Tectonophysics 642 (2015) 46–57

Contents lists available at ScienceDirect

Tectonophysics

j ourna l homepage: www.e lsev ie r .com/ locate / tecto

Stress fields in and around metropolitan Osaka, Japan, deduced frommicroearthquake focal mechanisms

Reiken Matsushita, Kazutoshi Imanishi ⁎Geological Survey of Japan, AIST, Tsukuba Central 7, 1-1-1 Higashi, Tsukuba, Ibaraki 305-8567, Japan

⁎ Corresponding author. Tel.: +81 29 861 3836.E-mail address: imani@ni.aist.go.jp (K. Imanishi).

http://dx.doi.org/10.1016/j.tecto.2014.12.0110040-1951/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 2 July 2014Received in revised form 14 November 2014Accepted 18 December 2014Available online 3 January 2015

Keywords:Stress fieldFocal mechanismSlip tendencyUemachi faultOsakaBorehole observation

Present-day stress fields in and around metropolitan Osaka, Japan, have been investigated based on 238 micro-earthquake focal mechanism solutions from approximately the past 10 years. The successful retrieval of thefocal mechanisms in the noisy urban area owed much to the clean seismic data recorded at four borehole obser-vatories deeper than 500 m depth in Osaka prefecture, in which P-wave onset polarity of the microearthquakecould be clearly identified. We found many microearthquakes with a pure reverse-faulting mechanism, whichis consistent with the faulting style of major active faults within the area (42-km long east-dipping Uemachiand 38-km long east-dipping Ikoma fault zones); however, a considerable number of microearthquakes withstrike-slip components were also found to occur. Most of the P-axes are oriented in the ESE–WNW to ENE–WSW directions sub-horizontally, conforming to the general tectonic trend of the area. The analysis of a stresstensor inversion delineated small but noticeable differences in stress regime, on a scale of at least 10 km,where strike-slip faulting prevails throughout the region but the middle and southern sections of the Uemachifault zone contain some reverse-faulting components. Based on the estimated stress fields, together with as-sumptions regarding the deep geometry of the Uemachi fault zone, we assessed the reactivation potential ofthe fault zone through slip-tendency analysis. The results indicate higher potential in the middle and southernsections of the fault zone, which is related to the locally increased reverse component detected there.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

Knowledge of the tectonic stress field is of great importance forvarious fields of geoscience including the modeling of geodynamicprocesses and the evaluation of seismic hazards (e.g., Zoback, 1992).As for the seismic hazard assessment, slip tendency analysis (Morriset al., 1996) is an effective tool and has been successfully used to evalu-ate the seismic slip potential along known or suspected faults in a givenstress field (e.g., Worum et al., 2004). However, the evaluation cruciallydepends on the adopted stress field, so we need to know a local-scalestress pattern near the faults that is as detailed as possible.

The Osaka urban region, which is the financial center of westernJapan, has a population exceeding eight million. A number of activefaults are concentrated within the area, including E–W- to NE–SW-trending strike-slip faults and N–S-trending reverse faults (e.g., Nakataand Imaizumi, 2002; Research Group for Active Faults of Japan, 1991).The Uemachi fault zone, an east-dipping blind reverse fault, traversesthe center of the city of Osaka (Fig. 1). According to the Headquartersfor Earthquake Research Promotion (2004), the Uemachi fault zonehas the potential for generating an earthquake with a magnitude(M) of about 7.5, which is a significant seismic source for the city and

the surrounding region. The Central Disaster Prevention Council of theCabinet Office of the Japanese Government (2007) estimated that inthe worst-case scenario, an earthquake in the Uemachi fault zonewould cause about 42,000 deaths, a million collapsed buildings, andeconomic damage of 74 trillion Yen. Various geological, geomorpho-logical, and geophysical surveys have been undertaken in the Osaka re-gion to reveal the paleoseismicity, fault slip rate, and deep geometry ofthe fault zone, together with the subsurface structure (e.g., Horikawaet al., 2003; Ministry of Education, Culture, Sports, Science andTechnology in Japan, 2013; Nakata et al., 1996; Okada and Chida,1996; Sugiyama et al., 2003). Information obtained from these surveysis of great importance for the long-term evaluation of seismic activity,evaluation of strong groundmotion, anddevelopment of possible earth-quake scenarios (e.g., Kase et al., 2003). Regarding the crustal stressfieldof the Japanese islands, there exist stress maps based on earthquakedata (e.g., Terakawa and Matsu'ura, 2010; Townend and Zoback, 2006;Tsukahara and Kobayashi, 1991), showing that the present-day south-west Japan is generally under E-W compression. To data, however, thestress field has not been described within the Osaka region (i.e. on alocal scale), which is information necessary for the evaluation of theslip potential along the Uemachi fault zone as well as for the improve-ment of the accuracy of the above studies.

Among the various stress indicators such as in-situ stress measure-ments and strain data, earthquake focal mechanisms are considered

130˚ 135˚ 140˚ 145˚

35˚

40˚

45˚

Eurasian or Amurian plate

Pacific plate

Philippine Sea plate

North American plate

135˚ 135.5˚ 136˚ 136.5˚34˚

34.5˚

35˚

0 10 20

km

UMFUMFUMFUMF

IKFIKFIKFIKF

OsakaOsakaOsakaOsaka

1995 M1995 Mw6.9 Kobe EQ6.9 Kobe EQ1995 Mw6.9 Kobe EQ

study areastudy areastudy areastudy area

Arima-Takatsuki Tectonic Line

Arima-Takatsuki Tectonic Line

Arima-Takatsuki Tectonic Line

Arima-Takatsuki Tectonic Line

Median Tectonic Line

Median Tectonic Line

Median Tectonic Line

Median Tectonic Line

Osaka BayOsaka BayOsaka BayOsaka BayKTZKTZKTZKTZ

Fig. 1. (a) Tectonic setting of Japanese islands. Kinki Triangle Zone (KTZ) (Huzita, 1962) is shownby a green triangle. (b) Enlargedmap of red rectangle shown in a depicting active faults andshallow background seismicity in and around the study area. Red lines show active faults after Nakata and Imaizumi (2002); Uemachi fault zone (UMF) and Ikoma fault zone (IKF) are in-dicatedby bold lines. Circles representM1+earthquake locations shallower than 20 kmdetermined by the JapanMeteorological Agency (JMA)during the period January2001 toDecember2011. Blue rectangle defines the present study area. The 1995 Mw6.9 Kobe earthquake is marked with a Centroid Moment Tensor solution by JMA (http://www.data.jma.go.jp/svd/eqev/data/mech/pdf/cmt1995.pdf (Accessed: June 30, 2014)).

47R. Matsushita, K. Imanishi / Tectonophysics 642 (2015) 46–57

the most effective means by which to constrain the stress field at thedepth of earthquake occurrence. The main obstacle to seismic observa-tion in urban areas is background noise produced by human activitiessuch as road traffic and industrial work. The National Research Institutefor Earth Science and Disaster Prevention (NIED) constructed four deepborehole observatories in the Kanto area including Tokyo, the capitalcity of Japan, showing a remarkable improvement of microearthquakedetection capability (Suzuki, 1996; Takahashi, 1982). These boreholeobservatories together with the surrounding high-sensitivity seismo-graph network contributed to the understanding of seismotectonicsand seismogenic processes beneath Tokyo (e.g., Ishida, 1992). Followingthe disastrous 1995 Mw 6.9 Hyogoken-nanbu (Kobe) earthquake, adensely distributed high-sensitivity borehole seismograph networkcovering all Japan (Hi-net) has been constructed, which is operatedby NIED (Okada et al., 2004). Most of the Hi-net seismometers wereinstalled at depths of a few hundred meters, but some specific siteshave boreholes deeper than 1000 m (three sites in Osaka prefecture).In addition, the Geological Survey of Japan, AIST (GSJ) has created anintegrated borehole network for observing subsurface water levels,water temperatures, crustal strain, tilt, and seismic waves. This networkis concentrated mainly in southwest Japan, where high-sensitivityseismometers have been installed at depths of about 30–800 m(Imanishi et al., 2011b; see also https://gbank.gsj.jp/wellweb/GSJ_E/tmp/gaiyoue.html#0 (Accessed: June 30, 2014)). In particular, there isan observatory at a depth of 543 m in the central part of the Uemachifault zone. These borehole seismic observations can significantly en-hance the capability for microearthquake detection, providing a uniqueopportunity for the retrieval of focal mechanisms and tectonic stressesin noisy urban Osaka area.

