express letter remotely triggered earthquakes in south...

February 8, 2015 13:17 Geophysical Journal International ggv037

5

10

15

20

25

30

35

40

45

50

55

60

67

72

77

82

87

92

97

102

107

112

117

122

Geophys. J. Int. (2015) doi: 10.1093/gji/ggv037

GJI Seismology

Q1 E X P R E S S L E T T E R

Remotely triggered earthquakes in South-Central Tibet following the2004 Mw 9.1 Sumatra and 2005 Mw 8.6 Nias earthquakes

Dongdong Yao, Zhigang Peng and Xiaofeng MengSchool of Earth and Atmospherical Sciences, Georgia Institute of Technology, Atlanta, GA 30332, USA. E-mail: [email protected]

Accepted 2015 January 19. Received 2015 January 19; in original form 2014 October 28

S U M M A R YWe conduct a systematic search for remotely triggered earthquakes in South-Central Tibetfollowing the 2004 December 26 Mw 9.1 Sumatra and 2005 March 28 Mw 8.6 Nias earthquakes.We apply a Network Waveform Matched Filter Technique (NWMFT) to continuous seismicdata recorded by the Hi-CLIMB array to obtain more complete earthquake catalogues in thisregion. Local earthquakes with magnitudes up to 4 were triggered during the large-amplitudesurface waves, and most of them occurred in Gaize immediately north of the Bangong-NujiangSuture Zone separating the Lhasa and Qiangtang Terranes. The triggered seismicity lasted50 hr following the Sumatra and a few hours following the Nias main shocks, respectively.The difference in triggering durations could be explained by the fact that the Sumatra mainshock excited long-duration surface waves with cumulative energy density 15 times more thanthat of the Nias main shock. In both cases, the transient increase in seismicity is followed bya moderate transient decrease, likely reflecting a ‘dynamic shadow effect’, where there is atemporary lack of seismic events following the remotely triggered seismicity. In comparison,we do not find any clear evidence of dynamic triggering (with a magnitude of completenessMc = 1.7) in the Zhongba region, where a Mw 6.3 normal faulting earthquake occurred 10 dafter the 2005 Nias event.

Key words: Earthquake interaction, forecasting, and prediction; Seismicity and tectonics;Statistical seismology.

1 I N T RO D U C T I O N

Earthquakes can interact with each other in a wide range of spatial-temporal windows. With the deployment of high-quality seismome-ters and advancement in processing techniques, many recent studieshave found that large earthquakes can trigger microearthquakes anddeep tectonic tremors far away from the rupture zone (e.g. Peng &Gomberg 2010; Hill & Prejean 2015, and references therein). Trig-gered events mostly occur during or immediately following theteleseismic surface waves and can be explained by frictional failureon critically stressed faults with dynamic Coulomb failure criterion(Hill 2012). However, in some cases, triggered events occur longafter the arrival of seismic waves, and this phenomenon is gener-ally referred to as ‘delayed dynamic triggering’ (e.g. Parsons 2005;Brodsky 2006).

So far most studies on dynamic triggering focus on plate bound-ary regions where seismic instrumentations are abundant (Hill &Prejean 2015). Only a few observations are made in intraplate re-gions (e.g. Velasco et al. 2008; Jiang et al. 2010; Wu et al. 2012).Q2

In this study, we conduct a systematical investigation of dynamictriggering in South-Central Tibet following the 2004 December 26Mw 9.1 Sumatra and 2005 March 28 Mw 8.6 Nias earthquakes.

These two events triggered numerous microearthquakes and tec-tonic tremors around the world (e.g. West et al. 2005; Miyazawa &Mori 2006; Ghosh et al. 2009; Peng et al. 2009; Tang et al. 2010;Chao et al. 2012). We select South-Central Tibet mainly because theMw 6.3 Zhongba earthquake occurred in this region on 2005 April07, ∼10 d after the Nias earthquake (Fig. 1a). Ryder & Burgmann(2011) speculated that the 2005 Zhongba earthquake may be delay-triggered by the Nias earthquake, due to their close timings. If true,the 10-d interval between the two events indicates that one or moresecondary triggering mechanisms (e.g. triggered creep; fluid migra-tion; fault weakening) may take place. Elevated seismic activitiescan be used to prove the existence of such secondary triggeringmechanisms (Anderson et al. 1994; Papadopoulos 2002; Shellyet al. 2011; van der Elst et al. 2013). Hence, a detailed examina-tion of continuous waveforms is needed to determine the existence(or absence) of dynamic triggering and reveal possible connectionsbetween the two events.

