE in the region would be similar to that involcanic regions.

4.2 Source Spectrum of the Tremor

It is well-known that LFT show peak spectrumaround 2–3 Hz. Fig. 17 shows an example of spec-trum of the tremor. The spectrum is calculateddirectly from ground velocity record obtained witha broadband instruments (STS-1) without instru-mental response correction. The black bold andblue thin curves show spectra of tremor-activeperiod and period without tremor, respectively.Spectrum level is clearly different in a frequencyrange of 1–10 Hz. But no clear difference is seenin other frequency ranges. In the frequency rangeof 0.1–1 Hz, high-level ground noise due to micro-


2 0 0 1 / 0 6 / 0 3 0 8 : 4 0 H : 3 0 km NE A I CH I PREF YASUOK

0 . 0 5 . 0 1 0 . 0 1 5 . 0 2 0 . 0 2 5 . 0 3 0 . 0 3 5 . 0 4 0 . 0 4 5 . 0

YASUOKNS

D= 2 3 . 9 km A= 3 4 . 3 d e g

S

( P ) ( S )

0 . 0 5 . 0 1 0 . 0 1 5 . 0 2 0 . 0 2 5 . 0 3 0 . 0 3 5 . 0 4 0 . 0 4 5 . 0

EW

S

( P ) ( S )

0 . 0 5 . 0 1 0 . 0 1 5 . 0 2 0 . 0 2 5 . 0 3 0 . 0 3 5 . 0 4 0 . 0 4 5 . 0 1 0 0 ( nm)

UD

S

( P ) ( S )

1

2

3

4

5

6

7

8

9

1 0

1 1

1 2

1 3

1 4

1 5

1 6

1 7

1 8

1 9

2 0

2 1

2 2

2 3

2 4

2 5

2 6

2 7

2 8

2 9

3 0

3 1

3 2

3 3

3 4

3 5

3 6

3 7

1

5 0 ( nm)

T

RR1 3 2 . 6 T

R

Z

T

Z

2

2 0 ( nm)

T

RR1 0 8 . 1 T

R

Z

T

Z

3

2 0 ( nm)

T

RR1 1 6 . 9 T

R

Z

T

Z

4

2 0 ( nm)

T

RR1 2 2 . 7 T

R

Z

T

Z

5

2 0 ( nm)

T

RR1 2 0 . 0 T

R

Z

T

Z

6

2 0 ( nm)

T

RR - 4 2 . 5 T

R

Z

T

Z

7

1 0 ( nm)

T

RR1 3 0 . 3 T

R

Z

T

Z

8

1 0 ( nm)

T

RR - 4 2 . 4 T

R

Z

T

Z

9

1 0 ( nm)

T

RR0 . 7 T

R

Z

T

Z

1 0

1 0 ( nm)

T

RR - 2 6 . 5 T

R

Z

T

Z

Figure 13: Particle motions of the tremor. The upper panel shows waveforms of three components, and thelower panel shows particle motions on radial-transverse (R-T), vertical-radial (Z-R) and vertical-transverse (Z-T)planes of every two second period with overlapping of one second.

seisms is recognized. In the range of 0.01–0.1 Hzin which the ground noise is low enough, power ofthe tremor is not recognized.

Hirose and Obara (2003) found slow-slip eventssimultaneous with deep tremor activities. Iftremor is directly due to a slow-slip event, spec-trum level at low frequency is higher than thatat high frequency based on source spectrum mod-els [e.g., Aki, 1967]. But no power is observedat frequency ranges of 0.01-0.1Hz. This indicatesthat there would be different power source for thetremor from the slow-slip.

5 Fluid Generation in the

Tremor Source Region

The continuous tremors remind us of volcanictremors, which are caused by some kinds of fluid[e.g., Chouet, 1996]. Deformation of ductile mate-rial may also show long duration like tremor.

Since a large quantity of water is conveyedby subduction of the Philippine Sea plate, waterwould be related to the deep tremor, directly or in-

directly. Water would alter physical properties ofminerals, and may cause tremor directly by mov-ing or vibration.

We assume some possible sources of the waterwhich is related to the deep tremor.

1. Extraction of interstitial water from oceaniccrust.

2. Dehydration from minerals in the oceaniccrust.

3. Dehydration from minerals in the mantle ofthe Philippine Sea plate.

4. Dehydration from minerals in the mantlewedge.

Although extraction of interstitial water fromthe oceanic crust might be a possible source, itwould be difficult to explain why the extractionoccurs at the restricted depth range of 25–40km.Since extraction of interstitial water should occurgradually, the limited depth range of water supplyis considered to be related to limited pressure andtemperature condition.


