Contribution of Slow Study for Assessing the Occurrence Potential of Megathrust

Review: Contribution of Study for Assessing the Occurrence Potential of Megathrust Earthquakes

Kazushige Obara

Earthquake Research Institute, The University of 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-0032, Japan E-mail: [email protected] [Received February 17, 2014; accepted May 12, 2014]

Studies of slow earthquakes during the last decade nomena, may be related to the stress regimes that cause have suggested a relationship between various types megathrust earthquakes, as the source regions of the two of earthquakes occurring at the interface between types of earthquakes are adjacent to one another. subducting oceanic plates and overlying continental Slow earthquakes can be broadly classified into seismic plates. Such a relationship has been postulated for and geodetic phenomena. For example, in 1992, an after- slow earthquakes, which are distributed between the slip event (in which the slip deficit accumulating adjacent stable sliding zone and the locked zone, and megath- to a large earthquake source is released) followed rust earthquakes, which are located in the locked by a magnitude (M) 6.9 earthquake off northern Honshu, zone. The adjacency of the respective sources of slow Japan, was first reported by Kawasaki et al. (1995) [1] and megathrust earthquakes suggests expected inter- as a geodetic slow earthquake (Fig. 2). The development actions between these two types of earthquakes. Ob- of GPS observation networks since the 1990s has enabled served interactions between different types of slow not only the detection of some afterslip events, but also earthquakes located at neighbor area suggest a com- the discovery of slow slip events (SSEs), in which small mon triggering mechanism in the seismogenic zone. amounts of crustal deformation occur spontaneously. Af- Also, it is expected that stress accumulations in the ter the construction of the GPS network GEONET, op- locked zone should influence stress regimes in sur- erated by the Geospatial Information Authority of Japan rounding regions; thus, slow earthquake activity in the (GSI) (Sagiya, 2004 [2]), an SSE lasting for about one stable sliding zone may change in response to stress year was detected in 1997 in the Bungo Channel between build-up in the locked zone. Numerical simulations Kyushu and Shikoku islands, southwestern Japan (Fig. 2). reproducing both megathrust and slow earthquakes Hirose et al. (1999) [3] estimated that the SSE event was show a shortening of the recurrence interval between generated by a thrusttype focal mechanism located on the slow earthquake episodes leading up to the occurrence downdip side of the Nankai earthquake seismogenic zone. of a . Similarities between the In the Tokai region, central Japan, similar SSEs were de- activities of slow and megathrust earthquakes, such as tected between late 2000 and 2005 (e.g., Ozawa et al., those related to periodicity and patterns of multiseg- 2002 [4]; Miyazaki et al., 2006 [5]). The Tokai SSE was ment ruptures, are useful for understanding megath- of interest because during the initial stage, the SSE oc- rust earthquakes, particularly given the higher fre- curred along the downdip part of the slab interface neigh- quency of occurrence of slow earthquakes. From this boring the source area of an anticipated Tokai earthquake. perspective, the continuous and accurate monitoring Similarly, SSEs with large magnitudes (M7) and long du- of slow earthquake activity is important for evaluating rations (months to years) have been detected in other sub- the occurrence potential of megathrust earthquakes. duction zones, for example in (e.g., Ohta et al., 2006 [6]) and in Mexico (e.g., Vergnolle et al., 2010 [7]) Keywords: slow earthquake, nonvolcanic tremor, slow (Fig. 3). slip, zone, megathrust earthquake In the , along the western coast of North America, a significant SSE was discovered by a network of densely distributed GPS sites (Dragert 1. Introduction et al., 2001 [8]) (Fig. 3). This SSE (Mw6.7) was a reverse faulting event occurring on the interface of the deeper part Slow earthquakes, which are characterized by a wide from the megathrust seismogenic zone along the subduct- spectral range, and megathrust earthquakes both occur in ing plate interface. The duration of the SSE of weeks was subduction zones, along the interface between oceanic much shorter than those of the SSEs in Japan, Alaska, and and overlying plates (Fig. 1). The source of slow earth- Mexico described above. To discriminate SSEs based on quakes is in the transition zone between the locked and their durations, events such as the SSE in Cascadia are re- stable sliding zones. Therefore, the occurrence of slow ferred to as short-term SSEs, while SSEs with durations earthquakes, which are considered as transitional phe- of months to years (such as those in the Bungo Channel,

Journal of Disaster Research Vol.9 No.3, 2014 317 Obara, K.

Near Downdip side Characteristic time (tc) Sensor Network

Long-term slow slip event GPS GSI 100day (tc:0.5~5years) GEONET Short-term slow slip event 1day Tiltmeter (tc:2~6days)

1000sec Deep very low frequency Tiltmeter NIED Earthquake䠄VLF䠅 (tc:20sec䠅 Hi-net Shallow very low 10sec frequency earthquake Deep low frequency tremor Short-period (S-VLF) (tc:10sec䠅 0.1sec (tc:1.5~5Hz)

Accretionary prism Nankai Trough Long-term SSE Short-term SSE S-VLF VLF Tremor

Fig. 1. Cross sectional schematic illustration of slow earthquakes in southwest Japan. On the top panel, the difference in characteristic time of each slow earthquake is shown. On the right side of the top panel, adequate sensor and observation network for each slow earthquake are indicated.

Tokai, Alaska, and Mexico) are referred to as long-term SSEs. Interestingly, the short-term SSE in Cascadia was Afterslip characterized by an along-strike migration of the source Long-term SSE location at a speed of several km per day. After the dis- covery of the Cascadia SSE, Miller et al. (2002) [9] found Boso-type SSE 1 that it recurred at an interval of about 14 months. There- Shallow VLF Tokachi-oki fore, this SSE is interpreted as a stickslip phenomenon ETS(Deep VLF, tremor, 11 Short-term SSE) occurring on the downdip side of the megathrust seis- 2 Sanriku-oki mogenic zone, which is estimated to have ruptured in Tohoku SSE AD 1700 based on historical descriptions of in 15 Japan (Satake et al., 2003 [10]) (Fig. 4). 3 Miyagi-oki A nonvolcanic tremor was the first seismic slow earth- 4 quake to be recorded in southwestern Japan (Obara, Fukushima-oki 2002 [11]; Obara and Shiomi, 2009 [12]) (Figs. 1 and 2) Tokai 10 14 based on the analysis of Hi-net operated by National Re- Bungo Cha. 6 search Institute for Earth Science and Disaster Prevention 14 Boso 14 (NIED) (Okada et al., 2004 [13]; Obara et al., 2005 [14]). 7 12 12 8 12 12 The tremor was distributed along a belt-shaped zone on 5 13 9 the downdip side of the seismogenic zone of the Nankai 12 Trough megathrust earthquake, parallel to the strike of the Hyuga-nada subducting . The tremor source lo- cation usually migrated along strike at a velocity of ap- Fig. 2. Distribution of slow earthquakes in Japan. proximately 10 km/day. Based on the similarity of the Afterslip: 1, Miyazaki et al. (2004) [16]; 2, Kawasaki et short-term SSEs in Cascadia and the tremor in south- al. (1995) [1]; 3, Miura et al. (2006) [17]; 4, Suito et al. western Japan, with respect to their locations relative to (2011) [18]; 5, Yagi and Kikuchi (2003) [19], Long-term the seismogenic zone and to their migration properties, SSE: 6, Miyazaki et al. (2006) [5]; 7, Kobayashi (2012) [20]; Rogers and Dragert (2003) [15] searched for a seismic 8, Hirose et al. (1999) [3]; 9, Yarai and Ozawa (2013) [21], tremor during the occurrence period of the SSEs in Cas- Boso-type SSE: 10, Hirose et al. (2012) [22], Shallow VLF cadia and discovered a coupled phenomenon composed earthquake: 11, Asano et al. (2008) [23]; 12, Obara and Ito of a geodetic SSE and a seismic tremor, referred to as (2005) [24]; 13, Hirose et al. (2010) [25], ETS: 14, Obara et episodic tremor and slip (ETS) (Fig. 4). A coupling phe- al. (2004) [26], Ito et al. (2007) [27], Tohoku SSE: 15, Kato nomenon that includes both short-term SSE and tremor et al. (2012) [28], Ito et al. (2013) [29].

318 Journal of Disaster Research Vol.9 No.3, 2014 Contribution of Slow Earthquake Study for Assessing the Occurrence Potential of Megathrust Earthquakes

ETS(Episodic tremor and slip) Alaska Deep VLF earthquake 21 14 9 60N Aleutian Queen Charlotte Fault Ambient tremor without SSE 9 (including triggered tremor) 2 Cascadia

8 San Andreas Fault Triggered tremor only Japan 30N Mexico Taiwan 4 19 20 Long-term SSE 13 9 10 1 31517 Haiti

Boso-type shallow SSE 22 Costa Rica 0 Shallow VLF earthquake 12 7 18 Tremor excited by SSE 30S Shallow VLF excited by SSE New Zealand 6 Southern Chile 11 5 16 60S

0 60E 120E 180W 120W 60W

Fig. 3. World-wide distribution of various types of slow earthquakes. ETS: 1, Obara et al. (2004) [26], Maeda and Obara (2009) [30]; 2, Rogers and Dragert (2003) [15], Deep VLF earthquake: 3, Ito et al. (2007) [27], Ambient tremor: 4, Peng and Chao (2008) [31], Chao et al. (2013) [32]; 5, Ide (2012) [33], Kim et al. (2011) [34]; 6, Gallego et al. (2013) [35]; 7, Brown et al. (2009) [36]; 8, Nadeau and Dolenc (2005) [37], Triggered tremor: 9, Chao et al. (2013) [32], Long-term SSE: 10, Hirose et al. (1999) [3], Ozawa et al. (2002) [4]; 11, Wallace and Beavan (2010) [38]; 12, Jiang et al. (2012) [39]; 13, Vergnolle et al. (2010) [7]; 14, Ohta et al. (2006) [6], Boso-type shallow SSE: 15, Hirose et al. (2012) [22]; 16, Wallace and Beavan (2010) [38], Shallow VLF earthquake: 17, Obara and Ito (2005) [24], 18, Walter et al. (2013) [40], Tremor excited by SSE: 19, Obara et al. (2010) [41]; 20, Zigone et al. (2012) [42]; 21, Peterson and Christensen (2009) [43], Shallow VLF excited by SSE: 22, Hirose et al. (2010) [25].