In this study, we infer the present-day stress fields in and aroundmetropolitan Osaka from the population of focal mechanism solutionsover approximately the past 10 years. Because most earthquakes thatoccur within the area are smaller than M2.0, we attempt to determinethe focal mechanisms using P-wave polarity data in conjunction withbody wave amplitudes, which permits us to obtain numerous well-determined solutions. Based on the estimated stress fields, we can forthe first time evaluate the slip potential of the Uemachi fault zone

using slip tendency analysis (Lisle and Srivastava, 2004; Morris et al.,1996).

2. Seismological and tectonic setting

In southwestern Japan, the Philippine Sea Plate is being subductednorthwestward beneath the Japanese Islands at a convergence rate of�40 mm/yr (e.g., Seno et al., 1993) (Fig. 1). M8-class megathrust earth-quakes have occurred repeatedly along the plate boundary with aninterval of 100–150 yrs (e.g., Ando, 1975). The latest events were the1944 M7.9 Tonankai and 1946 M8.0 Nankai earthquakes, which rup-tured different segments of the plate boundary (Baba and Cummins,2005). The crustal stress field in southwest Japan is basically charac-terized by a subhorizontal E-W compression as indicated by P-axis dis-tribution of earthquake focal mechanisms (Tsukahara and Kobayashi,1991), stress tensor inversion using earthquake focal mechanisms(Townend and Zoback, 2006) or centroid moment tensors (Terakawaand Matsu'ura, 2010), in-situ stress measurements (e.g., Tanaka,1985), and geological evidence (e.g., Sugiyama, 1994). The strike-slip faulting stress regime predominates the southwest Japan, butthe reverse-faulting one regionally prevails in the eastern part(e.g., Terakawa and Matsu'ura, 2010; Townend and Zoback, 2006;Tsukahara and Kobayashi, 1991). The region roughly correspondsto the Kinki Triangle Zone (Huzita, 1962), which is characterized bya dense distribution of primarily N–S-trending active reverse faults(Nakata and Imaizumi, 2002; Research Group for Active Faults ofJapan, 1991).

The present study area (a blue box in the right of Fig. 1) is situatedwithin the western part of the Kinki Triangle Zone, bounding to thenorth by Arima–Takatsuki Tectonic Line, which is a dextral strike-slipfault and to the south by the E–W-striking Median Tectonic Line. The1995 Mw6.9 Kobe earthquake occurred on NE-striking strike-slip faultsto the west of the region, causing over 6000 fatalities. The study areaincludes two major active faults: the Uemachi fault zone, traversingthe center of the city of Osaka, and the Ikoma fault zone, which runs al-most parallel to the Uemachi fault zone about 12 km further east. TheHeadquarters for Earthquake Research Promotion (2004) defined the

Mj3

21

GSJ NIED JMA DPRIERI

134˚ 135˚ 136˚ 137˚

34˚

35˚

0 50

km

Fig. 3.Distribution of seismic stations used for the hypocenter and focalmechanismdeter-mination:filled squares, Geological Survey of Japan, AIST; open squares, National ResearchInstitute for Earth Science and Disaster Prevention; diamonds, JapanMeteorological Agen-cy; inverted triangles, Earthquake Research Institute; triangles, Disaster Prevention Re-search Institute. The gray filled circles represent earthquake locations analyzed in thisstudy.

135˚ 135.5˚ 136˚ 136.5˚34˚

34.5˚

35˚

0 10 20

M 4 3 2

study area

km

Fig. 2. Spatial distribution of focal mechanism solutions shallower than 20 kmdeterminedby the Japan Meteorological Agency during the period October 1997 to December 2011(black) and by Nakamukae et al. (2003) (gray). All mechanisms are on equal-area projec-tions of the lower focal hemisphere.

48 R. Matsushita, K. Imanishi / Tectonophysics 642 (2015) 46–57

Uemachi fault zone as an approximately 42-km long east-dippingreverse fault with a strike ranging from N10°W to N20°E (UMF inFig. 1). The Ministry of Education, Culture, Sports, Science andTechnology in Japan (MEXT, 2013) conducted comprehensive researchon the Uemachi fault zone, which provided important informationregarding the development of plausible earthquake scenarios. That re-port suggested that the late Quaternary cumulative deformation couldnot be identified in the northern 5 km of the Uemachi fault zone. In-stead, the 26-km-long coastal fault from approximately 34.35°N135.23°E to 34.51°N 135.42°E, with a strike of about N45°E, was newlyrecognized as part of the Uemachi fault zone. Deformation recordedin growth strata indicates that the average vertical slip rate is at most0.6 mm/yr, the recurrence interval is longer than about 7000 yr,and that the most recent earthquake on the fault zone occurred afterapproximately 2700 yr BP. Regarding the Ikoma fault zone, whichis an approximately 38-km-long east-dipping reverse fault (IKF inFig. 1), the Headquarters for Earthquake Research Promotion (2001)showed that the average vertical slip rate is 0.5–1.0 mm/yr, the recur-rence interval is 3000–6000 yr, and that the most recent earthquakeoccurred approximately 400–1000 yr BP. Based on careful considerationof ancient writings, Hagiwara (1989) suggested that an earthquake in734 (M7.0) could be a candidate for the most recent destructive earth-quake on the fault zone.

Fig. 1 also illustrates the shallow seismicity in and around the studyarea during theperiod from January 2001 toDecember 2011, as routine-ly determined by the JapanMeteorological Agency (JMA). It can be seenthat there are some regions with high levels of seismicity: the Tanbadistrict lying to the north of the study area, swarm-like activity in theWakayama district lying to the southwest of the study area, and after-shocks of the 1995 Kobe earthquake. Background seismicity withinthe study area is quite low compared with these regions, but M4.0 orlarger earthquakes do occasionally occur.

The black “beach balls” in Fig. 2 show the focal mechanisms of earth-quakes that occurred during the period fromOctober 1997 to December2011 at depths shallower than 20 km, which have been determined bythe JMA based on observations of P-wave first motion polarities. TheJMA generally reports focal mechanisms for earthquakes greater thanM3.0, which is insufficient for discussions on the local tectonic stressprevailing within the study area. Using the newly deployed seismicobservatories of the Committee of Earthquake Observation and Re-search in the Kansai Area, which are distributed within the study area,Nakamukae et al. (2003) determined the focal mechanisms of earth-quakes down to M1.5 that occurred during the period from 1995 to1999 (gray beach balls in Fig. 2). The two data sets show that in additionto earthquakes with a pure reverse-fault mechanism, earthquakes withstrike-slip components do occur, which is inconsistent with the faultingstyle of the Uemachi and Ikoma fault zones. Although the P-axes are onaverage oriented to the general E–W trend of the region (e.g., Terakawaand Matsu'ura, 2010; Townend and Zoback, 2006; Tsukahara andKobayashi, 1991), they do appear to differ slightly depending on theirlocation. These observations imply spatial variation within the localstress field, but the small number of available focal mechanisms pre-cludes detailed investigation.

3. Data

We selected a rectangular area (blue box in Fig. 1 and dashed box inFig. 2) for the present study area, which includes both the Uemachi andIkoma fault zones, part of the Rokko fault zone, and theMedian TectonicLine active fault zone. Fig. 3 shows the distribution of the permanentstations used in this study, operated by the NIED, JMA, GSJ, EarthquakeResearch Institute (University of Tokyo), and Disaster Prevention Re-search Institute (Kyoto University). Each station is equipped with a setof three-component velocity transducers having a natural frequency of1 or 2 Hz. As already mentioned, seismometers operated by the NIEDand GSJ are installed at the bottom of boreholes.