1 0 - 4 1 0 - 3 1 0 - 2 1 0 - 1 1 0 0 1 0 1


1 0 - 1 2

1 0 - 1 1

1 0 - 1 0

1 0 - 9

Sp

ec

tr

al

Am

pl

it

ud

e

(m/

s)

2 0 0 4 / 0 6 / 2 7 1 6 : 0 0 - - 1 7 : 0 0

Figure 17: Spectrum of LFT observed at Asahi sta-tion (Fig. 18) of the F-net installed by the NIED. Theblack bold and blue thin curves show spectra of tremor-active period and period without tremor, respectively.

3 5 N

1 3 7 E 1 3 7 3 0 ’ E 1 3 8 E

A s a h i

Figure 18: Station of Asahi (F-net of the NIED) andepicenters of LFT in June 2004 (red circles).

Fig. 19 shows the phase relationships forwater-saturated mid-ocean ridge basalt obtainedby Schmidt and Poli (1998). This phase relation-ships are applicable to minerals in a oceanic crustin a subducting slab. The broken curves in thefigure show geotherms at the top of the subduct-ing Philippine Sea plate estimated by Hyndman

et al. (1995) for Tonankai and Nankaido profiles,respectively (Fig. 5). Depth contours in Fig. 8were used to convert distance from the trench axisinto depth at the top of the slab. The Philip-pine Sea plate subducts with a steeper incline be-neath Kii Peninsula than beneath Shikoku. Thegeotherms cross the phase boundary shown by theheavy solid curve. Because chlorite can not exist

in the region to the right of line ”A”, water isreleased by dehydration of chlorite along this line.

On the other hand, the temperature profile be-neath Kii Peninsula crosses a thick line ”B” atsimilar depth. Crossing this line, clinopyroxeneappears. Because clinopyroxene is not a hydrousmineral, the total water content of the basalt de-creases.

Nor

thea

ster

n J

apan

Kii Peninsula

Shikoku A

B

Figure 19: Phase relationships for water-saturatedmid-ocean ridge basalt [Schmidt and Poli, 1998] andtemperature profiles for the upper surfaces of the Pa-cific Plate under northeastern Japan [Iwamori, 1998]and the Philippine Sea Plate beneath the Kii Peninsulaand Shikoku [Hyndman et al., 1995] in southwesternJapan: amph = amphibole, chl = chlorite, cld = chlo-ritoid, cpx = jadeitic or omphacitic clinopyroxene, epi= epidote, gar = garnet, law = lawsonite, zo = zoisite.The temperature profile for the Philippine Sea Platebeneath Shikoku crosses a thick line ”A” at a depthof about 40km. On the other hand, the temperatureprofile beneath Kii Peninsula crosses a thick line ”B”at similar depth.

Fig. 20 shows Phase diagram for water-saturated average mantle peridotite, modifiedfrom a figure by Schmidt and Poli (1998), andtemperature profiles for the top of the PacificPlate under northeastern Japan [Iwamori, 1998]and the Philippine Sea Plate beneath the KiiPeninsula and Shikoku [Hyndman et al., 1995] insouthwestern Japan. Serpentine and some otherhydrous minerals including chlorite and amphi-bole in the mantle peridotite also dehydrate under


the conditions of 600–700◦C and 30–50 km. Thisindicates a possibility that hydrous minerals inthe descending mantle wedge along with the slabwould also release water. Though, temperature atthe top of the mantle wedge beneath Shikoku wasestimated at about 450-500◦C by Hyndman et al.

(1995). The temperature is considered too low forserpentine in the mantle wedge to release much ofwater.

Nor

thea

ster

n J

apan

Kii Peninsula

Shikoku

breakdown of serp

breakdown

of chl

breakdown

of serp

breakdown

of chl

Figure 20: Phase diagram for water-saturated av-erage mantle peridotite, modified from a figure bySchmidt and Poli (1998), and temperature profiles forthe top of the Pacific Plate under northeastern Japan[Iwamori, 1998] and the Philippine Sea Plate beneaththe Kii Peninsula and Shikoku [Hyndman et al., 1995]in southwestern Japan: ’A’ = phase A, amph = amphi-bole, chl = chlorite, cpx = clinopyroxene, gar = garnet,ol = olivine, opx = orthopyroxene, serp = serpentine,sp = spinel, tc = talc. The breakdown of serpentineand of chlorite, which mean releasing water, are in-dicated along the temperature profiles and the phaseboundaries.