Previous studies on earthquake triggering mainly use mi-croearthquakes based on manual phase picking (e.g. Gomberg et al.2004; Peng et al. 2007; Wu et al. 2011; Aiken & Peng 2014), orautomatic detectors (e.g. Velasco et al. 2008) based on the short-time average/long-time average (STA/LTA) ratio method. However,

C⃝ The Authors 2015. Published by Oxford University Press on behalf of The Royal Astronomical Society. 1

mailto:[email protected]


129

134

139

144

149

154

159

164

169

174

179

184

191

196

201

206

211

216

221

226

231

236

241

246

2 D. Yao, Z. Peng and X. Meng

80˚ 90˚ 100˚0˚

10˚

20˚

30˚

2005 M8.6 Nias

2004 M9.0 Sumatra

Tibet Plateau

b)

2

4

6

8

10M

agnit

ude

0 400 800 1200 1600 2000Day since 2004/01/01

2004 2005 2006 2007 2008 2009 2010

07/11/2004Zhongba Eq

12/26/2004Sumatra Eq

03/28/2005Nias Eq



a)

29

30

31

32

33

34

83 84 85 86

2008/08/25 M6.69

2005/04/07 M6.28

2004/07/11 M6.18

2008/01/09 M6.40

H0800H1010

H1030H1050H1070

H1090H1110

H1130H1150

H1170H1190

H1210H1230

H1250H1270H1290

H1310

H1340H1360

H1380H1400

H1410

H1421H1423

H1450H1470

H1490H1510

H1530H1550

H1570

H1590

H1620

H0790

H0810

QiangtangTerrane

Lhasa Terrane

HimalayaTerrane

Zhongba

Gaize

Bangong Suture

50 km

Figure 1. (a) Magnitude versus time from 2004 to 2009 near Zhongba. (b) Map of Southeast Asia. Cyan and purple lines are major plate boundaries and majorblocks in China. The study region is shown as the white rectangle, and the triangles are stations in Hi-CLIMB project. The stations marked in red recordedboth Sumatra (green star) and Nias (blue star) main shocks. Beachballs are focal mechanisms of events with magnitude larger than 6.0 in the past 30 yr fromthe global CMT catalogue. (c) Map of study region in South Central Tibet. Yellow and cyan dots mark events used in the waveform matched filter analysis. Thefocal mechanisms of the 2004, 2005 and 2008 Zhongba earthquakes and the 2008 Nima-Gaize earthquake are plotted. Smaller beachballs are focal mechanismsof M < 6 events from the CMT catalogue in the past 30 yr. Green and blue lines are normal and strike-slip faults in Tibetan Plateau, respectively (Taylor & Yin2009).

manual picking can be time consuming, and while both methodsprovide timings and magnitudes, they do not provide locationsof the identified events. Recent studies mostly utilized a NetworkWaveform Matched Filter Technique (NWMFT) to detect miss-ing events immediately before or after large earthquakes (Shellyet al. 2007; Peng & Zhao 2009; Kato et al. 2012; Lengline et al.2012; Meng et al. 2013; Kato & Nakagawa 2014; Meng & Peng2014). This technique utilizes waveforms of known earthquakes astemplates to detect new events with high similarities in the con-tinuous waveforms over a network of seismometers. Depending onthe detection threshold and station coverage, this technique canidentify many times more events than that listed in standard cata-logues. Recently, the NWMFT has also been applied to detect/locateremotely triggered seismicity following large distant earthquakes(Yukutake et al. 2013; Wang et al. 2015). In this study, we used thesame technique to detect missing local events in the South-CentralTibet around the origin times of the 2004 Sumatra and 2005 Niasearthquakes.

2 S T U DY R E G I O N A N D DATA

The continuous collision between the Indian Plate and the EurasianPlate gives rise to the highest plateau in the world, the TibetanPlateau (Royden et al. 2008). It consists of different terranes(Fig. 1b), namely the Himalaya, Lhasa, Qiangtang, Songpan-Ganzi(or Bayan Har), and Kunlun terranes. The convergence betweenthe India and Eurasian plates causes relative motions among theseterrains and makes Tibetan Plateau one of the most complex tec-tonic environments in the world. Recent geodetic studies indicatethat the motion within the Tibetan Plateau is predominantly E–Wextension and N–S shortening (Zhang et al. 2004). The occurrencesof large normal faulting earthquakes with M > 6.0 in the past decadeare consistent with this observation (Fig. 1b). Among those normalfaulting earthquakes, three occurred in Zhongba county in 2004,2005 and 2008 (Ryder et al. 2012), including the Mw 6.3 earth-quake that occurred ∼10 d after the 2005 March 28 Mw 8.6 Niasearthquake.