We analyzed 261 earthquakes that occurred between June 2002and October 2011, which had JMA magnitude (Mj) N1 and focal depthb20 km. The locations of these earthquakes are shown by shaded circlesin Fig. 3. Of these earthquakes, 227 events (87%) were less than Mj2.0;the largest event was Mj4.1. Fig. 4 shows an example of the verticalcomponent seismograms of an Mj1.1 earthquake that occurred nearthe middle of the Uemachi fault zone. Four seismograms recorded inboreholes are enlarged, showing that the P-wave onset polarity of themicroearthquake can be clearly identified. These borehole data providea unique opportunity for retrieving the focal mechanisms and tectonicstress, even in noisy urban environments.

4. Hypocenter determination

We determined the hypocenters of the earthquakes by applying amaximum-likelihood estimation algorithm (Hirata and Matsu'ura,

0

5

10

15

20

25

30

35

40

45

50

Dep

th (

km)

3 4 5 6 7 8 9 10

Vp (km/s)

Fig. 5. P-wave velocity structure models used for hypocenter determination. The S-wave

models are assumed by scaling the P-wave velocities by a factor of 1=ffiffiffi

3p

.

0

10

20

30

40

50

Dis

tanc

e (k

m)

0 5 10 15 20

Time (s)

GS.TNN (543 m)

N.OSKH (1012 m)

N.KNHH (2032 m)

N.TJRH (1532 m)

max = 4.99 m/s

max = 1.03 m/s

max = 0.81 m/s

max = 0.55 m/s

0 10 20N.TJRH

km

N.KNHH N.KNHH N.KNHH N.KNHH

N.OSKH N.OSKH N.OSKH N.OSKH GS.TNN GS.TNN GS.TNN GS.TNN

21:03:46 (LT) on 9 October 2010

Fig. 4. Example of vertical component seismograms of a microearthquake (Mj1.1, 9.9 kmdepth) that occurred near the Uemachi fault zone at 2103 LT on 9 October 2010. Ampli-tude is normalized for each station to its maximum value. P-wave portions recorded atfour deep borehole stations are enlarged, in which the number next to the station codeshows the borehole depth. The inset map shows the location of the stations togetherwith the epicenter of the earthquake.

49R. Matsushita, K. Imanishi / Tectonophysics 642 (2015) 46–57

1987) to picked arrival times. The P- and S-wave arrival times wereidentified manually using the WIN system developed by Urabe andTsukada (1991). The P-wave velocity model used in our investigationis a one-dimensional velocity structure based on the tomography resultsof Matsubara et al. (2008), which consists of four layers (Fig. 5). The S-wave models are assumed by scaling the P-wave velocities by a factorof 1=

ffiffiffi

3p

. We first located the hypocenters of all the events withoutstation corrections. We then relocated them by introducing the stationcorrections, which were obtained using the average of the differencesbetween the observed and theoretical travel times at each station. Werepeated the above procedure until the reduction of the root meansquare of the arrival time residuals (RMS) became less than 0.01 s andobtained the final locations as well as station corrections for each sta-tion. After three iterations, the RMS decreased from 0.123 to 0.071 sfor the P-wave and from 0.380 to 0.143 s for the S-wave. As for the sta-tion corrections, theywere nearly the same after two or three iterations.The average spatial errors calculated using themaximum-likelihood esti-mation algorithmwere 125m in the horizontal direction and 420m ver-tically, which are sufficient for the purposes of the present study. Fig. 6shows the hypocenter distribution determined in this study. In compari-sonwith the distribution determined by JMA, our locations clustered andbecame shallower a few km on average. Earthquakes can be seen distrib-uted near mapped surface fault traces, except for the seismicity in theOsaka Bay area and the region between the Uemachi and Ikoma faultzones. Considering the earthquakes near the Uemachi fault zone, it canbe seen that the earthquakes are concentrated in an areawhere the strikeof the fault changes (34.6–34.7°N). The depth distribution reveals thatseismicity in the present study area is approximately confined at depthsof 3–15 km. The seismicity below the Uemachi fault zone is sparse in

the depth range of 0–9 km. The blue circles in Fig. 6 show earthquakesof M3.0 or greater, indicating that larger events tend to occur at thebase of the seismogenic zone. A similar pattern has also been reportedin other continental areas around the world (e.g., Boese et al., 2012;Spada et al., 2013; Yang and Hauksson, 2011). Fig. 7 presents the earth-quake hypocenters projected onto profiles almost perpendicular to theUemachi and Ikoma fault zones. We also show the deep fault geometriesof the Uemachi fault zone, which are inferred from a 3-dimensional bal-anced cross-sectional analysis assuming that the strata of the marineclay on the hanging wall side has been deformed as fault-related folds(MEXT, 2013). Considering the deep fault geometry of the Uemachifault zone, it can be seen that the seismicity distributed near the surfacetraces of the fault occurs primarily on the footwall side. The profile A–A′shows three alignments that appear related to fault structures. The align-ment a is located on the east side of the Uemachi fault zone and has aneastward-dipping plane with an angle of about 30°. However, the align-ment location does not always coincide with the inferred deep faultgeometry of the Uemachi fault zone, but appears located at the pointwhere the fault dip changes from steep to gentle. The alignments b andc are located on the east side of the Ikoma fault zone and have aneastward-dipping planes with angles of about 50° and 20°, respectively.Both alignments could be part of the deeper portion of the Ikoma faultzone. The seismicity in the profiles B–B′ and C–C′ shows some small-sized clusters, but it is difficult to identify clear alignments associatedwith the fault zones.

5. Stress field in and around the Uemachi Fault Zone

5.1. Focal mechanism solution of microearthquakes using body waveamplitudes

In this study, we determined focal mechanism solutions using P-wave polarity together with absolute P and SH amplitudes. The sameapproach has been used in previous studies (e.g., Imanishi et al.,2011a) and shown to be effective for microearthquakes. We analyzedearthquakes with at least 10 P-wave polarities. After applying correc-tions for instrument response, we determined the spectral levels andcorner frequencies of the spectra by fitting the ω2 model (Boatwright,1978) with an attenuation correction. The spectral levels were used asobserved amplitudes in the preceding analysis. Theoretical amplitudes

135.2˚ 135.4˚ 135.6˚ 135.8˚

34.4˚

34.6˚

34.8˚

M3

21

Fig. 6. Hypocenter distributions determined in the present study. The bottom six panels show earthquakes in each consecutive 3-km depth interval from 0 to 18 km. Circles in blue rep-resent earthquakes ofM3 or greater. Boxes in the top panel labeled A–A′ through C–C′ are used for plots in Figs. 7 and 9. Inverted triangles show reference locations of the Ikoma fault zonefor each box.

50 R. Matsushita, K. Imanishi / Tectonophysics 642 (2015) 46–57

were computed based on the far-field solutions of a shear point-sourcedislocation in a homogeneous infinite medium, with corrections for theincident angles at the surface andgeometrical spreading. The best-fit so-lution for each event was determined by minimizing the residual be-tween the observed and theoretical amplitudes, for which a gridsearch was performed using a 5° grid for strike, dip, and slip angles.