LFT is observed in the limited range of slabtop depth, 25–40 km. Since the amount of waterfrom dehydration processes in the oceanic crustwould increase as temperature and pressure in-crease, a wider belt area would be expected. Ac-cording to the structure shown by Kurashimo et

al. (2002), descending oceanic crust starts con-tact with the mantle wedge under the northernrim of the belt distribution (Fig. 6). Fluid waterrelating to the tremor would not be supplied tothe inland crust in the north of the rim becausetemperature in the mantle wedge would be lowenough for mantle peridotite to absorb ascendingwater forming hydrous minerals. There would beanother possibility that tremor would not occurin the boundary region between the inland crust

and the mantle wedge due to a different mechan-ical condition from that between the inland crustand the oceanic crust.

When the Philippine Sea plate reach at thedepth around 50km, melting of the slab wouldstart. On the other hand, the lowest part ofthe mantle wedge, which consists of hydrous peri-dotite, descends with slab and release water be-cause of dehydration of serpentine and chlorite ataround the depth of 60-80km. At deeper point,the hydrous peridotite starts melting partially.Such high temperature fluid or magma wouldcause LFE.

Fig. 21 shows schematic view of the processthat generates LFT and LFE in the southwestJapan.

Belt of LFT epicenters

Water is produced by dehydration of Chlorite and formation of Clinopyroxene.

Oceanic Sediment

Water is fixedby forming serpentine.

Peridotite

Inland Crust

MantleOceanicCrust

LFT

Hydrous Peridotite

Dehydration ofChlorite

Dehydration ofSerpentine

Partial Melting

LFEMantle Wedge

Basalt

Water

Eclogite

Figure 21: Schematic view of the process that gener-ates LFT and LFE distributed within and of a belt andnorth of the belt in western Japan. Short arrows showwater being released from the descending PhilippineSea Plate. A considerable amount of water is releasedby both the dehydration of chlorite and the formationof clinopyroxene in slab basalt, and the upwelling wa-ter would cause LFT. When the slab comes in contactwith the mantle wedge, the upwelling water is fixedas serpentine in the lowest part of the mantle wedge.Therefore, LFT cannot occur because of the absenceof fluid. The gray stars show the hypocentral area forLFT and LFE.

Belt LFT activities are not recognized in theeast of 138◦E line and in Kyushu Island. In addi-tion, a gap is recognized around the Kii channel(Fig. 5). Since the activity is considered to berelated to dehydration process of hydrous miner-als such as chlorite, the inactive areas would in-dicate absence of dehydration process due to thefollowing reasons. The Philippine Sea plate movesnorthwestward in this area [e.g., Seno et al., 1993],


and the plate moves almost along the iso-depthcurves of the plate under the channel (Fig. 22). Ifthe dehydration process has almost done aroundthe Kii Peninsula before reaching this area, waterwould not be supplied. On the other hand, in thearea east of 138◦E, there are some volcanoes, Mt.Fuji, Mt. Hakone, and so on. Volcanic activitieswould affect thermal condition and dehydrationprocess of the subducting plate.

Figure 22: Horizontal displacement vector map basedon GPS measurements conducted by the Geographi-cal Survey Institute [Earthquake Research Committee,2001], and depth contours for the Philippine Sea Platein southwestern Japan. Arrows show horizontal dis-placement vectors from May in 1997 to June 2000. Thefixed station is Fukui. Circles show epicenters of low-frequency events from September 1 in 1999 to February20 in 2003. Open circles = Low-frequency events farfrom active volcanoes and the Quaternary volcanoes,Closed circles = Low-frequency events near active vol-canoes and the Quaternary volcanoes. Contours arelabeled with depth in km.

6 Locating the Tremor with De-

convolution Technique

The JMA determines the source location of thetremor with an ordinary hypocenter determina-tion method based on onset times of enlargingparts of the tremors. With this method, we maymiss to detect the tremor with stationary ampli-tude levels. The unavailability of P-wave onsetsmakes the locating precision poor. We are de-veloping a tremor-locating method using deconvo-lution which often used to get receiver functions[e.g., Ammon, 1991].

We get a function of H(ω) as

H(ω) =Uj(ω)U∗

i (ω)

φ(ω)G(ω),

φ(ω) = max{Ui(ω)U∗

i (ω), cUi(ω)U∗

i (ω)}

where Ui(ω), Uj(ω), G(ω) and c are observed spec-tra at ith and jth stations, gauss filter [Langston,1979], and a constant called as water level.

Inverse Fourier-transformed function of H(ω)should show a peak at a time of arrival-time differ-ence of stations of i and j. Fig. 24 shows examplesof the deconvolved functions. The original wave-forms are shown in Fig. 23. Some of time series inFig. 24 show clear peaks. We take times of thosepeaks to locate the source location.