SW Japan Cascadia

Philippine Sea plate

Fig. 4. Distribution of ETS and rupture area of the megathrust earthquake in southwest Japan and Cascadia. (Left) Southwest Japan. Distribution of ETS is represented by tremor (red dot), deep VLF earthquake (blue star) and short-term SSE (black square) (Obara, 2011 [44]). The source area of anticipated huge earthquake (Headquarters for Earthquake Research Promotion, 2001 [45]) is plotted by yellow. (Right) Cascadia. Distribution of ETS and rupture area of the megathrust earthquake is represented by tremor (red shaded area) (Wech et al., 2009 [46]) and Tsunami source zone for 1700 huge earthquake (Satake et al., 2003 [10]), respectively.

Journal of Disaster Research Vol.9 No.3, 2014 319 Obara, K.

was also detected in southwestern Japan by Obara et al. sensors, whereas the crustal deformation caused by the (2004) [26]. The durations of short-term SSE associated short-term SSEs in southwestern Japan is mainly detected with the tremor in southwestern Japan (several days to a by borehole tiltmeters or strainmeters. However, in Cas- week) are slightly shorter than that of the SSE in Cas- cadia, long-term SSEs, deep VLF earthquakes, and shal- cadia. Tremor during an ETS is interpreted as a seismic low VLF earthquakes have thus far not been detected. signature during slow slip on the plate interface, and may In subduction zones other than those of Japan and Cas- represent seismic rupture along small patches on the slow cadia, tremor activity has been detected on the downdip slip fault plane. The ETSs in southwestern Japan are also side of the source region of long-term SSEs in both Mex- accompanied by very low frequency (VLF) earthquakes ico and Alaska (Payero et al., 2008 [49]; Peterson and with long periods of about 20 s (Ito et al., 2007 [27]) Christensen, 2009 [43]) after the detection of long-term (Figs. 1 and 2). The long-period seismograms from the SSEs. However, clear short-term SSEs have not been VLF earthquakes, which are useful for moment tensor recognized, whereas small short-term SSEs may be in- analysis, indicate that the VLF earthquakes have a thrust- cluded in the GPS time sequence in Mexico (Vergnolle et type focal mechanism on the subducting plate interface. al., 2010 [7]). In New Zealand, two types of SSEs have The discovery of VLF earthquakes associated with ETS been reported (Wallace and Beavan, 2010 [38]): long- episodes in southwestern Japan supports the interpretation term SSEs are located in the deeper parts of the south- of ETS as stickslip interplate phenomena associated with ern part of the North Island, and SSEs seemingly simi- reverse faulting, which is also the main feature of megath- lar to those in Boso occur at the depth of shallower than rust earthquakes. that of the short-term SSE in southwest Japan and Cas- Similar VLF earthquakes have been detected near the cadia. Tremor activity has also been recognized on the Nankai Trough (Obara and Ito, 2005 [24]) (Figs. 1 and downdip side of the source area of these SSEs (Kim et al., 2). In this region, shallow and deep VLF earthquakes are 2011 [34]; Ide, 2012 [33]). defined as events occurring along the Nankai Trough and Tremor activity is detected not only in the reverse fault those associated with ETSs in the deeper transition zone, system in the subduction zone but also in the strikeslip respectively. The shallow VLF earthquakes are consid- fault system. In North America, Nadeau and Dolenc ered as seismic ruptures occurring at slow speeds within (2005) [37] discovered tremor-like seismic events beneath the accretionary prism, as determined by centroid moment the San Andreas fault, the strikeslip plate boundary be- tensor analysis using land-based broadband seismograph tween the North American and Pacific plates. data (Ito and Obara, 2006 [47]). Recently, temporary It is well known that tremor activity can be triggered oceanbottom seismic observations made to the southeast by external influences. In both Cascadia and southwest- of the Kii Peninsula have revealed that shallow VLF earth- ern Japan, tremor activities are activated at intervals of 12 quakes exhibit focal mechanisms of reverse faulting with or 24 hours during major tremor episodes as a response subhorizontal fault planes around the decollement discon- to Earth tides (Rubinstein et al., 2008 [50]; Nakata et al., tinuity (Sugioka et al., 2012 [48]). Therefore, the shallow 2008 [51]). Moreover, minor tremor activity is often trig- VLF earthquakes seem to occur at the plate interface near gered temporarily during the passage of large-amplitude the trench axis. surface waves from distant huge earthquakes because the The observations in southwestern Japan show that var- tremor wavetrain shows repeating amplitude peaks corre- ious types of slow earthquakes are distributed in both the lating to the long-period surface waves (Miyazawa and updip and downdip transition zones. Slow earthquakes in Mori, 2006 [52]; Rubinstein et al., 2007 [53]). Such the transition zone at the deeper part from the seismogenic tremor activities triggered by surface waves are detected zone are classified as long-term SSEs and ETSs coupled not only in Cascadia and southwestern Japan, but also in with tremor, deep VLF earthquakes, and short-term SSEs. many other regions of the world. Tremor events triggered In contrast, a slightly different type of SSE is found near by surface waves are easily detected and such tremor the Boso Peninsula, central Japan; these SSEs occur at in- may indicate the existence of slow slip (Chao et al., tervals of 6-7 years and have a duration of about one week 2013 [32]). Spontaneously occurring tremor is commonly (Hirose et al., 2012 [22]). The depth of the Boso SSE is known as ambient tremor, to differentiate them from trig- shallower than that of the short-term SSE in southwest gered tremor. For example, Peng and Chao (2008) [31] Japan and Cascadia. recorded tremor activity in central Taiwan triggered by a Various combinations of the different types of slow teleseismic event, and very recently Chao et al. (2014, earthquakes are found in other subduction zones (Fig. 3), in preparation [54]) detected ambient tremor in the same which may reflect contrasts in the properties of the dif- location of the source of the triggered tremor previously ferent plate interfaces and/or in the capability of detection detected. The triggering of tremor by both surface waves of the equipment arrays deployed in different areas. In and earth tides suggests that the source area of the slow Cascadia, ETSs coupled with both tremor episodes and earthquakes is of very low strength. The high sensitivity short-term SSEs occur over wide lateral domains of about of these slow earthquakes suggests that such earthquakes 1200 km. The magnitudes of the short-term SSEs with may be easily affected by the stress change caused by the ETSs in Cascadia are generally greater than those ob- accumulation of slip deficit in the locked zone of the plate served in southwestern Japan. This is one reason why interface. If so, slow earthquake activity may represent short-term SSEs in Cascadia are clearly detected by GPS a type of precursory change prior to the occurrence of

320 Journal of Disaster Research Vol.9 No.3, 2014 Contribution of Slow Earthquake Study for Assessing the Occurrence Potential of Megathrust Earthquakes

megathrust earthquakes. 2. Style of Slow Earthquake Activity as a Proxy Regarding the relationship between slow earthquakes for the Occurrence of Megathrust Earth- and megathrust earthquakes, Fig. 4 shows the distribution quakes of tremor source regions and megathrust earthquake rup- ture areas in southwestern Japan and in Cascadia. The In this section, the style of slow earthquake activity is distributions of the two types of events in each location reviewed from the perspective of its similarity to the style are parallel to each other and run along the strike of the of megathrust earthquake activity. respective subducting oceanic plates. Notably, the lateral extents of the tremor zones are exactly the same as those of the source areas of megathrust earthquakes. This sug- 2.1. Segmentation and the Periodicity of ETSs gests that these two types of event are strongly related to The ETS episodes in southwestern Japan are composed each other or that they are controlled by common condi- of three slowearthquake phenomena; tremor, deep VLF tions. earthquakes, and short-term SSEs. Comparing the mag- There are three ways in which studies of slow earth- nitudes of each phenomenon, short-term SSEs are con- quakes can contribute to the understanding of megathrust sidered to be the primary events and trigger other seis- earthquakes. The first is the similarity in activity style mic events. However, the ability to detect geodetic SSEs of the slow and megathrust earthquakes. For example, and seismic tremor differs according to both the sensi- megathrust earthquakes in the Nankai subduction zone tivity of the sensor and the signal-to-noise ratio. In the occur at an average interval of 100-150 years in three present study, tremor and short-term SSEs are discussed or four segments of the zone (Ando, 1975 [55]). How- because they are commonly detected in both Cascadia and ever, the recurrence interval is variable, and neighboring southwestern Japan. In the case of a major episode of segments can rupture simultaneously or individually sepa- events, the slip region of short-term SSE is estimated us- rated by a short time interval. In contrast, the ETS zone on ing geodetic data, and the source of the tremor is located the downdip side of the locked zone is divided into several using seismic data. However, a minor episode is recog- segments, each of which has a regular recurrence inter- nized only in the form of tremor activity detected by seis- val. The recurrence interval of ETS epsiodes is variable, mometers, and the crustal deformation is not measured and ETS episodes in one segment sometimes propagate because of the low signal-to-noise ratio for geodetic ob- to a neighboring segment. The greatest attraction of the servations. study of slow earthquakes is that the recurrence intervals Aguiar et al. (2009) [58] found that the duration of of these events are much shorter than those of megathrust tremor activity and the moment magnitude of short-term earthquakes. In southwestern Japan, 200-300 ETS events SSEs in ETS episodes are proportional to each other. A would be expected at each segment during an interseismic precise timeevolution analysis using GPS data in Casca- period between megathrust earthquakes. This allows the dia showed the collocation of slip distribution and tremor ETS cycle to be studied very frequently. during the migration episode (Bartlow et al., The second way is that slow earthquake activity may 2011 [59]). Therefore, it can be assumed that the inten- reflect some precursory change caused by the accumu- sity of tremor episode is proportional to the magnitude of lation of stress at the locked zone prior to an upcoming short-term SSEs and that the spatiotemporal distribution megathrust earthquake. Because of the sensitivity of slow of tremor (including minor tremor episodes) reflects the earthquakes, as described above, slow earthquake activity pattern of ETS activity. may be expected to vary according to the pattern of stress Based on the tremor distribution as shown in figure 2 of accumulation. Obara (2011) [44], the ETS zone in southwestern Japan The third way is that slow earthquakes may have the can be divided into several segments, each of which has a potential to trigger megathrust earthquakes. Deep slow regular recurrence interval between tremor episodes. The earthquakes such as ETSs show reverse faulting on the longer segments, which have lengths of about 100 km and plate interface on the downdip side of the megathrust seis- are located in northeastern Kii and western Shikoku, have mogenic zone. The faulting in the ETS zone is hypothe- a recurrence interval similar to that of tremor episodes sized to transfer stress to the updip locked zone, which (about six months). By contrast, on the shorter segments, encourages reverse faulting at the locked zone (Dragert et which have lengths of <50 km and are located in cen- al., 2004 [56]; Mazzotti and Adams, 2004 [57]). If the tral and western Kii and eastern Shikoku, episodes recur stress accumulated at the locked zone reaches a particular at an interval of approximately three months. Therefore, critical level, a slow earthquake may be sufficient trigger the recurrence interval of tremor episodes appears to be the occurrence of a megathrust earthquake (Kato et al., controlled by segment size. A careful check of the spa- 2012 [28]). tiotemporal distribution of tremor episodes indicates that Based on these three ways in which knowledge of slow both the recurrence interval and spatial extent of tremor earthquakes can inform our understanding of megathrust episodes vary. For example, in the northeastern Kii seg- earthquakes, this paper discusses the study of slow earth- ment, some tremor episodes have a short recurrence in- quakes for evaluating the occurrence potential of megath- terval, suggesting that parts of the segment may be com- rust earthquakes. Therefore, we focus on slow earth- posed of small units. quakes occurring spontaneously.