We first determined the focal mechanisms and seismicmoments forall events and then re-determined them by introducing the amplitudestation correction, obtained using the logarithmic average of the ratiosbetween the theoretical and observed amplitudes at each station. Thestability of the solution was verified by plotting all focal mechanismswhose residual was within 1.1 times that of the minimum value. Werejected ambiguous solutions where multiple solutions were possible.In total, we obtained 238 focalmechanismswhosemomentmagnitudes

ranged from 1.4 to 3.7. We defined focal mechanism uncertainties foreach event based on the average of the Kagan angles (Kagan, 1991) be-tween the best-fitting solution and all the solutions whose residual wasless than 1.1 times the minimum residual value. Here, the Kagan anglebecomes 0° when the two mechanisms are the same and 120° whenthey differ most. The average uncertainty of all the focal mechanismswas 20°. The focal mechanisms and their faulting type are plotted inFig. 8a. A triangle diagram (Flohlich, 1992) with a color scale is present-ed to the right of Fig. 8a, in which the different colors illustrate thefaulting type: reverse (green), strike-slip (red), and normal (blue).Following the definition of Frohlich (1992), we categorized reverseevents as those having T-axis plunges of less than 40°, and strike-slipand normal fault earthquakes as those having B- and P-axis plunges ofless than 30°, respectively. All remaining events were defined as

0

10

20

Dep

th (

km)

0

10

20

Dep

th (

km)

0

10

20

Dep

th (

km)

M3

21

A A'

C C'

B B'

a b c

Fig. 7. Earthquake locations projected onto the corresponding profiles shown in Fig. 6.Circles in blue represent earthquakes of M3 or greater. A, B, and C in profile A–A′ representearthquake alignments. Gray curves show the deep geometry of the Uemachi fault zoneinferred from a balanced cross-section analysis (Ministry of Education, Culture, Sports,Science and Technology in Japan, 2013), where three representative curves (one half,one quarter and three quarters of each box) are shown in each profile. Inverted trianglescorrespond to the surface trace of the Ikoma fault zone.

51R. Matsushita, K. Imanishi / Tectonophysics 642 (2015) 46–57

“other.” The numbers of categorized reverse, strike-slip, normal, andother eventswere 81, 33, 3, and 121, respectively. Fig. 8b shows the spa-tial distribution of focal mechanisms for each faulting type. It can beseen that the reverse faulting type earthquakes occur throughout thestudied area, which is consistent with the fact that the active faultswithin the area represent reverse faulting (Fig. 1). It can also be notedthat a considerable number of microearthquakes with strike-slip com-ponents occur, which suggests that the area is characterized by a mix-ture of reverse- and strike-slip-faulting stress regimes. Fig. 9represents the focal mechanisms and their faulting type projectedonto three profiles, whose sections are the same as those in Fig. 7. Asshown in the previous section, the relocated hypocenter reveals threeearthquake alignments in the profile A–A′. It is noted that most of thefocal mechanisms within these alignments show reverse faulting andthat their dip angles are consistent with those of each alignment. Thisobservation further supports that these alignments represent fault-plane structures associated with the deeper portion of the Uemachiand Ikoma fault zones. The earthquake depth profiles also suggest thatreverse-faulting earthquakes occur throughout the seismogenic zone,whereas a relatively larger number of strike-slip earthquakes occur inthe deeper part of the seismogenic zone, which is specifically identifiedin profile C–C′ and below the Uemachi fault in profile B–B′. This depthdependency in the stress regime will be examined further later in thediscussion.

The directions of the P- and T-axes are shown in Fig. 10, wheredifferent colors are used to represent different plunge angles. Most ofthe P-axes are sub-horizontal and oriented in ESE–WNW to ENE–WSW directions, which conforms to the general tectonic trend withinthis area (e.g., Terakawa and Matsu'ura, 2010; Townend and Zoback,2006; Tsukahara and Kobayashi, 1991). The T-axes have a wide rangeof plunge, indicating the coexistence of reverse and strike-slip faultingearthquakes.

The focal mechanisms alone do not adequately represent the stressfield, because the direction of the P- and T-axes can differ from themax-imum and minimum principal stresses by 45°, respectively (McKenzie,1969). In the following, we invert for the orientations of the principalstresses and their relative magnitude from the population of focalmechanisms.

5.2. Stress tensor inversion

Using the focal mechanism solutions determined in the previoussection, we estimated the stress field within the study area by applyingthe inversionmethod ofMichael (1984). The inversion solves the orien-tation of the three principal stress axes together with the relative mag-nitude of the principal stresses defined by ϕ = (S2 − S3) / (S1 − S3),where S1, S2, and S3 are the maximum, intermediate, and minimumcompressive principal stresses, respectively. One plane must be chosenfrom each focal mechanism as the actual fault plane, because the inver-sion uses the direction of the tangential traction on the fault plane. Wechose the preferred planes while inverting for the stress tensor basedon Michael's (1987) grid search algorithm, which is considered to be arational approach in situations where the choice of fault plane cannotbe made based on relative hypocentral locations or geological informa-tion (e.g., Kastrup et al., 2004). There are four parameters in the gridsearch algorithm: the trend and plunge of the S3 axis, a rotation angleof the S2 axis about the S3 axis, and ϕ. The grid search was performedusing a 5° grid for all of the angular variables and a step of 0.1 for ϕ.To calculate confidence regions for the stress tensor, we applied thebootstrap resampling technique by assuming that a certain percentageof the planes would be picked incorrectly (Michael, 1987). In thepresent study, for the bootstrap, we assumed that each nodal planehad the same probability of being selected during the resampling. Fol-lowing Michael (1987), we used 2000 bootstrap samples to obtain the95% confidence region.

We divided the focal mechanism data set into nine areas (A1 to A9)based on earthquake locations (Fig. 11). The number within each set ofparentheses indicates the number of events used for the inversion. Weexcluded 3 normal fault events together with 11 events whose P-axisdirections obviously deviated from the general tectonic trend, becausetheymade the inversion unstable.Webelieve that these eventswere as-sociated with a local heterogeneous stress field. The results of the stresstensor inversion are shown in Fig. 12. The average misfit angles (β)between the predicted tangential traction on the fault planes and theobserved slip direction on each plane range from 13° (area A2) to 34°(area A8), suggesting a valid fit to the single stress tensor (Michael,1991). For each area, themaximumprincipal stress S1 is clearly differen-tiated from S2 and S3, trending sub-horizontally from ESE–WNW toENE–WSW. The S2–S3 plane rotates around the S1 direction, whichgives rise to a spatial dependence in the stress field. Based on the defini-tion of Zoback (1992), and using the best stress axes for each area, thestress regime is categorized as reverse faulting (RF) for area A6, pre-dominantly reverse faulting with a strike-slip component (RS) forareas A5 and A9, and strike-slip faulting (SS) for the remaining areas.The stress ratio ϕ is basically less than 0.5 and therefore, the magnitudeof S2 is closer to S3 than S1. In other words, themagnitude of S1 is prom-inent throughout the studied area, showing smaller variation in the S1orientation compared with the other two principal stresses.

Following the method of Lund and Townend (2007), we computedthe true axis of maximum horizontal compressive stress (SHmax) from

135.2˚ 135.4˚ 135.6˚ 135.8˚

34.4˚

34.6˚

34.8˚

0 10

Mw

3

21

strike-slip

reversenormal

other

Triangle diagram

135.2˚ 135.4˚ 135.6˚ 135.8˚

34.4˚

34.6˚

34.8˚

0 10

135.2˚ 135.4˚ 135.6˚ 135.8˚

34.4˚

34.6˚

34.8˚

0 10

135.2˚ 135.4˚ 135.6˚ 135.8˚

34.4˚

34.6˚

34.8˚

0 10

135.2˚ 135.4˚ 135.6˚ 135.8˚

34.4˚

34.6˚

34.8˚

0 10

Reverse (81) Strike slip (33)

Normal (3) Other (121)

(a)

(b)

km

km km

km km

Fig. 8. (a) Focal mechanism solutions of 238 events determined in the present study (lower hemisphere of equal-area projection), where different colors are used to differentiate reverse-(green), strike-slip- (red), and normal- (blue) faulting mechanisms. A triangle diagram (Flohlich, 1992) with color scale is presented in the upper right, where each focal mechanism isplotted by open circles. (b) Distributions of focal mechanisms for each faulting mechanism. The number within each set of parentheses indicates the number of events for that faultingmechanism. All mechanisms are on equal-area projections of the lower focal hemisphere.