ND .MSKD l t = 2 0 . 6A zm= 8 5 . 9ND . HOUD l t = 2 1 . 8A zm= 2 0 0 . 9LG . HRND l t = 2 5 . 8A zm= 1 3 0 . 1ND . ASHD l t = 2 8 . 2A zm= 2 7 8 . 1ND . KSHD l t = 2 9 . 0A zm= 2 8 7 . 2ND . ACHD l t = 3 2 . 8A zm= 3 . 9ND . SMYD l t = 3 9 . 6A zm= 2 4 6 . 3ND . HKWD l t = 3 9 . 9A zm= 1 0 3 . 9ND . KGWD l t = 4 5 . 1A zm= 1 4 1 . 2

5 . 0 ( s )

Figure 23: Waveform data used for the deconvolution.

0 . 0 2

- 0 . 0 2

ND . ASH d l t = 2 8 . 2 2 ( km) t p k= 0 . 0 0 ( s ) pw i d= 0 . 2 1 ( s ) t p k ( 2 ) = 6 . 2 6 ( s )

0 . 0 1

- 0 . 0 1

ND . KSH d l t = 2 9 . 0 4 ( km) t p k= 0 . 2 0 ( s )

67 0

pw i d= 0 . 1 9 ( s ) t p k ( 2 ) = 1 . 4 7 ( s )

0 . 0 0

0 . 0 0

ND . ACH d l t = 3 2 . 8 2 ( km) t p k= 1 1 . 1 4 ( s )

7 1

pw i d= 0 . 1 9 ( s ) t p k ( 2 ) = 0 . 9 1 ( s )

0 . 0 1

- 0 . 0 1

ND . SMY d l t = 3 9 . 6 7 ( km) t p k= 2 . 1 0 ( s )

7 2

pw i d= 0 . 2 1 ( s ) t p k ( 2 ) = 2 . 7 2 ( s )

0 . 0 0

0 . 0 0

ND . HKW d l t = 3 9 . 9 4 ( km) t p k= 2 . 7 3 ( s )

7 3

pw i d= 0 . 2 1 ( s ) t p k ( 2 ) = 5 . 0 1 ( s )

0 . 0 0

0 . 0 0

ND . KGW d l t = 4 5 . 1 9 ( km) t p k= 1 . 6 1 ( s )

7 4

pw i d= 0 . 2 8 ( s ) t p k ( 2 ) = 5 . 1 1 ( s )

0 . 0 0

0 . 0 0

ND .MKB d l t = 4 5 . 8 7 ( km) t p k= 3 . 3 1 ( s )

77 5

pw i d= 0 . 2 1 ( s ) t p k ( 2 ) = 4 . 6 4 ( s )

0 . 0 0

0 . 0 0

KUROMA d l t = 4 7 . 0 7 ( km) t p k= 4 . 7 5 ( s )

7 6

pw i d= 0 . 2 9 ( s ) t p k ( 2 ) = - 0 . 2 1 ( s )

0 . 0 0

0 . 0 0

ND . HTN d l t = 4 7 . 2 1 ( km) t p k= 4 . 2 1 ( s )

7 7

pw i d= 0 . 0 9 ( s ) t p k ( 2 ) = 0 . 9 2 ( s )

0 . 0 0

0 . 0 0

ND . NUK d l t = 4 7 . 6 8 ( km) t p k= 3 . 4 6 ( s )

7 8

pw i d= 0 . 2 2 ( s ) t p k ( 2 ) = 4 . 1 4 ( s )

0 . 0 0

0 . 0 0

KAKEGA d l t = 4 9 . 8 7 ( km) t p k= 4 . 9 5 ( s )

7 9

pw i d= 0 . 1 1 ( s ) t p k ( 2 ) = 5 . 5 1 ( s )

2 . 0 ( s )

Figure 24: Deconvolved waveforms. The top trace is aresult of deconvolution by itself. The primary and sec-ondary peaks are marked by red and green segments.

We determine the source locations as to reducesquare of residuals of the arrival-time differencesas

∑{(tj − ti)

obs − (tj − ti)cal}2 → min.

The summation is taken over pairs of arrival-timedifferences ti and tj .

Fig. 25 shows an estimated location (an circle)and iso-arrival-time-difference curves.