Journal of Disaster Research Vol.9 No.3, 2014 321 Obara, K.

2.2. Spatial Gaps in Tremor Activity and Migration rust earthquakes. The duration of long-term SSEs in the of ETSs Bungo Channel is about one year with a rapidslip period A plate boundary segment showing tremor activity is of a few months. Based on GPS observations, the re- currence interval of long-term SSEs is 6-7 years. The usually bounded by an aseismic portion within a belt- ∼ shaped zone along the plate interface. This along-strike moment magnitude of these events is Mw6.6 7.1 and variation may reflect local inhomogeneity of structure, the slip length is 10-20 cm. Kobayashi and Yamamoto given that both temperature and pressure are almost con- (2011) [64] discovered the regular recurrent behavior of stant because of the very similar depths of tremor. The ex- long-term SSEs before GPS monitoring data were avail- istence of spatial gaps in tremor activity suggests two end able, based on leveling and sealevel data from 1978. members of the plate interface: a firmly locked portion Kobayashi (2012) [20] detected other long-term SSEs in and a fully stable sliding portion. In southwestern Japan, the southern part of Shikoku Island, 100 km from the the belt-shaped zone of tremor is separated by two large Bungo Channel, although the recurrence interval of those gaps, located at Ise Bay and Kii Channel. The gap at Ise events is not known. Bay is the boundary between the Tokai and Kii segments. The long-term SSEs along the Tokai segment detected Tremor episodes in these two segments recurred indepen- between late 2000 and 2005 have relatively long durations as compared with those in the Bungo Channel. The fi- dently at an interval of about six months before 2006, as ∼ . shown in figure 2 of Obara and Sekine (2009) [60]. The nal moment magnitude of this event is Mw 7 1andthe directions of migration of tremor episodes in each seg- slip amount is up to 25 cm (Suito and Ozawa, 2009 [65]). ment have different tendencies. Major tremor episodes in The Tokai SSE is also observed to occur repeatedly. Ya- the Kii segment propagate mainly southwestwards from mamoto et al. (2005) [66] reported a similar change in tilt the vicinity of Ise Bay (Obara, 2010 [61]), whereas in the from 1988 to 1990. Therefore, the recurrence interval of Tokai segment, an eastward migration of tremor from Ise the long-term SSEs may be about 10 years; however, no Bay is frequently observed. This suggests that the plate long-term SSEs have been recorded in this region since interface in the vicinity of Ise Bay may rupture initially 2005. Therefore, either the recurrence intervals and du- owing to stress loading caused by stable sliding. rations of these long-term SSEs have large fluctuations or In contrast, the tremor episode of January 2006 initi- these events do not have fixed properties. ated in the southern part of the Kii segment and migrated northeastwards toward Ise Bay. This episode then propa- 2.4. Recurrence Intervals of Boso SSEs gated across Ise Bay and continued to migrate eastwards East of the Boso Peninsula, Japan, SSEs occurred in through the Tokai segment (figure 3 of Obara and Sekine, 1983, 1990, 1996, 2002, 2007, 2011, and 2014 (e.g., Hi- 2009 [60]). Tremor seismicity is low in the area of Ise rose et al., 2012 [22]). The recurrence interval of these Bay; however, the tremor in the Tokai segment is initi- events is 6-7 years and the duration is about one week. ated along the extrapolated path of migration of tremors Therefore, the Boso SSEs are categorized as short-term in the Kii segment. Crustal deformation measured by tilt- SSEs based on their short durations; however, the recur- meters suggests that the short-term reverse-faulted SSEs rence interval is similar to that of long-term SSEs. The propagate laterally in a continuous fashion along the belt- moment magnitude of the Boso SSEs varies from Mw6.3 shaped zone of tremor, including the Ise Bay area. This to 6.8 and the slip length is about 10 cm. The depth of the suggests that the plate interface in the area of Ise Bay is fault causing the SSEs is 10-20 km whereas other long- not a stable sliding region, but is able to accumulate a cer- and short-term SSEs in Nankai are located at depths of 20- tain amount of slip deficit, and short-term SSEs are trig- 40 km and 30-50 km, respectively. The recurrence inter- gered by the slip front approaching this region. The con- val of the Boso SSEs is rather variable compared with that nected nature of the rupture zone of SSEs, covering two of the short-term SSEs with tremor. In particular, the 2011 segments, may be controlled by the direction of propaga- SSE occurred four years after the previous SSE in 2007. tion of the rupture of slow earthquakes and by the amount This shortening of the recurrence interval from 6-7 years of slip deficit built up at the boundary between the seg- is interpreted as an effect of the 2011 M9.0 Tohoku earth- ments. This connected rupturing of neighboring segments quake, whereby the static stress change caused by that is very similar to the multisegment seismic ruptures of earthquake hastened the reverse fault slip along the Philip- megathrust earthquakes. pine Sea Plate in this region (Hirose et al., 2012 [22]). Moreover, the 2014 SSE occurred in January 2014 two 2.3. Recurrence Intervals of Long-Term SSEs years after the previous SSE in 2011 (Ozawa, 2014 [67]). Long-term SSEs are characterized by recurrence in- tervals longer than those of short-term SSEs and much 2.5. Shallow VLF Earthquakes shorter than those of updip megathrust earthquakes. In The slow earthquakes described in Section 2.1-2.4 are the Bungo Channel, southwestern Japan, long-term SSEs located on the downdip side of the seismogenic zone were detected by GPS in 1997, 2003, and 2010 (Hirose of large earthquakes. However, the shallow VLF earth- et al., 1999 [3]; Ozawa et al., 2004 [62]; Hirose et al., quakes occur episodically in clusters along the Nankai 2010 [25]; Ozawa et al., 2013 [63]) between the tremor Trough (Obara and Ito, 2005 [24]; Ito and Obara, zone and the seismogenic zone of the Nankai megath- 2006 [47]). Asano et al. (2008) [23] reported another

322 Journal of Disaster Research Vol.9 No.3, 2014 Contribution of Slow Earthquake Study for Assessing the Occurrence Potential of Megathrust Earthquakes