253

258

263

268

273

278

283

288

293

298

303

308

315

320

325

330

335

340

345

350

355

360

365

370

Remotely triggered earthquakes 3

Figure 2. (a) 5-Hz high-pass filtered vertical-component waveforms aligned with epicentral distances during the 2005 Mw 8.6 Nias main shock. The dashedred and blue lines mark the predicted Love and Rayleigh wave arrival (with a nominal phase velocity of 4.1 and 3.5 km s−1). (b) Log10 envelope function of5-Hz high-passed seismograms at stations near Gaize (H1490 and H1500) and Zhongba (H1190 and H1200). The bottom three traces are broadband recordingsrotating to transverse, vertical and radial components. P, S, Love and Rayleigh phases are marked. Pink vertical dashed lines denote detected events, and cyanlines denote templates used for detection. (c) Spectrogram of vertical component recorded at station H1490.


377

382

387

392

397

402

407

412

417

422

427

432

439

444

449

454

459

464

469

474

479

484

489

494


The Tibetan Plateau has been extensively instrumented byPASSCAL and other temporary seismic deployments since 1990s.From 2002 to 2005, the Himalayan-Tibetan Continental Lithosphereduring Mountain Building (Hi-CLIMB) experiment (XF network)was conducted in South-Central Tibet, extending from the Gangeslowland, across the Himalayas and onto the central Tibetan Plateau(Fig. 1c). Over 200 sites were occupied during this experiment(Figs 1b and c), providing unprecedented continuous recordings forimaging crustal and upper mantle structures in this region (Nabeleket al. 2009). This network also recorded the 2004 Sumatra and the2005 Nias earthquakes and their numerous aftershocks. Fig. 2(a)shows an example of waveforms recorded at selected stations dur-ing the 2005 Nias main shock. After applying a nominal 5 Hzhigh-pass filter, it is evident that many high-frequency signals oc-curred during and immediately following the large-amplitude sur-face waves. By comparing the envelope function (Fig. 2b) and spec-trogram (Fig. 2c), we confirm that those are locally triggered earth-quakes. After picking their P- and S-wave arrivals, many of theseevents were located north of the Bangong suture zone near Gaize(Fig. S1). This finding motivated us to look further into the seismic-ity pattern in the Gaize region, as well as the epicentral region ofthe 2005 Zhongba earthquake.

3 A NA LY S I S P RO C E D U R E

Our analysis procedure mainly follows that of Wang et al. (2015) andis briefly described here. Using the Antelope software, we first man-ually picked and located events within two 1◦ × 1◦ grids: N30◦–31◦

and E83.5◦–84.5◦ around the Zhongba region and N32.5◦–33.5◦,E84◦–85◦ around Gaize county (Fig. 1c) between 2004 December01 and 2005 May 01. We also computed their local magnitudes us-ing the ‘dbevproc’ command within Antelope. During the 6-monthstudy period, we obtained 623 and 547 events near Zhongba andGaize, respectively. We then used these events as templates forwaveform detection. We applied a bandpass filter of 2–10 Hz toboth template and continuous waveforms in order to suppress tele-seismic signals. We utilized a 12 s time window (2 s before and10 s after the S and P arrivals are used for the two horizontal chan-nels and the vertical channel, respectively) to compute the cross-correlation coefficient (CCC) near Zhongba due to relatively largesource–receiver distance. For templates near Gaize, we used a 5 stime window (1 s before and 4 s after) to compute the CCC, mainlybecause most events occurred within 100 km of the array. Longerwindow with both P- and S-wave information helps suppress thebackground noise and enhance detection quality, but significantlyincrease computational cost. We conducted a simple test by utiliz-ing windows with different lengths to test how the window lengthaffects the detection result. As shown in Fig. S2, different detectingwindow length results in slightly different detection results, but theoverall temporal evolution of seismicity rate remains essentially thesame. Finally, we only employed template waveforms with signal-to-noise ratio (SNR) no less than 5, following Peng & Zhao (2009).

We defined the detection threshold as the mean CCC plus 12 timesthe median absolute deviation (MAD) near Zhongba. Since there arefewer stations near Gaize, we used a higher detection threshold (i.e.15 times MAD) in this region to remove possible false detectionsdue to the small number of stations. We assigned the epicentreof a detected event to be the same as the corresponding templatewith the highest CCC value (Peng & Zhao 2009). In addition, weestimated magnitudes of detected events by computing the peakS-wave amplitude ratio of the detected and corresponding template

Q3

events. Fig. 3 shows an example of a positive detection near Zhongba(Peng & Zhao 2009). The template event has a local magnitudeof 3.74 and occurred on 2005 April 08, 10:47:04. The detectedevent occurred earlier on the same day at 01:25:21, and the inferredmagnitude is 2.34. The detected event is not identified manually,mainly because of its relative low SNR.