7 Conclusions

Unified processing of seismic data in Japan low-ered the detectable magnitudes of earthquakes,


0 5 0 km

3 4 4 0 ’ N

3 5 N

3 5 2 0 ’ N

3 5 4 0 ’ N

1 3 7 E 1 3 7 2 0 ’ E 1 3 7 4 0 ’ E 1 3 8 E 1 3 8 2 0 ’ E

t s d e p= 3 6 . 5 + / - 0 . 7

ND .MSK

ND . HOU

ND .MSK

LG . HRN

ND .MSKND . ASH

ND .MSK

ND . KSH

ND .MSK

ND . ACH

ND .MSK

ND . SMY

ND .MSK

ND . HKW

ND .MSK

ND . KGW

ND .MSK

ND .MKB

ND .MSK

KUROMA

ND .MSK

ND . HTN

ND .MSK

ND . NUK

ND .MSK

KAKEGA

ND .MSK

ND . OKB

ND .MSK

ND . S I Z

ND .MSK

KUROKA

ND .MSK

ND . KG I

ND .MSK

ND . OHS

ND .MSKE . OKY

ND .MSK

ND . AN J

ND .MSK

ND . KGN

ND .MSK

ND . HAZ

ND .MSK

ND . GER

ND .MSK

ND . ABN

ND . HOULG . HRN

ND . HOU

ND . ASH

ND . HOU

ND . KSH

ND . HOU

ND . ACH

ND . HOU

ND . SMY

ND . HOU

ND . HKW

ND . HOU

ND . KGW

ND . HOU

ND .MKB

ND . HOU

KUROMA

ND . HOU

ND . HTN

ND . HOU

ND . NUK

ND . HOU

KAKEGA

ND . HOU

ND . OKB

ND . HOU

ND . S I Z

ND . HOU

KUROKA

ND . HOU

ND . KG I

ND . HOU

ND . OHS

ND . HOU

E . OKY

ND . HOU

ND . AN J

ND . HOU

ND . KGN

ND . HOU

ND . HAZ

ND . HOU

ND . GER

ND . HOU

ND . ABN

LG . HRN

ND . ASH

LG . HRN

ND . KSH

LG . HRN

ND . ACH

LG . HRNND . SMY LG . HRN

ND . HKW

LG . HRN

ND . KGW

LG . HRN

ND .MKB

LG . HRNKUROMA

LG . HRN

ND . HTN

LG . HRN

ND . NUK

LG . HRN

KAKEGA

LG . HRN

ND . OKB

LG . HRN

ND . S I Z

LG . HRN

KUROKA

LG . HRN

ND . KG I

LG . HRN

ND . OHS

LG . HRN

E . OKY

LG . HRN

ND . AN J

LG . HRN

ND . KGN

LG . HRN

ND . HAZ

LG . HRN

ND . GER

LG . HRN

ND . ABN

ND . ASH

ND . KSH

ND . ASH

ND . ACH

ND . ASH

ND . SMY

ND . ASH

ND . HKW

ND . ASH

ND . KGW

ND . ASH

ND .MKB

ND . ASH

KUROMA

ND . ASH

ND . HTN

ND . ASH

ND . NUK

ND . ASH

KAKEGA

ND . ASH

ND . OKB

ND . ASH

ND . S I Z

ND . ASH

KUROKA

ND . ASH

ND . KG I

ND . ASH

ND . OHS

ND . ASH E . OKY ND . ASH

ND . AN J

ND . ASH

ND . KGN

ND . ASH

ND . HAZ

ND . ASH

ND . GER

ND . ASH

ND . ABN

ND . KSH

ND . ACH

ND . KSH

ND . SMY

ND . KSH

ND . HKW

ND . KSH

ND . KGW

ND . KSH

ND .MKB

ND . KSH

KUROMA

ND . KSH

ND . HTN

ND . KSH

ND . NUK

ND . KSH

KAKEGA

ND . KSH

ND . OKB

ND . KSH

ND . S I Z

ND . KSH

KUROKA

ND . KSH

ND . KG I

ND . KSH

ND . OHS

ND . KSHE . OKY

ND . KSH

ND . AN J

ND . KSH

ND . KGN

ND . KSH

ND . HAZ

ND . KSH

ND . GER

ND . KSH

ND . ABN

ND . ACH

ND . SMY

ND . ACH

ND . HKW

ND . ACH

ND . KGW

ND . ACH

ND .MKB

ND . ACH

KUROMA

ND . ACH

ND . HTN

ND . ACH

ND . NUK

ND . ACH

KAKEGA

ND . ACH

ND . OKB

ND . ACH

ND . S I Z

ND . ACH

KUROKA

ND . ACH

ND . KG I

ND . ACH

ND . OHS

ND . ACH

E . OKY

ND . ACH

ND . AN J

ND . ACH

ND . KGN

ND . ACH

ND . HAZ

ND . ACH

ND . GER

ND . ACH

ND . ABN

ND . SMY

ND . HKW

ND . SMY

ND . KGW

ND . SMY

ND .MKB

ND . SMYKUROMA

ND . SMY

ND . HTN

ND . SMY

ND . NUK

ND . SMY

KAKEGA

ND . SMY

ND . OKB

ND . SMY

ND . S I Z

ND . SMY

KUROKA

ND . SMY

ND . KG I

ND . SMY

ND . OHS

ND . SMY

E . OKY

ND . SMY

ND . AN J

ND . SMY

ND . KGN

ND . SMY

ND . HAZ

ND . SMY