episode of shallow VLF earthquake activity near the junc- noted that minor tremor episodes are concentrated on the tion of the Japan and Kuril trenches off the Tokachi re- downdip edge of the tremor zone and occur frequently, gion, Hokkaido. This VLF seismicity was dramatically whereas the shallowest tremor episodes are infrequent. activated shortly after a thrust-type earthquake of Mw8. in Wech and Creager (2011) [68] demonstrated a shortening 2003, and then continued for a period with a gradual de- of the recurrence interval of tremor episodes with increas- crease in activity over time. The time sequence curve of ing depth. The recurrence intervals of tremor activity (as the cumulative number of VLF earthquakes, as shown in discussed in Section 2.1) reflect the updip tremor activ- figure 3 of Asano et al. (2008) [23], is very similar to the ity. Major ETS events usually initiate from the deeper pattern of crustal deformation caused by the afterslip fol- side of a tremor zone and then migrate upwards. There- lowing the 2003 earthquake. This similarity suggests that fore, a major ETS event ruptures the entire depth range of the shallow VLF earthquakes occur within or next to the the transition zone, whereas minor episodes are limited to afterslip fault plane on the plate interface. This is consis- rupturing the downdip portion. tent with the results of the analysis of oceanbottom seis- If is it assumed that frictional strength decreases with mometer observations in the Nankai Trough region (Su- depth, then minor episodes of tremor occur frequently on gioka et al., 2012 [48]), which reveal the shallow VLF the downdip edge of the tremor zone because of the stress earthquakes to be caused by slow seismic slip along the concentration due to the deeper stable sliding and low plate interface. On the basis of these recent studies, it is frictional strength. The deeper, minor episode transfers expected that the level of activity of shallow VLF earth- the stress to the updip portion; however, the episode stops quake seismicity may be affected by the accumulation of at around mid-way up the tremor zone if the accumulated stress in the seismogenic zone because of the neighboring stress does not exceed the frictional strength. This fractal- locations of the seismogenic zone and the locus of VLF like stress transfer process controls the ETS rupture in the earthquakes. Also, temporal correspondence between the transition zone. afterslip sequence and shallow VLF earthquake seismic- If we consider the megathrust earthquake, long-term ity in the Tokachi-oki region indicates that shallow VLF SSE and ETS, the recurrence interval becomes shorter earthquakes along the Nankai Trough are triggered by with increasing distance from the seismogenic zone to- SSEs, as are deeper ETS phenomena. To date, no regu- wards the stable sliding zone. The systematic variation in lar recurrence interval has been measured for the shallow the recurrence interval of ETS activity with depth agrees VLF earthquake activity. well with the general trend of the properties of all slip phenomena along the direction of dip of the plate inter- face (Obara et al., 2010 [41]). Therefore, it is expected 3. Interactions Between Slow Earthquakes and that major ETS events load stress on the updip long-term Megathrust Earthquakes SSE source fault and/or on the seismogenic zone. How- ever, the relationship between ETSs and long-term SSEs As slow earthquakes are located in both the downdip is opposite to this situation, as discussed below in Sec- and updip sides of the seismogenic zone on the plate in- tion 3.1.2. terface, some interaction is expected between these slow earthquakes and megathrust events, including precursory changes in slow earthquake activity prior to the occur- 3.1.2. Long-Term SSEs and Tremors in the Bungo rence of megathrust earthquakes. In fact, there is no Channel obvious interaction between these two types of event. In the Bungo Channel, long-term SSEs, which recur However, some examples of interaction between differ- at intervals of 6-7 years and with durations of one year, ent types of slow earthquakes have been clearly observed, affect the activity style of ETSs. In western Shikoku, and if the mechanism of the interaction between different the regular recurrence interval of ETSs is about six types of slow earthquakes can be understood, then such months, but this interval was shortened to a period of an understanding may be able to be applied to the study three months during the 2003 long-term SSE (Hirose and of the interaction between slow and megathrust earth- Obara, 2005 [69]). Also, the tremor activity in the Bungo quakes. In this section, the interaction between differ- Channel persisted for a few months during a period of ent slow earthquakes is discussed, following which some rapid slip (Obara and Hirose, 2006 [70]), whereas tremor examples are given of the relationship between slow and episodes in that area usually occur at an interval of a few megathrust earthquakes. months. The persistent tremor activity is characterized by the frequent occurrence of minor tremor episodes with an 3.1. Interactions Between Different Types of Slow interval of a few days at the updip edge of the tremor zone. Earthquakes Obara et al. (2010) [41] showed that the response of the tremor activities to the long-term SSEs varies with depth 3.1.1. Depth Dependency of Tremor Activity Within (i.e., with distance of the tremors from the source fault of the ETS Zone the long-term SSE). In the updip part of the tremor zone, Tremor activity is characterized not only by along- the cumulative number of tremor activities is closely re- strike segmentation but also by depthdependent proper- lated to the GPS displacement caused by the long-term ties. Wech et al. (2009) [46] and Obara et al. (2010) [41] SSE; however, tremor activity in the deeper part of the

Journal of Disaster Research Vol.9 No.3, 2014 323 Obara, K.

tremor zone is very stable and does not seem be affected data (Suito and Ozawa, 2009 [65]). During this stage by the occurrence of the long-term SSE. of higher long-term SSE activity, tremor episodes and If the initiation of tremor episodes and the occurrence short-term SSEs also frequently occurred (Kobayashi et of long-term SSEs are compared, it is observed that ma- al., 2006 [72]; Obara, 2010 [61]). If the tremor activity jor tremor episodes start at the initiation of the rapidslip is divided into updip and downdip groups, the cumulative stage of long-term SSEs. Although it is difficult to deter- number of tremor activity in the updip group is seen to in- mine what happens prior to the rapid-slip stage, it seems crease significantly compared with that of the downdip that long-term SSEs start a few months before the rapid group. This depth dependence of tremor triggering is slip and involve very minor crustal deformation. There- very similar to the case of tremor activity in the Bungo fore, long-term SSEs are initiated prior to the beginning Channel. However, the correspondence between triggered of tremor activity. tremor seismicity and long-term SSEs is much clearer in the Bungo Channel than it is in Tokai. This difference 3.1.3. Long-Term SSEs and Shallow VLF Earth- may be caused by the difference between the two regions quakes in the slip velocity of the long-term SSEs or in the dis- tance between the tremor source area and the SSE fault. Hirose et al. (2010) [25] recognized another activa- The Bungo Channel SSEs have slip lengths of 10-20 cm tion of shallow VLF earthquakes on the shallower side of over a period of 3-4 months whereas the Tokai SSEs need the long-term SSE source region south of Cape Ashizuri five years to achieve a slip of 30 cm. At present, the re- at the southwestern edge of Shikoku island. The VLF lationship between tremor activity and long-term SSEs is earthquake seismicity is concentrated in clusters along the difficult to define because of the smoothness in the esti- Nankai Trough (Obara and Ito, 2005 [24]). In the Hyuga- mated slip distribution of SSEs derived from GPS analy- nada area, east of Kyushu, there is a cluster with frequent sis. shallow VLF earthquake activity. Shallow VLF earth- Similar relationships between long-term SSEs and quakes activated by long-term SSEs are located next to tremor activity are observed in Mexico and Alaska. the Hyuga-nada VLF cluster and usually show low lev- Husker et al. (2012) [73] identified three types of tremor els of activity. Over the last 10 years, VLF earthquakes activity in Guerrero, Mexico: a deeper “sweet spot” re- have been activated only twice, corresponding to the 2003 gion characterized by persistent activity; intermediate- and 2010 long-term SSEs. The occurrence of VLF earth- deep episodic tremor bursts at an interval of several quakes corresponds to the initiation of the rapid slip stage months; and shallow tremor that were activated during of the long-term SSEs. This temporal correspondence of a period of long-term SSEs in 2006. The “sweet spot” shallow VLF earthquakes and long-term SSEs strongly activity is similar to the frequent activity on the deeper suggests that shallow VLF earthquakes are triggered by side of the tremor zone in southwestern Japan reported by long-term SSEs. However, the source fault of the long- Obara et al. (2010) [41]. The case of shallower tremor term SSEs estimated by GPS is distinct from the source being associated with long-term SSEs is similar to that of area of the triggered, shallow VLF earthquakes. If the triggered tremors recorded in the Bungo Channel. The triggered events are located next to the triggering long- episodic tremor bursts at intermediate depths in Guer- term SSEs, then the source fault of the long-term SSEs rero occurred rather frequently during the 2006 SSEs. must extend towards the Nankai Trough. At present, the Brudzinski et al. (2010) [74] pointed out that tremor ac- spatial extent of the long-term SSEs is not resolved by the tivity in Oaxaca, Mexico, was also activated during a pe- inland GPS network. An oceanbottom measurement sys- riod of SSEs in 2007. The shortening of the recurrence tem may be able to detect the existence of the SSEs. interval is very similar to the case of western Shikoku dur- The triggering effect depends on the location of source ing the 2003 SSEs (Hirose and Obara, 2005 [69]). Peter- faults of VLF earthquakes and long-term SSEs. The son and Christensen (2009) [43] detected tremor-like sig- linked slow earthquakes along the dip direction are lo- nals during SSEs in Alaska compared with periods with- cated west of the largeslip source fault of the 1946 M8.4 out SSEs. The tremor activity, recognized only during the Nankai earthquake. Therefore, these slow earthquakes summer season because of limited data, is located on the may act as a barrier to coseismic rupture. However, Fu- downdip side of the SSE source region, similar to other rumura et al. (2011) [71] proposed that the 1707 Houei areas. These phenomena suggest that tremor activity is included the slow earthquake source easily affected by long-term SSEs that occur next to the region in the Bungo Channel, based on the study of tremor source zone. tsunami deposits and numerical simulation. This suggests On the North Island of New Zealand, tremors have been that frictional properties change with slip velocity. detected in the deeper parts of the plate interface, where long-term SSEs occur (Ide, 2012 [33]). However, because 3.1.4. Other Long-Term SSEs and Tremors there has been no systematic, long-term study in this area Tremor activity triggered by long-term SSEs are ob- of New Zealand, the interaction between tremor activity served not only in the Bungo Channel but also in other and long-term SSEs cannot be defined. regions of the world. In the Tokai area, the frequency of long-term SSEs recorded from late 2000 to 2005 in- creased during 2003 and 2004, as estimated from GPS

324 Journal of Disaster Research Vol.9 No.3, 2014 Contribution of Slow Earthquake Study for Assessing the Occurrence Potential of Megathrust Earthquakes