52 R. Matsushita, K. Imanishi / Tectonophysics 642 (2015) 46–57

the four stress parameters determined by the stress tensor inversion(the directions of the three principal stresses and the stress ratio ϕ).Fig. 13 displays rose diagrams of SHmax orientations computed fromthe stress parameters that are within the 95% confidence regions.Here, the bars are colored according to the stress regime, which we de-scribed in the above paragraph. Also shown are the SHmax orientationsand faulting regimes reported by Townend and Zoback (2006), whichwere extracted from the World Stress Map database (Heidbach et al.,2008). The SHmax orientation observed throughout the region is almostuniform from E–W to ESE–WNW, although small local fluctuationsin the counterclockwise direction can be identified (e.g., A6 and A9).

With regard to the stress regime, strike-slip faultingprevails throughoutthe region; however, we also found a reverse-faulting regime, especiallyin the southern part of the region that includes area A6. In other words,it is indicated that there is a locally increased reverse component in themiddle and southern parts of the Uemachi fault zone.

6. Slip tendency analysis

We assessed slip potential of the Uemachi fault zone under the esti-mated stress fields using a slip tendency analysis (Morris et al., 1996). Inthe present study, we do not examine the slip tendency of the Ikoma

135.2˚ 135.4˚ 135.6˚ 135.8˚

34.4˚

34.6˚

34.8˚

0 10

A1 (48)A1 (48)A1 (48)A1 (48)

A2 (16)A2 (16)A2 (16)A2 (16)

A3 (13)A3 (13)A3 (13)A3 (13)

A7 (23)A7 (23)A7 (23)

A8 (18)A8 (18)A8 (18)A8 (18)

A6 (17)A6 (17)A6 (17)A6 (17) km

A4 (22)A4 (22)A4 (22)A4 (22)

A5 (30)A5 (30)A5 (30)A5 (30)

A9 (37)A9 (37)A9 (37)A9 (37)

Fig. 11. Definition of sub-areas (A1 to A9) used in the stress tensor inversion. White filledcircles represent focal mechanisms used in the inversion. The number within each set ofparentheses indicates the number of events used for the inversion.

0

10

20

Dep

th (

km)

0

10

20

Dep

th (

km)

0

10

20

Dep

th (

km)

Mw

3

21

A A'

C C'

B B'

Fig. 9. Focal mechanisms and their faulting type projected onto three profiles, whose sec-tions are the same as those in Fig. 7. The color is based on the triangle diagram in Fig. 8. Seecaption of Fig. 7 for additional information.

53R. Matsushita, K. Imanishi / Tectonophysics 642 (2015) 46–57

fault zone and a newly recognized part of the Uemachi fault zone asmentioned in the Section 2, because the deep geometry of these faultsis not yet clearly understood. Slip tendency (Ts) is based on the notionthat slip on a fault is controlled by the ratio of shear stress (τ) to effec-tive normal stress (σn) acting on the plane of weakness; hence, it is de-fined as Ts = τ / σn. Lisle and Srivastava (2004) introduced normalized

135.2˚ 135.4˚ 135.6˚ 135.8˚

34.4˚

34.6˚

34.8˚

0 10

P−axis

3

3

3

0 30Plung

km

Fig. 10. P- and T-axis distributions of focal mechanism solutions.

slip tendency as Ts′ = Ts/max(Ts) = Ts / μ, where μ is the coefficient ofstatic friction. Ts′ ranges from 0 to 1, where the larger value indicatesgreater slip probability.We supposed that the frictional sliding envelopeis tangential to the S1–S3 Mohr circle, which makes it possible to com-pute Ts′ without knowledge of the absolute stress value (Lisle andSrivastava, 2004). Following Collettini and Trippetta (2007), we definedfavorably oriented planes for 0.5 b Ts′ ≤ 1.0 and poorly oriented planesfor 0 ≤ Ts′ ≤ 0.5.

Based on mapped fault traces, we considered three different faultorientations: 350°, 5°, and 20° for the northern, middle, and southernsections of the fault zone, respectively (Fig. 14a). We assumed that thefault plane dips at 60° in the shallower part and at 30° in the deeperpart for the northern and middle sections, and at 45° for the southernsection, as suggested by a balanced cross-sectional analysis (MEXT,2013) (Fig. 7). We applied the stress fields estimated in areas A4, A5,and A6 (Fig. 12) to the northern, middle, and southern sections of thefault zone, respectively. The coefficient of static friction μ was set to0.6, which in general is considered suitable for seismogenic faults

135.2˚ 135.4˚ 135.6˚ 135.8˚

4.4˚

4.6˚

4.8˚

0 10

T−axis

60 90e (°)

km

Different colors are used to indicate different plunge angles.

0

10

20

Num

ber

0 60 120 180Misfit angle (°)

0

500

1000

Num

ber

0.0 0 .5 1.0Stress ratio,

0

10

20

Num

ber

0 60 120 180Misfit angle (°)

0

500

1000

Num

ber

0.0 0 .5 1.0Stress ratio,

0

10

20

Num

ber

0 60 120 180Misfit angle (°)

0

500

1000

Num

ber

0.0 0 .5 1.0Stress ratio,

Stress ratio, Stress ratio, Stress ratio,

0

10

20

Num

ber

0 60 120 180Misfit angle (°)

0

500

1000

Num

ber

0.0 0 .5 1.0

0

10

20

Num

ber

0 60 120 180Misfit angle (°)

0

500

1000

Num

ber

0.0 0 .5 1.0

0

10

20

Num

ber

0 60 120 180Misfit angle (°)

Stress ratio, Stress ratio, Stress ratio,

Misfit angle (°) Misfit angle (°) Misfit angle (°)

0

500

1000

Num

ber

0.0 0 .5 1.0

A1

A2

A3

0

10

20

Num

ber

0 60 120 180

0

500

1000

Num

ber

0.0 0 .5 1.0

A4

A5

A6

0

10

20

Num

ber

0 60 120 180

0

500

1000

Num

ber

0.0 0 .5 1.0

A7

A8

A9

0

10

20

Num

ber

0 60 120 180

0

500

1000

Num

ber

0.0 0 .5 1.0

S1S2S3

95% confidence regions of S195% confidence regions of S295% confidence regions of S3

×

×

×

Fig. 12. Stress tensor inversion results for each area. Left-hand panels show principal stress axes with their 95% confidence regions plotted on lower hemisphere stereonets. Upper-rightpanels show the misfit angle for the data with respect to the best stress tensor determined by the stress tensor inversion, where the misfit angle represents the angle between the tan-gential traction predicted by the best solution and the observed slip direction on each plane determined from the focal mechanism. Lower-right panels show the frequency of the stressratio ϕ, which belongs to the 95% confidence region.

54 R. Matsushita, K. Imanishi / Tectonophysics 642 (2015) 46–57

(e.g., Sibson and Xie, 1998; Townend, 2006). The assumption of thevalue of μ does not change the main conclusion. Fig. 14c shows the fre-quency of Ts′ computed from the 95% confidence regions of the assumedstress fields (solid bars), together with the Ts′ of the best stress solution(solid inverted triangles). The results suggest that the present-daystress fields are favorably oriented in the Uemachi fault zone exceptfor the shallower part (dip 60°) of the middle section.