ND . GER

ND . SMY

ND . ABN

ND . HKW

ND . KGW

ND . HKW

ND .MKB

ND . HKWKUROMA

ND . HKW

ND . HTN

ND . HKW

ND . NUK

ND . HKW

KAKEGA

ND . HKW

ND . OKB

ND . HKWND . S I Z

ND . HKW

KUROKA

ND . HKW

ND . KG I

ND . HKW

ND . OHS

ND . HKW

E . OKY

ND . HKW

ND . AN J

ND . HKW

ND . KGN

ND . HKW

ND . HAZ

ND . HKW

ND . GER

ND . HKW

ND . ABN

ND . KGW

ND .MKB

ND . KGW

KUROMA

ND . KGW

ND . HTN

ND . KGW

ND . NUK

ND . KGW

KAKEGA

ND . KGW

ND . OKB

ND . KGW

ND . S I Z

ND . KGW

KUROKA

ND . KGW

ND . KG I

ND . KGW

ND . OHS

ND . KGW

E . OKY

ND . KGW

ND . AN J

ND . KGW

ND . KGN

ND . KGW

ND . HAZ

ND . KGW

ND . GER

ND . KGW

ND . ABN

ND .MKB

KUROMA

ND .MKB

ND . HTN

ND .MKB

ND . NUK

ND .MKB KAKEGAND .MKB

ND . OKB

ND .MKB

ND . S I Z

ND .MKB

KUROKA

ND .MKB

ND . KG I

ND .MKB

ND . OHS

ND .MKB

E . OKY

ND .MKB

ND . AN J

ND .MKB

ND . KGN

ND .MKBND . HAZ ND .MKB

ND . GER

ND .MKB

ND . ABN

KUROMA

ND . HTN

KUROMA

ND . NUK

KUROMA

KAKEGA

KUROMA

ND . OKB

KUROMA

ND . S I Z

KUROMA

KUROKA

KUROMA

ND . KG I

KUROMA

ND . OHS

KUROMA

E . OKY

KUROMA

ND . AN J

KUROMA

ND . KGN

KUROMA

ND . HAZ

KUROMA

ND . GER

KUROMA

ND . ABN

ND . HTN

ND . NUK

ND . HTN

KAKEGA

ND . HTN

ND . OKB

ND . HTN

ND . S I Z

ND . HTN

KUROKA

ND . HTNND . KG I ND . HTN

ND . OHS

ND . HTN

E . OKY

ND . HTN

ND . AN J

ND . HTN

ND . KGN

ND . HTN

ND . HAZ

ND . HTN

ND . GER

ND . HTN

ND . ABN

ND . NUK

KAKEGA

ND . NUKND . OKB

ND . NUK

ND . S I Z

ND . NUK

KUROKA

ND . NUK

ND . KG I

ND . NUK

ND . OHS

ND . NUK

E . OKY

ND . NUKND . AN J ND . NUK

ND . KGN

ND . NUK

ND . HAZ

ND . NUK

ND . GER

ND . NUK

ND . ABN

KAKEGA

ND . OKB

KAKEGA

ND . S I Z

KAKEGA

KUROKA

KAKEGA

ND . KG I

KAKEGA

ND . OHS

KAKEGA

E . OKY

KAKEGA

ND . AN J

KAKEGA

ND . KGN

KAKEGAND . HAZ KAKEGA

ND . GER

KAKEGA

ND . ABN

ND . OKB

ND . S I Z

ND . OKB

KUROKA

ND . OKB

ND . KG I

ND . OKB

ND . OHS

ND . OKB

E . OKY

ND . OKBND . AN J

ND . OKB

ND . KGN

ND . OKB

ND . HAZ

ND . OKB

ND . GER

ND . OKB

ND . ABN

ND . S I Z

KUROKA

ND . S I Z

ND . KG I

ND . S I Z

ND . OHS

ND . S I Z

E . OKY

ND . S I Z

ND . AN J

ND . S I Z

ND . KGN

ND . S I Z

ND . HAZ

ND . S I Z

ND . GER

ND . S I Z

ND . ABN

KUROKA

ND . KG I

KUROKA

ND . OHS

KUROKA

E . OKY

KUROKA

ND . AN J

KUROKA

ND . KGN

KUROKA

ND . HAZ

KUROKA

ND . GER

KUROKA

ND . ABN

ND . KG I

ND . OHS

ND . KG I

E . OKY

ND . KG I

ND . AN J

ND . KG I

ND . KGN

ND . KG I

ND . HAZ

ND . KG I

ND . GER

ND . KG I

ND . ABN

ND . OHS

E . OKY

ND . OHS

ND . AN J

ND . OHS

ND . KGN

ND . OHS

ND . HAZ

ND . OHS

ND . GER

ND . OHS

ND . ABN

E . OKY

ND . AN J

E . OKY

ND . KGN

E . OKY

ND . HAZ

E . OKY

ND . GER

E . OKY

ND . ABN

ND . AN J

ND . KGN

ND . AN J

ND . HAZ

ND . AN J

ND . GER

ND . AN J

ND . ABN

ND . KGN

ND . HAZ

ND . KGNND . GER

ND . KGN

ND . ABN

ND . HAZ

ND . GER

ND . HAZ

ND . ABN

ND . GER

ND . ABN

2 1

2 32 6

2 7

2 9

3 9

7 8

9 6

9 81 1 4

1 7 8

1 8 7

2 6 4

2 7 6