3.2. Interactions Between Slow Earthquakes and earthquake. A high-quality tremor catalog constructed by Regular Earthquakes Shelly et al. (2009) [79] based on a waveform correlation 3.2.1. Boso technique revealed that the tremor was located at the base of the crust beneath the seismogenic zone, and that they The depths of the SSEs in the Boso area, which are migrated laterally along the San Andreas Fault (Shelly, shallower than those of the Nankai subduction zone, are 2009 [80]). The bidirectional along-strike migration was always associated with seismic swarms of regular earth- observed during 2001 and 2008, except in the few months quakes. The seismic swarms include repeating earth- before the Parkfield earthquake. During the three months quakes (Kimura et al., 2010 [75]), interpreted as seismic preceding the 2004 earthquake, there were no north-south rupture on isolated, small asperities that are surrounded migration episodes, and all episodes migrated towards by slow slip fault planes (Uchida et al., 2003 [76]). There- the south from just beneath the point of rupture initia- fore, repeaters are located in the transition zone between tion of the Parkfield earthquake. After the Parkfield earth- the SSE source fault and the brittle-ductile transition. quake, the pattern of bidirectional along-strike migration Kimura et al. (2010) [75] found that these repeaters are of tremor resumed. This described change in the migra- located at an intraslab discontinuity a few kilometers be- tion pattern may reflect the accumulation of stress in the low the surface of the subducting Philippine Sea Plate. seismogenic zone. These changes in tremor activity be- The discontinuity is interpreted to be caused by underplat- fore the occurrence of large earthquake have not been rec- ing through subduction-related processes. Therefore, the ognized in either southwestern Japan or Cascadia; how- Boso SSEs are considered to occur along the fault con- ever, some change in the activity style of slow earthquakes necting the updip plate interface with the downdip under- may be expected to occur prior to a megathrust earthquake plating discontinuity. In other words, the SSEs may rep- in both of these areas. resent deformation within the top layer of the subducting oceanic plate. The triggering of the seismic swarm on the downdip side of the SSE source area may be caused by 3.3.2. Simulation of Megathrusts and Slow Earth- the concentration of stress by slip movement. quakes Although changes in tremor activity before a megath- 3.2.2. New Zealand rust earthquake along a subducting plate interface have Off the east coast of the North Island of New Zealand, not yet been recorded, it is possible to simulate the SSEs similar to those of the Boso area occur frequently in interaction between these phenomena. Matsuzawa et some regions. The similarities include both the shallow al. (2010) [81] used simulation to reproduce interplate depth and the association with regular earthquake seismic slip events including megathrust earthquakes, long-term swarms. Wallace et al. (2010) [38] reported a clear exam- SSEs, and short-term SSEs based on the assumption of a ple of triggered seismic swarms at the downdip edge of rate- and state-dependent friction law with cutoff veloc- the SSE region in Hawke Bay. The triggering mechanism ities and a distribution of pore fluid that controls the re- of the swarms may be the same as that of the swarms in currence interval of both types of SSE. With appropriate Boso. Kim et al. (2011) [34] described tremor activity in parameterization, the simulation explained the recurrence the downdip part of the SSE source area. However, the intervals of megathrust earthquakes and of both types of tremor-like signals in this area are not particularly clear, SSE. In addition, using a two-dimensional model of the and there is a need to accumulate more reliable evidence subduction zone, a shortening of the recurrence interval to constrain the relationship between tremors and shallow of both types of SSE was recognized in the latter stage SSEs in the vicinity of North Island, New Zealand. of the period between mega-earthquakes. This is inter- preted as a transient slip developing gradually and repre- 3.3. Response of Slow Earthquake Activity to Large senting the locus of nucleation between the source areas Earthquakes of SSEs and those of large earthquakes. Therefore, an ac- celeration in transient slip may promote the occurrence of 3.3.1. Tremors Along the San Andreas Fault SSEs and lead to a shortening of their recurrence inter- Tremor activity along the San Andreas Fault has been val, which in turn may favor the occurrence of megathrust closely studied since Nadeau and Dolenc (2005) [37] dis- earthquakes. However, the change in the recurrence in- covered tremor activity near Parkfield, the site of a Mw6.0 terval of SSEs over time shows more variation in a three- earthquake in 2004. Some papers have reported precur- dimensional model than in a twodimensional model, as sory changes in tremor activity preceding the 2004 Park- shown in figure 10 of Matsuzawa et al. (2010) [81]. This field earthquake. Nadeau and Guilhem (2009) [77] in- suggests that a shortening of the recurrence interval may vestigated the temporal properties of tremor activity both be useful only for the long-term evaluation of megath- before and after the 2004 Parkfield earthquake. High- rust earthquakes, as short-term prediction is very diffi- level tremor activity after the earthquake is interpreted as cult because of the high variation in recurrence interval. continuing afterslip (Brenguiar et al., 2008 [78]). How- This result is just one of many possible scenarios derived ever, the tremor was activated three weeks before the from a set of simple modeling rules and parameters, and occurrence of the earthquake, and therefore the tremor other scenarios might be possible using other parameter- activity represented precursory phenomena for the 2004 izations. Therefore, the precise monitoring of these slow

Journal of Disaster Research Vol.9 No.3, 2014 325 Obara, K.

earthquakes and the exploration of the subsurface struc- currently too limited to reveal the entire distribution of ture in and around slow earthquake source regions are frictional properties. Therefore, the detection of various very important for providing tighter constraints on sim- slow earthquakes will help to constrain the inhomoge- ulations to produce more realistic models. neous distribution of frictional properties and to delineate the seismogenic zone. The interaction between different slip phenomena rep- 3.4. Triggering of Megathrust Earthquakes by Slow resents the response of the triggered phenomena to the Earthquakes static or dynamic stress transfer caused by the triggering Slow earthquakes have been detected mainly in warm event. Therefore, it is expected that information about subduction zones characterized by subducting young the frictional properties of the plate interface could be oceanic plate. In southwest Japan, tremor is widely dis- obtained based on the behavior of triggered slow earth- tributed over a length of 600 km along the subducting, quakes because the temporal changes and spatial distri- warm Philippine Sea Plate. On the other hand, except af- bution of triggered event activity must be controlled by terslip, no slow earthquakes had been detected for many frictional parameters and friction law. An understanding years in eastern Japan where the cold Pacific Plate is of frictional properties will be valuable in explaining the subducting. However, ocean-bottom pressure gauges de- mechanisms of interactions between different slow earth- tected SSEs in 2008 and 2011 close to the locus of rup- quakes. Therefore, we need to clarify space-time evolu- ture initiation of the 2011 M9.0 Tohoku earthquake in tion of the source process of the triggering slow earth- the shallower part of the plate interface near the Japan quake as an input event and the triggered slow earthquake Trench, where the Pacific Plate is subducting (Ito et al., as a response event. The interaction mechanisms could 2013 [29]). Kato et al. (2012) [28] reported that two SSEs be used to explain megathrust earthquakes in terms of the occurred before the Tohoku earthquake and that the rup- effect of such interactions on the behavior of megathrust ture front of the SSEs propagated towards the of source faults. the Tohoku earthquake twice, based on the study of fore- According to numerical simulation studies, the friction shock activity, including repeaters. Just after the slip front law with slipvelocity dependence is able to explain the reached close to the locus of rupture initiation of the To- slip properties of slow earthquakes. Recently, material hoku earthquake, the earthquake started to rupture, with experiments on core samples obtained from deep ocean- a maximum slip of nearly 50 m. Therefore, SSEs may bottom drilling have provided evidence for rapid rupture trigger megathrust earthquakes if the prior stress build- near the Nankai Trough, where stable sliding or slow slip up in the seismogenic zone reaches a critical level. The were expected (Sakaguchi et al., 2011 [83]). The region second SSE detected by Kato et al. (2012) [28] is some- near the off northeastern Japan, where SSEs times considered as an afterslip caused by Mw7.3 fore- occurred in 2008 and 2011 (Ito et al., 2013 [29]), rup- shock two days before the Tohoku earthquake (Miyazaki tured seismically during the large slip of the 2011 Tohoku et al., 2011 [82]). However this transient slip phenom- earthquake. This indicates that friction changes consider- ena might be different from ordinary afterslip because the ably according to slip velocity. Understanding the slipve- slip rate is higher than that of other afterslip events and it locity dependence of frictional properties is important, is characterized by clear migration front associated with not only for understanding the occurrence mechanisms of seismic events. In either case, the slow earthquake in the both slow and megathrust earthquakes, but also for eval- broad sense may interact to the occurrence of huge earth- uating tsunami hazard based on the precise estimation of quake. slip length during a megathrust earthquake. After the 2011 Tohoku earthquake, the Japanese gov- ernment evaluated the likely maximum magnitude of a 4. Discussion and Conclusions possible huge earthquake in the Nankai Trough region as being M9.0 (Cabinet Office, Government of Japan, A subducting plate interface contains numerous 2012 [84]). The maximum zone of rupture includes the patches with distinctive slip properties. The differences belt-shaped tremor zone. Tremor is a type of seismic sig- between the slip velocities of ruptures on each separate nal, and the tremor zone may have the potential to radiate patch may reflect variations in frictional properties, which strong seismic signals. Therefore, there is a need to evalu- may in turn reflect inhomogeneous distributions of both ate the contribution of the tremor zone to the generation of pore fluid pressure and material content at the interface strong motion. Clarification of the frictional properties of between the subducting and overriding plates. Quantify- the tremor source zone is needed to allow the behavior of ing the inhomogeneous distribution of frictional proper- the tremor zone as a response to the large and rapid slip in ties based on the study of seismic structure and on ex- the updip firmly locked zone to be evaluated. Moreover, periments of material properties is required if the occur- the accumulation of slip deficit at the tremor zone must rence mechanisms of slip phenomena, including megath- affect the generation of strong motion when a huge earth- rust earthquakes, are to be understood. Knowledge of the quake occurs. However, the tremor zone usually releases distribution of frictional properties should prove highly strain energy through the periodic occurrence of short- useful for assessing future seismic activity. However, the term SSEs. Therefore, the monitoring of the spatiotempo- data on the frictional properties of the plate interface are ral distribution of slip is required for a proper evaluation