TheWallace–Bott hypothesis (Bott, 1959;Wallace, 1951) states thatslip on faults occurs in the direction of maximum resolved shear stress.Based on their hypothesis, the currently assumed stress fields favorreverse faulting in the middle section, but oblique slip faulting in thenorthern and southern sections. High-resolution seismic reflectionprofiles and boring core data indicate that the vertical cumulative dis-placements by the Uemachi fault reach a few hundred meters, whilethe lateral displacements are not remarkable (e.g., Headquarters forEarthquake Research Promotion, 2004; Sugiyama et al., 2003). Follow-ing this observation, we computed the normalized slip tendency usingthe component of shear stress acting in the slip direction of 90° on afault (a pure dip-slip faulting) instead of the shear stress itself. Wetermed this the normalized dip-slip tendency (Tds′) to avoid confusion

with the original definition of Ts′ (Fig. 14b). Open bars and openinverted triangles in Fig. 14d show the frequency of Tds′ computedfrom the 95% confidence regions of the assumed stress fields and theTds′ of the best stress solution, respectively. Because the directions ofmaximumshear stress resolved on the fault planes of themiddle sectionare approximately 90°, the computed values of Tds′ are almost consis-tentwith those of Ts′. Therefore, the shallower part of themiddle section(dip 60°) is considered unfavorably oriented for failure in the present-day stress fields. This is a natural consequence because it is difficult forhigh-angle reverse faults to reactivate under horizontal compression.However, for the northern and southern sections, where obliquefaulting is expected, the values of Tds′ generally become smaller com-paredwith those of Ts′. The computed values of Tds′ of the best stress so-lution, together with the majority of the 95% confidence regions, fallbelow 0.5 for the northern section (both the shallower and the deeperparts), but still exceed 0.5 for the southern section. The reason whythe northern section results in an unfavorable orientation in terms ofpure reverse faulting is largely associated with the lack of a reverse-faulting component in its applied stress field. It is noted that this resultsupports the geological observation that the late Quaternary cumulative

135˚ 135.5˚ 136˚

34.5˚

35˚

0 10 20

RFSS

RS

Stress regime

34˚

A1A1A1

A2A2A2

A4A4A4

A5A5A5

A6A6

A7A7A7

A8A8A8

A9A9A9

A3A3A3

A1

A2

A4

A5

A6

A7

A8

A9

A3

km

Fig. 13.Map showing the axes of maximum horizontal compressive stress (SHmax) deter-mined in the present study (A1 to A9) together with that reported by Townend andZoback (2006). The present result displays rose diagrams of SHmax computed from thestress parameters that belong to the 95% confidence region. The color denotes the stressregime (green: reverse faulting (RF), red: strike-slip faulting (SS), orange: predominantlyreverse faulting with strike-slip component (RS)) based on the definition of Zoback(1992).

55R. Matsushita, K. Imanishi / Tectonophysics 642 (2015) 46–57

deformation cannot be identified in the northern 5 km of the Uemachifault zone (MEXT, 2013). Thus, we conclude that the middle (only thedeeper part) and southern sections have greater potential for reactiva-tion of reverse faulting.

7. Discussion

We determined earthquake distribution and local stress fields inand around metropolitan Osaka using microearthquake data downto M1.0, which has filled the gap in the stress fields left by previousstudies (e.g., Townend and Zoback, 2006) (see Figs. 2 and 13). The in-ferred stress fields are generally coincident with those observed insurrounding areas, especially with regard to the orientation ofSHmax. However, with a scale of at least 10 km, there are small but no-ticeable differences in both SHmax and the stress regime. These localfluctuations actually affect the evaluation of slip potential on thefault, emphasizing the importance of creating local-scale stressmaps that are as detailed as possible.

The depth distribution of focal mechanisms implies that reverse-faulting earthquakes occur throughout the seismogenic zone, while arelatively larger number of strike-slip earthquakes occur at the base ofthe seismogenic zone (Fig. 9). Iio (1996) found a similar depth variationin focal mechanisms of earthquakes in the Tanba district, which lies tothe north of the study area, where both strike-slip and reverse-faultevents are found throughout the seismogenic zone, and strike-slipearthquakes becomepredominant in the deeper part of the seismogeniczone. Iio (1996) inferred that the minimum horizontal compressionalstress (the N–S orientation) becomes lower than the vertical stressbelow the brittle layer due to brittle–semi-brittle transitional be-havior, which could possibly explain the observational evidence thatearthquakes occurring at deeper parts of the seismogenic zone aredominated by a strike-slip component. Here, based on the results of

deep crustal seismic surveys (Ito et al., 2006), we will present anothermechanism that could account for a change in stress field with depth.Ito et al. (2006) conducted seismic refraction and wide-angle-reflection surveys of the crust and upper mantle structure, the surveyline of which is located along the eastern part of the present studyarea (approximately from 33.72°N 135.95°E to 35.42°N 135.42°E), cut-ting almost perpendicular to the Median Tectonic Line and the Arima–Takatsuki Tectonic Line. They found clear reflectors in the lower crustat depths from about 15 and 35 km, which gently decline towards thenorth. If these reflectors represent fault planes and aseismic slip occursalong them, then the N–S tensional stress above the reflector planeswould be caused by the hanging wall side being pulled southward inthe case of reverse-faulting slip and northward in the case ofnormal-faulting slip. We infer that the induced tensional stress re-duces the N–S compressional stress and strike-slip-faulting earth-quakes become dominant at the base of the seismogenic zone.However, the present study alone cannot constrain the type offaulting that is favored along the reflector planes, because the stressfield there is unknown.

We have discussed the deviatoric part of the stress tensor, becausestress fields estimated from focal mechanisms do not yield absolutevalues of the principal stresses. However, information regarding abso-lute stress over a seismogenic zone is essential for understanding theprocess of earthquake generation. The b-value of theGutenberg–Richterlaw (Gutenberg and Richter, 1944) has been shown to be inversely pro-portional to differential stresses (e.g., Amitrano, 2003; Schorlemmeret al., 2005; Spada et al., 2013), suggesting that the b-value can beused as a proxy for a stress meter of the Earth's crust. In the presentstudy, we observed that larger events tend to occur at the base of theseismogenic zone (Figs. 6 and 7). Unfortunately, the small number ofearthquakes precludes the computation of a reliable b-value, but it islikely that the b-value for the deeper part of the seismogenic zonewould be lower than for the shallower part, implying that the deeperpart is highly stressed. We believe that further longer-term seismic ob-servation will reveal the spatial and depth distributions of b-valueswithin the study area, which together with the estimated stress fieldswould contribute to the development of more accurate seismic-hazardassessment for metropolitan Osaka.

8. Conclusion

We have investigated stress fields in and around metropolitanOsaka, Japan, based on numerous focal mechanism solutions of mi-croearthquakes over approximately the past 10 years. Borehole seis-mic observations allowed us to obtain 238 well-constrained focalmechanisms, even in a noisy urban environment, which has filled agap in the stress fields left by previous studies. We found manyearthquakes with a pure reverse-faulting mechanism throughoutthe studied area, which is consistent with the faulting style of activefaults within the area (Uemachi and Ikoma fault zones). It was alsofound that a considerable number of microearthquakes occur withstrike-slip components. This observation indicates that the area ischaracterized by a mixture of reverse- and strike-slip-faulting stressregimes. Most of the P-axes are sub-horizontal and oriented in theESE–WNW to ENE–WSW directions, which conform to the generaltectonic trend within this area. A stress tensor inversion revealedsmall but noticeable differences in stress regime, on a scale of atleast 10 km, in which strike-slip faulting prevails over the region,but where the middle and southern sections of the Uemachi faultzone contain some reverse-faulting components. We assessed thereactivation potential of the Uemachi fault zone using slip tendencyanalysis under the estimated stress fields and assumptions regardingthe deep fault geometry. The results indicate greater potential for re-activation in the middle and southern sections of the fault zone,which is associated with the locally increased reverse-faultingcomponent.