Figure 25: Iso-arrival-time-difference curves. Bluecurves, red line segments and a circle show iso-arrival-time-difference curves, the pair of stations, the esti-mated source location, respectively.

and activities of deep low-frequency earthquakesin non-volcanic regions became recognized. Afterthe development of the Hi-net in southwest Japan,deep tremor of a belt distribution along the strikeof the subducting Philippine Sea plate was found.

Source locations of the deep tremors was esti-mated with onsets of P and S waves, and it wasfound that source depth ranged from 25 to 40 kmand that they were distributed around the bound-ary region between island and oceanic crusts nearmantle wedge.

The east end of the tremor belt distribution isadjacent to the Tokai slow-slip area. The Tokaislow-slip has been continuing since 2000. Activitylevel of the tremor in the region has been kept highsince the beginning of 2003, which coincides withthe period of relatively fast slow-slip. The regionis located in the west of the probable source area ofthe Tokai Earthquake. We expect that the tremoractivity could be a new valuable indicator for theTokai Earthquake prediction.

Source force directions inferred from polariza-tion of S-waves were found to be along the di-rection of the Philippine Sea plate motion, whichwere clearly different from those of deep low-frequency earthquakes in volcanic regions. Thelack of frequency components lower than 0.1 Hzindicates that the tremor is not a direct outcomeof the slow slip at the boundary of the Philippine

Sea plate.It is assumed that the tremor is related to wa-

ter which is produced by dehydration of chloriteand formation of clinopyroxene in the subductingoceanic crust. The northern rim of the belt re-gion seems to correspond to the edge of the man-tle wedge. In the north of the edge, fluid waterwould be depleted to form hydrous minerals suchas serpentine in the mantle wedge. This wouldbe the cause of the relatively narrow width of thebelt distribution.

Acknowledgments

We used seismic data from the National Re-search Institute for Earth Science and DisasterPrevention, Hokkaido University, Hirosaki Uni-versity, Tohoku University, University of Tokyo,Nagoya University, Kyoto University, Kochi Uni-versity, Kyushu University, Kagoshima University,the National Institute of Advanced Industrial Sci-ence and Technology, Tokyo metropolitan gov-ernment, Shizuoka prefectural government, Kana-gawa prefectural government, the City of Yoko-hama, the Japan Marine Science and TechnologyCenter, and the Japan Meteorological Agency.

References

[1] Ammon, C. J., The isolation of receiver effectsfrom teleseismic P waveforms, Bull. Seism. Soc.

Am., 81, 2504–2510, 1991.

[2] Aki, K., Scaling law of seismic spectrum, J.

Geophys. Res., 72, 1217-1231, 1967.

[3] Chouet, B. A, Long-period volcano seismicity:its source and use in eruption forecasting, Na-

ture, 380, 309–316, 1996.

[4] Committee for Catalog of Quaternary Vol-canoes in Japan, eds., Catalog of Quater-nary volcanoes in Japan (CD-ROM with map),The Volcanological Society of Japan, 1999 (inJapanese).