326 Journal of Disaster Research Vol.9 No.3, 2014 Contribution of Slow Earthquake Study for Assessing the Occurrence Potential of Megathrust Earthquakes

of strong motion. Slow earthquakes have been discovered by a couple The accumulation of stress in firmly locked zones may of breakthroughs. One of the breakthroughs is develop- affect slow earthquake activity, according to both the ment of innovative observation technologies like as GPS. observations in Parkfield along the San Andreas Fault Another is densely distributed observation networks like and the results of numerical simulations. Ariyoshi et al. as GEONET (Sagiya, 2004 [2]) or Hi-net (Okada et al., (2012) [85] quantified the temporal change in migration 2004 [13]). The station spacing of these networks be- velocity of slow earthquakes leading up to the occurrence came one third of the previous observation network. If of a huge earthquake in the simulation study. To ver- we have more dense observation network, we might have ify the results of numerical simulation studies with ob- the next breakthrough for the earthquake science includ- servations, accurate, precise, and continuous monitoring ing improvement in long-term predictions of megathrust of slow earthquake activity is required. Matsuzawa et earthquakes. al. (2013) [86] conducted their numerical simulations with the assumption of an inhomogeneous distribution of friction parameters according to the tremor locations and Acknowledgements three-dimensional geometry of the Philippine Sea Plate I am very grateful to two anonymous reviewers for helpful com- obtained from receiverfunction analysis (Shiomi et al., ments. This review paper is based on studies on slow earthquakes 2008 [87]). The depth dependency of tremor activity ob- collaborated with NIED and ERI colleagues. I would like to ex- served by Obara et al. (2010) [41] was able to be ex- press my sincere thanks to the collaborators. We used Generic plained by the simulation. Moreover, the simulation suc- Mapping Tools (Wessel and Smith, 1998 [88]) for drawing the maps. ceeded in reproducing a newly detected long-term SSE in the southern part of Shikoku, which had been discovered by Kobayashi (2012) [20] based on an analysis of level- References: ing data. This suggests that adequate parameter setting [1] I. Kawasaki, Y. Asai, Y. Tamura, T. Sagiya, N. Mikami, Y. Okada, in such modeling may reasonably reproduce various ob- M. Sakata, and M. Kasahara, “The 1992 Sanriku-Oki, Japan, ultra- slow earthquake,” J. Phys. Earth., Vol.43, pp. 105-116, 1995. served phenomena. The study of subsurface structure and [2] T. Sagiya, “A decade of GEONET: 1994-2003 – The continuous the monitoring of slow earthquakes are needed to help GPS observation in Japan and its impact on earthquake studies –,” constrain the distribution of frictional properties and to Earth, Planets and Space, Vol.56, pp. xxix-xli, 2004. [3] H. Hirose, K. Hirahara, F. Kimata, N. Fujii, and S. Miyazaki, “A improve the resolution of simulations. slow thrust slip event following the two 1996 Hyuganada earth- Slow earthquakes have the potential to transfer stress quakes beneath the Bungo Channel, southwest Japan,” Geophys. to surrounding regions, including onto the megathrust Res. Lett., Vol.26, pp. 3237-3240, 1999. [4] S. Ozawa, M. Murakami, M. Kaidzu, T. Tada, T. Sagiya, Y. source fault. If the locked zone reaches a critical accumu- Hatanaka, H. Yarai, and T. Nishimura, “Detection and monitoring lation of stress, low stress loading caused by slow earth- of ongoing aseismic slip in the Tokai region, central Japan,” Sci- ence, Vol.298, pp. 1009-1012, 2002. quakes may trigger the large earthquake as the final stim- [5] S. Miyazaki, P. Segall, J. J. McGuire, T. Kato, and Y. Hatanaka, ulation. This triggering effect must be very rare, based on “Spatial and temporal evolution of stress and slip rate during the 2000 Tokai slow earthquake,” J. Geophys. Res., Vol.111, B03409, the consideration of the recurrence intervals of megath- doi:10.1029/2004JB003426, 2006. rust earthquakes and of slow earthquakes. If the accu- [6] Y. Ohta, J. T. Freymueller, S. Hreinsd´ottir, and H. Suito, “A large mulated strain energy and failure strength at the asper- slow slip event and the depth of the seismogenic zone in the south central Alaska subduction zone,” Earth Planet. Sci. Lett., Vol.247, ity of the large earthquake are known, the occurrence of pp. 108-116, 2006. the slow earthquake next to the seismogenic zone may be [7] M. Vergnolle, A. Walpersdorf, V. Kostoglodov, P. Tregoning, J. A. Santiago, N. Cotte, and S. I. Franco, “Slow slip events in Mexico useful for the short-term prediction of large earthquakes. revised from the processing of 11 year GPS observations,” J. Geo- Furthermore, slow earthquakes are very sensitive to slight phys. Res., Vol.115, B08403, doi:10.1029/2009JB006852, 2010. stress changes, and therefore useful information should be [8] H. Dragert, K. Wang, and T. S. James, “A silent slip event on the deeper Cascadia subduction interface,” Science, Vol.292, pp. 1525- able to be gained from the change in activity style of slow 1528, 2001. earthquakes in relation to the change in stress field at the [9] M. M. Miller, T. Melbourne, D. J. Johnson, and W. Q. Sumner, locked zone on the plate interface. At present, actually “Periodic slow earthquakes from the Cascadia subduction zone,” Science, Vol.295, No.2423, 2002. we don’t know what kind monitoring is effective to detect [10] K. Satake, K. Wang, and B. F. Atwater, “Fault slip and seismic the change in stress status at the locked zone. We expect moment of the inferred from Japanese tsunami descriptions,” J. Geophys. Res., Vol.108, No.2535, that not only the recurrence interval of slow earthquakes doi:10.1029/2003JB002521, 2003. but also the temporal changes in tidal response of tremor, [11] K. Obara, “Nonvolcanic deep tremor associated with subduction in triggered tremor activity by large amplitude surface wave, southwest Japan,” Science, Vol.296, pp. 1679-1681. 2002. migration pattern including speed during the major tremor [12] K. Obara and K. Shiomi, “Underground Structural Anomalies and Slow Earthquake Activities Around Seismogenic Megathrust Earth- episode might play rules as indicator for the stress status. quake Zone as Revealed by Inland Seismic Observations,” J. Disas- Therefore, the construction of the database based on the ter Research, Vol.4, pp. 83-93, 2009. [13] Y. Okada, K. Kasahara, S. Hori, K. Obara, S. Sekiguchi, H. Fuji- precise monitoring of slow earthquakes is very useful to wara, and A. Yamamoto, “Recent progress of seismic observation extract the above information. Of course, there might be networks in Japan – Hi-net, F-net, K-NET and KiK-net –,” Earth, many unrecognized indicators for the stress status. The Planets and Space, Vol.56, xv–xxviii, 2004. [14] K. Obara, K. Kasahara, S. Hori, and Y. Okada, “A densely database might be available for the further study. There- distributed high-sensitivity seismograph network in Japan: hi- fore, degradation of the quality in the database is not al- net by national research institute for earth science and disaster prevention,” Review of Scientific Instruments, Vol.76, 021301. lowed. doi:10.1063/1.1854197, 2005.

Journal of Disaster Research Vol.9 No.3, 2014 327 Obara, K.