0

500

1000

Num

ber

0.0 0.5 1.0Slip tendency

0

500

1000

Num

ber

0.0 0.5 1.0Slip tendency

0

500

1000

Num

ber

0.0 0.5 1.0Dip−slip tendency

0

500

1000

Num

ber

0.0 0.5 1.0Dip−slip tendency

Dip: 60° Dip: 30°

Northern section (Strike: 350°) Northern section (Strike: 350°)

Dip: 60° Dip: 30°

0

500

1000

Num

ber

0.0 0.5 1.0Slip tendency

0

500

1000

Num

ber

0.0 0.5 1.0Slip tendency

0

500

1000

Num

ber

0.0 0.5 1.0Dip−slip tendency

0

500

1000

Num

ber

0.0 0.5 1.0Dip−slip tendency

Middle section (Strike: 5°) Middle section (Strike: 5°)

0

500

1000

Num

ber

0.0 0.5 1.0Slip tendency

0

500

1000

Num

ber

0.0 0.5 1.0Dip−slip tendency

Dip: 60° Dip: 30° Dip: 60° Dip: 30°

Dip: 45°

Ts’

95% confidence region

best stress solution

Southern section (Strike: 20°) Southern section (Strike: 20°)

Dip: 45°

Tds’

95% confidence region

best stress solution

(c) (d)

135.2˚ 135.4˚ 135.6˚ 135.8˚

34.4˚

34.6˚

34.8˚

0 10

km

Northern sectionNorthern sectionNorthern sectionNorthern section

Southern sectionSouthern sectionSouthern sectionSouthern section

Middle sectionMiddle sectionMiddle sectionMiddle section

(a) (b)

Fig. 14. Slip tendency analysis of the Uemachi fault zone. (a) Location of northern, middle, and southern sections of the fault zone. (b) Definition of normalized slip tendency (Ts′) andnormalized dip-slip tendency (Tds′). (c) The frequency of the normalized slip tendency (Ts′) computed from the 95% confidence region of the assumed stress fields. The inverted trianglesrepresent the normalized slip tendency of the best stress solution (see the detail in the text). (d) The same as in c but for the normalized dip-slip tendency (Tds′).

56 R. Matsushita, K. Imanishi / Tectonophysics 642 (2015) 46–57

Acknowledgments

Seismograph stations used in this study include permanent stationsoperated by the National Research Institute for Earth Science and Disas-ter Prevention (Hi-net), Japan Meteorological Agency, EarthquakeResearch Institute (University of Tokyo), and Disaster Prevention Re-search Institute (Kyoto University), as well as the Geological Survey ofJapan, AIST. Comments by two anonymous reviewers were helpful inimproving the manuscript. Haruo Kimura kindly provided us with digi-tal data of the fault geometry of the Uemachi fault zone. We used “SlickPackage” (http://earthquake.usgs.gov/research/software/) softwarecoded by Andrew Michael for the stress tensor inversion. We modifieda program coded by Satoshi Ide for estimating focal mechanism

solutions. Figures were generated using Generic Mapping Tools(Wessel and Smith, 1998).

References

Amitrano, D., 2003. Brittle–ductile transition and associated seismicity: experimentaland numerical studies and relationship with the b value. J.geophys. Res. 108 (B1).http://dx.doi.org/10.1029/2001JB000680.

Ando, M., 1975. Source mechanisms and tectonic significance of historical earthquakesalong the Nankai Trough, Japan. Tectonophysics 27, 119–140.

Baba, T., Cummins, P.R., 2005. Contiguous rupture areas of two Nankai Trough earth-quakes revealed by high-resolution tsunami waveform inversion. Geophys. Res.Lett. 32, L08305. http://dx.doi.org/10.1029/2004GL022320.

Boatwright, J., 1978. Detailed spectral analysis of two small New York State earthquakes.Bull. Seismol. Soc. Am. 68, 1117–1131.

57R. Matsushita, K. Imanishi / Tectonophysics 642 (2015) 46–57

Boese, C.M., Townend, J., Smith, E., Stern, T., 2012. Microseismicity and stress in the vicin-ity of the Alpine Fault, central Southern Alps, New Zealand. J. Geophys. Res. 117,B02302. http://dx.doi.org/10.1029/2011JB008460.

Bott, M.H.P., 1959. The mechanics of oblique slip faulting. Geol. Mag. 96, 109–117.Central Disaster Prevention Council, Cabinet Office of the Japanese government, 2007.

Countermeasures against Chubu and Kinki regions island earthquakes. http://www.bousai.go.jp/jishin/index.html (in Japanese, Accessed: June 30, 2014).

Collettini, C., Trippetta, F., 2007. A slip tendency analysis to test mechanical and structuralcontrol on aftershock rupture planes. Earth Planet. Sci Lett. 255, 402–413.

Flohlich, C., 1992. Triangle diagrams: ternary graphs to display similarity and diversity ofearthquake focal mechanism. Phys. Earth Planet. Interiors. 75, 193–198.

Gutenberg, B., Richter, C.F., 1944. Frequency of earthquakes in California. Bull. seism. Soc.Am. 34, 185–188.

Hagiwara, T. (Ed.), 1989. Paleo-Earthquakes, Second Volume: Real and False Images.Tokyo Univ. Press, Tokyo (434 pp., in Japanese).

Headquarters for Earthquake Research Promotion, 2001. Long-term evaluation of theIkoma fault zone. http://jishin.go.jp/main/chousa/01may2/index.htm (in Japanese,Accessed: June 30, 2014).

Headquarters for Earthquake Research Promotion, 2004. Long-term evaluation of theUemachi fault zone. http://www.jishin.go.jp/main/chousa/04mar_uemachi/index.htm (in Japanese, Accessed: June 30, 2014).

Heidbach, O., Tingay, M., Barth, A., Reinecker, J., Kurfeß, D., Müller, B., 2008. The WorldStress Map Database Release 2008. http://dx.doi.org/10.1594/GFZ.WSM.Rel2008.

Hirata, N., Matsu'ura, M., 1987. Maximum likelihood estimation of hypocenter with origintime eliminated using nonlinear inversion technique. Phys. Earth Planet. Interiors 47,50–61.

Horikawa, H., Mizuno, K., Ishiyama, T., Satake, K., Sekiguchi, H., Kase, Y., Sugiyama, Y.,Yokota, H., Suehiro, M., Yokokura, T., Iwabuchi, Y., Kitada, N., Pitarka, A., 2003. Athree-dimensional subsurface structure model beneath the Osaka sedimentarybasin, southwest Japan, with fault-related structural discontinuities. Annual Reporton Active Fault and Paleoearthquake Researches No.3, 225–259 (in Japanese withEnglish abstract).

Huzita, K., 1962. Tectonic development of the Median zone (Setouchi) of southwestJapan, since the Miocene. J. Geosci. 6, 103–114 (Osaka City Univ.).

Iio, Y., 1996. Depth-dependent change in the focal mechanism of shallow earthquakes:implications for brittle–plastic transition in a seismogenic region. J. Geophys. Res.101 (B5), 11,209–11,216. http://dx.doi.org/10.1029/96JB00409.

Imanishi, K., Kuwahara, Y., Takeda, T., Mizuno, T., Ito, H., Ito, K., Wada, H., Haryu, Y., 2011a.Depth-dependent stress field in and around the Atotsugawa fault, central Japan,deduced from microearthquake focal mechanisms: Evidence for localized aseismicdeformation in the downward extension of the fault. J. Geophys. Res. 116, B01305.http://dx.doi.org/10.1029/2010JB007900.

Imanishi, K., Takeda, N., Kuwahara, Y., Koizumi, N., 2011b. Enhanceddetection capability ofnon-volcanic tremor using a 3-level vertical seismic array network, VA-net, in south-west Japan. Geophys. Res. Lett. 38, L20305. http://dx.doi.org/10.1029/2011GL049071.

Ishida, M., 1992. Geometry and relative motion of the Philippine Sea Plate and PacificPlate beneath the Kanto–Tokai District, Japan. J. Geophys. Res. 97 (B1), 489–513.http://dx.doi.org/10.1029/91JB02567.

Ito, K., Umeda, Y., Sato, H., Hirose, I., Hirata, N., Kawanaka, T., Ikawa, T., 2006. Deep seismicsurveys in the Kinki District: Shingu–Maizuru line. Bull. Earthq. Res. Inst. 81, 239–245.

Kagan, Y.Y., 1991. 3-D rotation of double-couple earthquake sources. Geophys. J. Int. 106,709–716.