[5] Earthquake Research Committee, On thelong-term evaluation of earthquakes in theNankai trough, 2001 (In Japanese).

[6] Funasaki and Earthquake Prediction Informa-tion Division, Revision of the JMA velocitymagnitude, Quart. J. Seism., 67, 11-17, 2004(in Japanese with English abstract).


[7] Hasegawa, A. and A. Yamamoto, Deep, low-frequency microearthquakes in or around seis-mic low-velocity zones beneath active volcanoesin northeastern Japan, Tectonophysics, 233,233-252, 1994.

[8] Hasegawa, A., D. Zhao, S. Hori, A. Ya-mamoto, and S. Horiuchi, Deep structure of thenortheastern Japan arc and its relationship toseismic and volcanic activity, Nature, 352, 683-689, 1991.

[9] Hirose, Hi and K. Obara, Repeating slowslip events which correlate deep low-frequencytremor activity in Southwest Japan, Prog. Ab-

str. Fall Meeting Seismol. Soc. Jpn., C098, 2003(in Japanese).

[10] Hyndman, R. D., K. Wang, and M. Yamano,Thermal constraints on the seismogenic portionof the southwestern Japan subduction thrust, J.Geophys. Res., 100, 15373–15392, 1995.

[11] Iwamori, H., Transportation of H2O andmelting in subduction zones, Earth Planet. Sci.

Lett., 160, 65–80.

[12] Katsumata, A., Revision of the JMA dis-placement magnitude, Quart. J. Seism., 67, 1-10, 2004 (in Japanese with English abstract).

[13] Kamaya, N. and A. Katsumata, Low-frequency events away from volcanoes in theJapan Islands, Zisin 2, 57, 11–28, 2004 (inJapanese with English abstract).

[14] Katsumata, A. and N. Kamaya, Low-frequency continuous tremor around the Mohodiscontinuity away from volcanoes in the south-west Japan, Geophys. Res. Lett., 30, 1,10.1029/2002GL015981, 2003.

[15] Kurashimo, E., M. Tokunaga, N. Hirata, T.Iwasaki, S. Kodaira, Y. Kaneda, K. Ito, R.Nishida, S. Kimura, and T. Ikawa, Geome-try of the subducting Philippine Sea plate andthe crustal and upper mantle structure be-neath eastern Shikoku Island revealed by seis-mic refraction/wide-angle reflection profiling,Zisin 2, 54, 489–505, 2002 (in Japanese withEnglish abstract).

[16] Langston, C. A., Structure under MountRainier, Washington, inferred from teleseismicbody waves, J. Geophys. Res., 84, 4749-4762,1979.

[17] Nishide, N., T. Hashimoto, J. Funasaki, H.Nakazawa, M. Oka, H. Ueno, N. Yamada,I. Sasakawa, K. Maeda, K. Sugimoto, andT. Takashima, Nationwide activity of low-frequency earthquakes in the lower crust inJapan, Abstr. Jpn. Earth and Planet. Sci. Joint

Meeting, sk-p002, 2000.

[18] Obara, K., Nonvolcanic deep tremor associ-ated with subduction in southwest Japan, Sci-

ence, 296, 1679–1681, 2002.

[19] Okada, Y., K. Kasahara, S. Hori, K. Obara,and S. Aoi, Hi-net (1): outline, Prog. Abstr.

Fall Meeting Seismol. Soc. Jpn., P004, 2000 (inJapanese).

[20] Tsuboi, C., Determination of the Gutenberg-Richter’s magnitude of earthquakes occurringin and near Japan, Zisin 2, 7, 185–193, 1954(In Japanese with English abstract).

[21] Schmidt, M. W. and S. Poli, Experimentallybased water budgets for dehydrating slabs andconsequences for arc magma generation, Earth

Planet. Sci. Lett., 163, 361–379, 1998.

[22] Seno, T., S. Stein, and A. E. Gripp, A modelfor the motion of the Philippine Sea plate con-sistent with NUVEL-1 and geological data, J.

Geophys. Res., 98, 17941–17948, 1993.

[23] Ukawa, M. and K. Obara, Low frequencyearthquakes around Moho beneath the volcanicfront in the Kanto district, central Japan, Bull.

Volcanol. Soc. Japan, 38, 187–197, 1993 (inJapanese with English abstract).

[24] Ukawa, M. and K. Ohtake, A monochro-matic earthquake suggesting deep-seated mag-matic activity beneath the Izu-Ooshinma vol-cano, Japan, J. of Geophys. Res., 92, 12,649–12,663, 1987.

abstract - usgs...depths of 20{40km in the japan islands. the jma has distinguished those deep...

Documents