[15] G. Rogers and H. Dragert, “Episodic tremor and slip on the Casca- [37] R. M. Nadeau and D. Dolenc, “Nonvolcanic tremors deep dia subduction zone: The chatter of silent slip,” Science, Vol.300, beneath the San Andreas Fault,” Science, Vol.307, No.389, pp. 1942-1943, 2003. doi:10.1126/science.1107142, 2005. [16] S. Miyazaki, P. Segall, J. Fukuda, and T. Kato, “Space time distribu- [38] L. M. Wallace and J. Beavan, “Diverse slow slip behavior at the tion of afterslip following the 2003 Tokachi-oki earthquake: Impli- Hikurangi subduction margin, New Zealand,” J. Geophys. Res., cations for variations in fault zone frictional properties,” Geophys. Vol.115, B12402, doi:10.1029/2010JB007717, 2010. Res. Lett., Vol.31, L06623, doi:10.1029/2003GL019410, 2004. [39] Y. Jiang, S. Wdowinski, T. H. Dixon, M. Hackl, M. Protti, and [17] S. Miura, T. Iinuma, S. Yui, N. Uchida, T. Sato, K. Tachibana, and V. Gonzalez, “Slow slip events in Costa Rica detected by contin- A. Hasegawa, “Co- and post-seismic slip associated with the 2005 uous GPS observations, 2002-2011,” Geochem. Geophys. Geosys., Miyagi-oki earthquake (M7.2) as inferred from GPS data,” Earth Vol.13, No.4, doi: 10.1029/2012GC004058, 2012. Planets Space, Vol.58, pp. 1567-1572, 2006. [18] H. Suito, T. Nishimura, M. Tobita, T. Imakiire, and S. Ozawa, “In- [40] J. I. Walter, S. Y. Schwartz, M. Protti, and V. Gonzalez, “The syn- terplate fault slip along the Japan Trench before the occurrence of chronous occurrence of shallow tremor and very low frequency the 2011 off the Pacific coast of Tohoku Earthquake as inferred from earthquakes offshore of the Nicoya Peninsula, Costa Rica,” Geo- GPS data,” Earth, Planets and Space, Vol.63, pp. 615-619, 2011. phys. Res. Lett., Vol.40, pp. 1517-1522, doi:10.1002/grl.50213, 2013. [19] Y. Yagi and M. Kikuchi, “Partitioning between seismogenic and aseismic slip as highlighted from slow slip events in Hyuga- [41] K. Obara, S. Tanaka, T. Maeda, and T. Matsuzawa, “Depth- nada, Japan,” Geophys. Res. Lett., Vol.30, No.2, p. 1087, dependent activity of non-volcanic tremor in southwest Japan,” doi:10.1029/2002GL015664, 2003. Geophys. Res. Lett., Vol.37, L13306, doi:10.1029/2010GL043679, [20] A. Kobayashi, “Long-term slow slip event around Kochi City from 2010. 1978 to 1980,” J. Seismol. Soc. Japan (Zisin), Second Series, [42] D. Zigone, D. Rivet, M. Radiguet, M. Campillo, C. Voisin, N. Vol.64, pp. 63-73, doi:10.4294/zisin.64.63, 2012 (in Japanese with Cotte, A. Walpersdorf, N. M. Shapiro, G. Cougoulat, P. Roux, V. English abstract and figure captions). Kostoglodov, A. Husker, and J. S. Payero, “Triggering of tremors [21] H. Yarai and S. Ozawa, “Quasi-periodic slow slip events in the after- and slow slip event in Guerrero, Mexico, by the 2010 Mw8.8 slip area of the 1996 Hyuga-nada earthquakes, Japan,” J. Geophys. Maule, Chile, earthquake,” J. Geophys. Res., Vol.117, B09304, doi: Res., Vol.118, p. 1801, doi:10.1002/jgrb.50161, 2013. 10.1029/2012JB009160, 2012. [22] H. Hirose, H. Kimura, B. Enescu, and S. Aoi, “Recurrent slow slip [43] C. Peterson and D. Christensen, “Possible relationship be- event likely hastened by the 2011 Tohoku earthquake,” Proc. Na- tween non-volcanic tremor and the 1998-2001 slow-slip event, tional Academy Sciences USA, Vol.109, No.38, pp. 15157-15161, south central Alaska,” J. Geophys. Res, Vol.114, B06302, doi: 10.1073/pnas.1202709109, 2012. doi:10.1029/2008JB006096, 2009. [23] Y. Asano, K. Obara, and Y. Ito, “Spatiotemporal distribution of [44] K. Obara, “Characteristics and interactions between non-volcanic very-low frequency earthquakes in Tokachi-oki near the junction tremor and related slow earthquakes in the Nankai subduction zone, of the Kuril and Japan trenches revealed by using array signal pro- southwest Japan,” J. Geodyn., Vol.52, No.3-4, pp. 229-248, doi: cessing,” Earth, Planets and Space, Vol.60, pp. 871-875, 2008. 10.1016/ j.jog.2011.04.002, 2011. [24] K. Obara and Y. Ito, “Very low frequency earthquake excited by the [45] Headquarters for Earthquake Research Promotion, “On the Long- 2004 off the Kii peninsula earthquake: A dynamic deformation pro- Term Evaluation of Earthquakes in the Nankai Trough,” 2001 cess in the large accretionary prism,” Earth Planets Space, Vol.57, (in Japanese), http://www.jishin.go.jp/main/chousa/01sep nankai/ pp. 321-326, 2005. f03.htm [accessed May 18, 2014] [25] H. Hirose, Y. Asano, K. Obara, T. Kimura, T. Matsuzawa, S. Tanaka, and T. Maeda, “Slow earthquakes linked along dip in the [46] A. Wech, K. C. Creager, and T. I. Melbourne, “Seismic and geode- Nankai subduction zone,” Science, Vol.330, No.1502, 2010. tic constraints on Cascadia slow slip,” J. Geophys. Res., Vol.114, B10316, doi:10.1029/2008JB006090, 2009. [26] K. Obara, H. Hirose, F. Yamamizu, and K. Kasahara, “Episodic slow slip events accompanied by non-volcanic tremors in south- [47] Y. Ito and K. Obara, “Dynamic deformation of the accretionary west Japan subduction zone,” Geophys. Res. Lett., Vol.31, L23602, prism excites very low frequency earthquakes,” Geophys. Res. Lett., doi:10.1029/2004GL020848, 2004. Vol.33, L02311, doi:10.1029/2005GL025270, 2006. [27] Y. Ito, K. Obara, K. Shiomi, S. Sekine, and H. Hirose, “Slow earth- [48] H. Sugioka, T. Okamoto, T. Nakamura, Y. Ishihara, A. Ito, K. quakes coincident with episodic tremors and slow slip events,” Sci- Obana, M. Kinoshita, K. Nakahigashi, M. Shinohara, and Y. Fukao, ence, Vol.315, pp. 503-506, 2007. “Tsunamigenic potential of the shallow subduction plate bound- [28] A. Kato, K. Obara, T. Igarashi, H. Tsuruoka, S. Nakagawa, ary inferred from slow seismic slip,” Nature Geosci., Vol.5, doi: and N. Hirata, “Propagation of slow slip leading up to the 10.1038/NGEO1466, 2012. 2011 Mw9.0 Tohoku-Oki Earthquake,” Science, Vol.335, No.705, [49] J. S. Payero, V. Kostoglodov, N. Shapiro, T. Mikumo, A. Igle- doi:10.1126/science.1215141, 2012. sia, X. Perez-Campos, and R. W. Clayton, “Nonvolcanic tremor [29] Y. Ito, R. Hino, M. Kido, H. Fujimoto, Y. Osada, D. Inazu, Y. Ohta, observed in the Mexican subduction zone,” Geophys. Res. Lett., T. Iinuma, M. Ohzono, S. Miura, M. Mishina, K. Suzuki, T. Tsuji, Vol.35, L07305, doi:10.1029/2007GL032877, 2008. and J. Ashi, “Episodic slow slip events in the Japan subduction zone [50] J. L. Rubinstein, M. La Rocca, J. E. Vidale, K. C. Creager, and A. G. before the 2011 Tohoku-Oki earthquake,” Tectonophysics, Vol.600, Wech, “Tidal modulation of nonvolcanic tremor,” Science, Vol.319, pp. 14-26, doi:10.1016/j.tecto.2012.08.022, 2013. pp. 186-189, doi:10.1126/science.1150558, 2008. [30] T. Maeda and K. Obara, “Spatio-temporal distribution of seis- mic energy radiation from low-frequency tremor in west- [51] R. Nakata, N. Suda, and H. Tsuruoka, “Non-volcanic tremor result- ern Shikoku, Japan,” J. Geophys. Res., Vol.114, B00A09, ing from the combined effect of Earth tides and slow slip events,” doi:10.1029/2008JB006043, 2009. Nature Geosci., Vol.1, pp. 676-678, 2008. [31] Z. Peng and K. Chao, “Non-volcanic tremor beneath the Central [52] M. Miyazawa and J. Mori, “Evidence suggesting fluid flow beneath Japan due to periodic seismic triggering from the 2004 - Range in Taiwan triggered by the 2001 Mw7.8 Kunlun earthquake,” Geophys. J. Int., Vol.175, pp. 825-829. 2008. Andaman earthquake,” Geophys. Res. Lett., Vol.33, L05303, [32] K. Chao, Z. Peng, H. Gonzalez-Huizar, C. Aiken, B. Enescu, H. doi:10.1029/2005GL025087, 2006. Kao, A. Wech, A. Velasco, K. Obara, and T. Matsuzawa, “A Global [53] J. L. Rubinstein, J. E. Vidale, J. Gomberg, P. Bodin, K. C. Creager, Search for Triggered Tremor Following the 2011 Mw9.0 Tohoku and S. D. Malone, “Non-volcanic tremor driven by large transient Earthquake,” Bulletin of the Seismological Society of America, shear stresses,” Nature, Vol.448, pp. 579-582, 2007. Vol.103, pp. 1551-1571, doi: 10.1785/0120120171, 2013. [54] K. Chao, Z. Peng, Y.-J. Hsu, K. Obara, H.-C. Pu, C. Wu, A. Wech, [33] S. Ide, “Variety and spatial heterogeneity of tectonic and P.-L. Leu, “Increasing of Tremor Rate Correlated with Possible tremor worldwide,” J. Geophys. Res., Vol.117, B03302, Slow Slip Preceding a Nearby Mw6.3 Earthquake in Taiwan,” 2014 doi:10.1029/2011JB008840, 2012. (in preparation). [34] M. J. Kim, S. Y. Schwartz, and S. Banister, “Nonvolcanic tremor [55] M. Ando, “Source mechanisms and tectonic significance of his- associated with the March 2010 Gisborne slow slip event at the torical earthquakes along the Nankai trough, Japan,” Tectonophys., Hikurangi subduction margin, New Zealand, Geophys. Res. Lett., Vol.27, pp. 119-140, 1975. Vol.38, L14301, doi:10.1029/2011GL048400, 2011. [56] H. Dragert, K. Wang, and G. Rogers, “Geodetic and seismic signa- [35] A. Gallego, R. M. Russo, D. Comte, V. Mocanu, R. E. Murdie, and tures of episodic tremor and slip in the northern Cascadia subduc- J. C. VanDecar, “Tidal modulation of continuous nonvolcanic seis- tion zone,” Earth Planets Space, Vol.56, pp. 1143-1150, 2004. mic tremor in the Chile triple junction region,” Geochem. Geophys. Geosys., Vol.14, No.4, doi: 10.1002/ggge.20091. 2011. [57] S. Mazzotti and J. Adams, “Variability of Near-Term Probability for [36] J. R. Brown, G. C. Beroza, S. Ide, K. Ohta, D. R. Shelly, S. the Next Great Earthquake on the Cascadia Subduction Zone,” Bull. Y. Schwartz, W. Rabbel, M. Thorwart, and H. Kao, “Deep low- Seismol. Soc. Am., Vol.94, pp. 1954-1959, 2004. frequency earthquakes in tremor localize to the plate interface in [58] A. C. Aguiar, T. I. Melbourne, and C. W. Scrivener, “Moment re- multiple subduction zones,” Geophys. Res. Lett., Vol.36, L19306, lease rate of Cascadia tremor constrained by GPS,” J. Geophys. doi:10.1029/2009GL040027, 2009. Res., Vol.114, doi: 10.1029/2008JB005909, 2009.