Kase, Y., Sekiguchi, H., Ishiyama, T., Horikawa, H., Satake, K., Sugiyama, Y., 2003. Simula-tion of rupture process in heterogeneous stress field estimated from active fault infor-mation: application to the Uemachi and Ikoma fault systems. Annual Report onActive Fault and Paleoearthquake Researches No.3, 261–272 (in Japanese withEnglish abstract).

Kastrup, U., Zoback, M.L., Deichmann, N., Evans, K.F., Giardini, D., Michael, A.J., 2004. Stressfield variations in the Swiss Alps and the northern Alpine foreland derived from in-version of fault plane solutions. J. Geophys. Res. 109. http://dx.doi.org/10.1029/2003JB002550.

Lisle, J.R., Srivastava, D.C., 2004. Test of the frictional reactivation theory for faults andvalid-ity of fault-slip analysis. Geology32 (7), 569–572. http://dx.doi.org/10.1130/G20408.1.

Lund, B., Townend, J., 2007. Calculating horizontal stress orientations with full or partialknowledge of the tectonic stress tensor. Geophys. J. Int. http://dx.doi.org/10.1111/j.1365-246X.2007.03468.x.

Matsubara, M., Obara, K., Kasahara, K., 2008. Three-dimensional P- and S-wave velocitystructures beneath the Japan Islands obtained by high-density seismic stations byseismic tomography. Tectonophysics 454, 86–103. http://dx.doi.org/10.1016/j.tecto.2008.04.016.

McKenzie, D.P., 1969. The relation between fault plane solutions for earthquakes and di-rections of the principal stress. Bull. Seis. Soc. Am. 59, 591–601.

Michael, A.J., 1984. Determination of stress from slip data: faults and folds. J. Geophys. Res.89, 11,517–11,526.

Michael, A.J., 1987. Use of focal mechanisms to determine stress: a control study.J. Geophys. Res. 92, 357–368.

Michael, A.J., 1991. Spatial variations in stress within the 1987 Whittier Narrows,California, aftershock sequence: new techniques and results. J. Geophys. Res. 9696.

Ministry of Education, Culture, Sports, Science and Technology in Japan, 2013. The com-prehensive research on the Uemachi fault zone. ttp://www.jishin.go.jp/main/chousakenkyuu/uemachi_juten/h24/index.htm (in Japanese, Accessed: June 30,2014).

Morris, A., David, A.F., Henderson, B., 1996. Slip-tendency analysis and fault reactivation.Geology 24, 275–278.

Nakamukae, M., Katao, H., Nemoto, H., Nakagawa, Y., 2003. Focal mechanisms of micro-earthquakes inOsaka basin, Zisin. J. Seismol. Soc. Jpn. 56 (Ser. 2), 303–309 (in Japanese).

Nakata, T., Imaizumi, T. (Eds.), 2002. Digital Active Fault Map of Japan. Univ. of TokyoPress, Tokyo (in Japanese, 60 pp.).

Nakata, T., Okada, A., Suzuki, Y., Watanabe, M., Ikeda, Y., 1996. Osaka Seinanbu. ActiveFault Map in Urban Area, scale 1:25,000, Geospatial Information Authority of Japan(in Japanese).

Okada, A., Chida, N., Nakata, 1996. Kishiwada. Active Fault Map in Urban Area, scale 1:25,000, Geospatial Information Authority of Japan (in Japanese).

Okada, Y., Kasahara, K., Hori, S., Obara, K., Sekiguchi, S., Fujiwara, H., Yamamoto, A., 2004.qq K-NET and KiK-net-. Earth Planets Space 56, xv–xxviii.

Research Group for Active Faults of Japan, 1991. Active Faults in Japan. revised ed.University of Tokyo Press (437 pp., in Japanese with English abstract).

Schorlemmer, D., Wiemer, S., Wyss, M., 2005. Variations in earthquake-size distributionacross different stress regimes. Nature 437, 539–542. http://dx.doi.org/10.1038/nature04094.

Seno, T., Stein, S., Gripp, A.E., 1993. Amodel for themotion of the Philippine Sea Plate con-sistent with NUVEL-1 and geological data. J. Geophys. Res. 98 (B10), 17941–17948.http://dx.doi.org/10.1029/93JB00782.

Sibson, R.H., Xie, G., 1998. Dip range for intracontinental reverse fault ruptures: truth notstranger than friction? Bull. Seismol. Soc. Am. 88, 1014–1022.

Spada, M., Tormann, T., Wiemer, S., Enescu, B., 2013. Generic dependence of thefrequency-size distribution of earthquakes on depth and its relation to the strengthprofile of the crust. Geophys. Res. Lett. 40, 709–714. http://dx.doi.org/10.1029/2012GL054198.

Sugiyama, Y., 1994. Neotectonics of Southwest Japan due to the right-oblique subductionof the Philippine Sea plate. Geofis. Int. 33, 53–76.

Sugiyama, Y., Mizuno, K., Nanayama, F., Sugai, T., Yokota, H., Hosoya, T., Miura, K.,Takemura, K., Kitada, N., 2003. Study of blind thrust faults underlying Tokyo andOsaka urban areas using a combination of high-resolution seismic reflection profilingand continuous coring. Ann. Geophys. 46, 1071–1085. http://dx.doi.org/10.4401/ag-3446.

Suzuki, H., 1996. Geology of the Koto deep borehole observatory and geological structurebeneath the metropolitan area, Japan. Rep. Natl. Res. Inst. Earth Sci. Disaster Prev. 56,17–123.

Takahashi, H., 1982. The deep borehole observatories and their contribution for revealingthe characteristics of microearthquake activity in the Kanto district. Rep. Natl. Res.Cent. Disas. Prev. 28, 1–104 (in Japanese with English abstract).

Tanaka, Y., 1985. Crustal stress measurement and earthquake prediction. J. Geod. Soc. Jpn.31, 73–85.

Terakawa, T., Matsu'ura, M., 2010. The 3-D tectonic stress fields in and around Japaninverted from centroid moment tensor data of seismic events. Tectonics 29,TC6008. http://dx.doi.org/10.1029/2009TC002626.

Townend, J., 2006. What do faults feel? Observational constraints on the stresses actingon seismogenic faults. In: Abercrombie, R., McGarr, A., Kanamori, H., Toro, G.D.(Eds.), Earthquakes: radiated energy and the physics of earthquake faulting. Geo-physical Monograph Series 170, pp. 313–327.

Townend, J., Zoback, M.D., 2006. Stress, strain, and mountain building in central Japan.J. Geophys. Res. 111, B03411. http://dx.doi.org/10.1029/2005JB003759.

Tsukahara, H., Kobayashi, Y., 1991. Crustal stress in the central and western parts ofHonshu, Japan. Zisin 44, 221–231 (in Japanese with English abstract).

Urabe, T., Tsukada, S., 1991. A workstation-assisted processing system for waveform datafrom microearthquake networks. Abstracts of Spring Meeting of Seismological Socie-ty of Japan, p. 70 (in Japanese).

Wallace, R.E., 1951. Geometry of shearing stress and relation to faulting. J. Geol. 59,118–130.

Wessel, P., Smith, W.H.F., 1998. New, improved version of the Generic Mapping Tools re-leased. EOS Trans. AGU 79, 579.

Worum, G., van Wees, J.-D., Bada, G., van Balen, R.T., Cloetingh, S., Pagnier, H., 2004. Sliptendency analysis as a tool to constrain fault reactivation: a numerical approach ap-plied to three-dimensional fault models in the Roer Valley rift system (southeastNetherlands). J. Geophys. Res. 109, B02401. http://dx.doi.org/10.1029/2003JB002586.

Yang,W., Hauksson, E., 2011. Evidence for vertical partitioning of strike-slip and compres-sional tectonics from seismicity, focal mechanisms, and stress drops in the east LosAngeles basin area, California. Bull. Seismol. So. Am. 101 (3), 964–974.

Zoback, M.L., 1992. First- and second-order patterns of stress in the lithosphere: theworldstress map project. J. Geophys. Res. 97, 11703–11728.