328 Journal of Disaster Research Vol.9 No.3, 2014 Contribution of Slow Earthquake Study for Assessing the Occurrence Potential of Megathrust Earthquakes

[59] N. M. Bartlow, S. Miyazaki, A. M. Bradley, and P. Segall, [80] D. R. Shelly, “Possible deep fault slip preceding the 2004 Parkfield “Space-time correlation of slip and tremor during the 2009 Cas- earthquake, inferred from detailed observations of tectonic tremor,” cadia slow slip event,” Geophys. Res. Lett., Vol.38, L18309, Geophys. Res. Lett., Vol.36, L17318, doi:10.1029/2008GL039589, doi:10.1029/2011GL048714, 2011. 2009. [60] K. Obara and S. Sekine, “Characteristic activity and migration of [81] T. Matsuzawa, H. Hirose, B. Shibazaki, and K. Obara, “Model- episodic tremor and slow-slip events in central Japan,” Earth Planets ing short- and long-term slow slip events in the seismic cycles of Space., Vol.61, pp. 853-862, 2009. large subduction earthquakes,” J. Geophys. Res., Vol.115, B12301, [61] K. Obara, “Phenomenology of deep slow earthquake family in doi:10.1029/2010JB007566, 2010. southwest Japan: Spatiotemporal characteristics and segmentation,” [82] S. Miyazaki, J. J. McGuire, and P. Segall, “Seismic and aseismic J. Geophys. Res., Vol.115, B00A25, doi: 10.1029/2008JB006048, fault slip before and during the 2011 off the Pacific coast of Tohoku 2010. Earthquake,” Earth Planets Space, Vol.63, pp. 637-642, 2011. [62] S. Ozawa, Y. Hatanaka, M. Kaidzu, M. Murakami, T. Imakiire, and [83] A. Sakaguchi, F. Chester, D. Curewitz, O. Fabbri, D. Goldsby, G. Y. Ishigaki, “Aseismic slip and low-frequency earthquakes in the Kimura, C. F. Li, Y. Masaki, E. J. Screaton, A. Tsutsumi, K. Ujiie, Bungo channel, southwestern Japan,” Geophys. Res. Lett., Vol.31, and A. Yamaguchi, “Seismic slip propagation to the up-dip end L07609, doi:10.1029/ 2003GL019381, 2004. of plate boundary subduction interface faults: Vitrinite reflectance geothermometry on Integrated Ocean Drilling Program NanTro- [63] S. Ozawa, H. Yarai, T. Imakiire, and M. Tobita, “Spatial and tem- SEIZE cores,” Geology, Vol.39, 395-398, doi:10.1130/G31642.1- poral evolution of the long-term slow slip in the Bungo Channel, 2011, 2011. Japan,” Earth, Planets and Space, Vol.65, pp. 67-73, 2013. [84] Cabinet Office, Government of Japan, 2012, http://www.bousai. [64] A. Kobayashi and T. Yamamoto, “Repetitive long‐term slow slip go.jp/jishin/nankai/taisaku wg/pdf/20130528 houkoku s1.pdf [ac- events beneath the Bungo Channel, southwestern Japan, identified cessed May 18, 2014] from leveling and sea level data from 1979 to 2008,” J. Geophys, [85] K. Ariyoshi, T. Matsuzawa, J. P. Ampuero, R. Nakata, T. Hori, Y. Res., Vol.116, B04406, doi:10.1029/2010JB007822, 2011. Kaneda, R. Hino, and A. Hasegawa, “Migration process of very [65] H. Suito and T. Ozawa, “Transient crustal deformation in the Tokai low-frequency events based on a chain-reaction model and its ap- district,” J. Seismol. Soc. Jpn., Vol.2, No.61, pp. 113-135, 2009 (in plication to the detection of preseismic slip for megathrust earth- Japanese with English abstract). quakes,” Earth, Planets and Space, Vol.64, pp. 693-702, 2012. [66] E. Yamamoto, S. Matsumura, and T. Ohkubo, “A slow slip event in [86] T. Matsuzawa, B. Shibazaki, K. Obara, and H. Hirose, “Comprehen- the Tokai area detected by tilt and seismic observation and its pos- sive model of short- and long-term slow slip events in the Shikoku sible recurrence,” Earth Planets Space, Vol.57, pp. 917-923, 2005. region of Japan,” incorporating a realistic plate configuration, Japan, [67] S. Ozawa, “Shortening of recurrence interval of Boso slow slip Geophys. Res. Lett., Vol.40, pp. 5125-5130, doi:10.1002/grl.51006, events in Japan,” Geophys. Res. Lett., doi:10.1002/ 2014GL060072, 2013. 2014. [87] K. Shiomi, M. Matsubara, Y. Ito, and K. Obara, “Simple relation- ship between seismic activity along Philippine Sea slab and geom- [68] A. Wech, and K. C. Creager, “A continuum of stress, strength and etry of oceanic Moho beneath southwest Japan,” Geophys. J. Int., slip in the Cascadia subduction zone,” Nature Geoscience, Vol.4, Vol.173, pp. 1018-1029, 2008. pp. 624-628, doi:10.1038/ngeo1215, 2011. [88] P. Wessel and W. H. F. Smith, “New, improved version of Generic [69] H. Hirose and K. Obara, “Repeating short- and long-term slow slip Mapping Tools released,” EOS Trans. AGU, Vol.79, No.47, p. 579, events with deep tremor activity around the Bungo channel region, 1998. southwest Japan,” Earth Planets Space, Vol.57, pp. 961-972, 2005. [70] K. Obara and H. Hirose, “Non-volcanic deep low-frequency tremors accompanying slow slips in the southwest Japan subduc- tion zone,” Tectonophys., Vol.417, pp. 33-51, 2006. [71] T. Furumura, K. Imai, and T. Maeda, “A revised tsunami source model for the 1707 Hoei earthquake and simulation of tsunami inundation of Ryujin Lake, Kyushu, Japan,” J. Geophys. Res., Name: Vol.116, B02308, doi: 10.1029/2010JB007918, 2011. Kazushige Obara [72] A. Kobayashi, T. Yamamoto, K. Nakamura, and K. Kimura, “Short- term slow slip events detected by the strainmeters in Tokai region in Affiliation: the period from 1984 to 2005,” J. Seismol. Soc. Jpn., Vol.2, No.59, Professor, Earthquake Research Institute, The pp. 19-18, 2006 (in Japanese with English abstract). University of Tokyo [73] A. L. Husker, V. Kostoglodov, V. M. Cruz-Atienza, D. Legrand, N. M. Shapiro, J. S. Payero, M. Campillo, and E. Huesca-Perez, “Temporal variations of non-volcanic tremor (NVT) locations in the Mexican subduction zone: Finding the NVT sweet spot,” Geochem. Geophys. Geosystems, Vol.13, No.3, doi:10.1029/2011GC003916, 2012. Address: [74] M. R. Brudzinski, H. R. Hinojosa-Prieto, K. M. Schlanser, E. 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-0032, Japan Cabral-Cano, A. Arciniega-Ceballos, O. Diaz-Molina, and C. DeMets, “Nonvolcanic tremor along the Oaxaca segment of the Brief Career: Middle America subduction zone,” J. Geophys. Res., Vol.115, 1985-2010 National Research Institute for Earth Science and Disaster B00A23, doi:10.1029/2008JB006061, 2010. Prevention [75] H. Kimura, T. Takeda, K. Obara, and K. Kasahara, “Seis- 1993-1994 Visiting Scientist, Massachusetts Institute of Technology mic evidence for active underplating below the megath- 2010- Earthquake Research Institute, The University of Tokyo rust earthquake zone in Japan,” Science, Vol.329, 210-212, Selected Publications: doi:10.1126/science.1187115, 2010. • “Nonvolcanic deep tremor associated with subduction in southwest [76] N. Uchida, T. Matsuzawa, T. Igarashi, and A. Hasegawa, “Inter- Japan,” Science, Vol.296, pp. 1679-1681, 2002. plate quasi-static slip off Sanriku, NE Japan, estimated from re- • “A densely distributed high-sensitivity seismograph network in Japan: peating earthquakes,” Geophys. Res. Lett., Vol.30, No.15, p. 1801, Hi-net by national research institute for earth science and disaster doi:10.1029/2003GL017452, 2003. prevention,” Review of Scientific Instruments, Vol.76, [77] R. M. Nadeau and A. Guilhem, “Nonvolcanic Tremor Evolution and 021301.doi:10.1063/1.1854197, 2005. the San Simeon and Parkfield, California,” Earthquakes, Science, • “Very low frequency earthquake excited by the 2004 off the Kii Vol.325, p. 191, doi:10.1126/science.1174155, 2009. peninsula earthquake: A dynamic deformation process in the large [78] F. Brenguiar, M. Campillo, C. Hadziioannou, N. M. Shapiro, R. accretionary prism,” Earth Planets Space, Vol.57, pp. 321-326, 2005. M. Nadeau, and E. Larose, “Postseismic Relaxation Along the San Andreas Fault at Parkfield from Continuous Seismological Obser- Academic Societies & Scientific Organizations: vations,” Science, Vol.321, p. 1478, Doi:10.1126/science.1160943, • American Geophysical Union (AGU) 2008. • Seismological Society of Japan (SSJ) • [79] D. R. Shelly, W. L. Ellsworth, T. Ryberg, C. Haberland, G. C. Japan Geoscience Union (JpGU) Fuis, J. Murphy, R. M. Nadeau, and R. Burgmann, “Precise loca- tion of San Andreas Fault tremors near Cholame, California, us- ing seismometer clusters: Slip on the deep extension of the fault?,” Geophys. Res. Lett., Vol.36, L01303, doi:10.1029/2008GL036367, 2009.

Journal of Disaster Research Vol.9 No.3, 2014 329