To avoid potential biases by pre-defining the two study regions(i.e. Zhongba and Gaize), we also performed a slightly differentanalysis. In this case, we first manually located all triggered earth-quakes during the surface wave trains of the 2004 Sumatra and 2005Nias events if possible. Next, we utilized these events (Fig. S1) astemplates for scanning through the continuous waveforms at longertimes before and after both main shocks. The detailed analysis pro-cedures and results were summarized in Supporting Information.

4 R E S U LT S

4.1 Detection results near Zhongba

Using 623 earthquakes near Zhongba as templates, we detecteda total of 1100 and 6453 events around the 2004 Sumatra event(2004 December 25–2005 January 09) and 2005 Nias event (2005March 21–2005 April 14), respectively. We computed the magni-tude of completeness (Mc) for the detected catalogues using thebest-combined method in ZMAP (Wiemer 2001). The obtainedvalues are 1.1 and 1.7 around the Sumatra and Nias main shocks,respectively (Fig. S3). Although we detected 6453 events aroundthe 2005 Nias main shock in Zhongba, 6278 of them (∼95 percent) occurred shortly after and around the epicentre of the 2005Zhongba event, which are considered as its aftershocks. We foundno clear change in seismic activity between the 2005 Nias and 2005Zhongba earthquakes, nor did we observe a clear change in seismicactivity following the 2004 Sumatra earthquake (Figs 4a and c).We evaluated the significance of seismicity changes by computingthe β-value, which measures the differences between the observednumbers of events after a main shock and the predicted numbersbased on the rates before the main shock (Matthews & Reasenberg1988; Aron & Hardebeck 2009). If the resulting β-value is greaterthan 2, it indicates a significant increase in the seismicity rate. Asignificant decrease occurs when the β-value is smaller than −2. Wecomputed the β-value in the time windows of 1–24 hr after the mainshocks with 1-hr increment (Fig. S4). To avoid potential bias in theresulting β-value with different pre-main shock time windows, weused the longest pre-main shock time to estimate the backgroundrate (250 hr before Sumatra and 180 hr before Nias). The result-ing β-value shows moderate to significant decrease of seismicityrate following the two main shocks in all time windows (Fig. S3), Q4

confirming our visual observation.

4.2 Detection results near Gaize

In comparison, many microearthquakes occurred during and imme-diately after the passage of the teleseismic surface waves of bothmain shocks near Gaize (Figs 2 and S5). The largest triggered earth-quakes have local magnitudes of 3.64 and 3.80, respectively, andboth were instantaneously triggered during the large-amplitude sur-face waves. Near Gaize, we used 547 templates for detection aroundthe 2004 Sumatra and 2005 Nias events. As a result, 1350 and 2500events are detected 10 d before and 15 d after the two main shocks,respectively. The Mc is −0.6 and −0.4 for the Sumatra and Niascases (Fig. S3), respectively. Clear increases of microearthquakes


501

506

511

516

521

526

531

536

541

546

551

556

563

568

573

578

583

588

593

598

603

608

613

618


5145 5150 5155 5160 5165 5170 5175 5180H1120.BHE 0.27H1120.BHN 0.41H1120.BHZ 0.19H1130.BHE 0.17H1130.BHN 0.27H1130.BHZ 0.26H1150.BHE 0.61H1150.BHN 0.56H1150.BHZ 0.38H1160.BHE 0.62H1160.BHN 0.68H1160.BHZ 0.28H1190.BHE 0.72H1190.BHN 0.60H1190.BHZ 0.48H1200.BHE 0.28H1200.BHN 0.62H1200.BHZ 0.56

c)

Seconds since 2005/04/08 00:00:00

4500 5000 5500 6000 6500 7000 7500 8000−0.2

00.20.4 a)

Seconds since 2005/04/08 00:00:00

Mea

n CC

105103101

b)

Number

Figure 3. An example of positive detection by template 20050408104704 (M 3.74) with mean cross-correlation coefficient (CCC) value = 0.583 and aninferred magnitude 2.34. (a) Distribution of CCC value 1000 s before and 3000 s after the origin time of the detection (red circle). Red dashed line marks thethreshold for positive detections (12 times the median absolute deviation). (b) Histogram of the CCC value. (c) Waveform comparison of template waveforms(blue) and continuous waveforms (black). The station name and channel as well as corresponding CCC value are shown on both sides.

Figure 4. Detection results near Zhongba. (a) Magnitude versus origin time of detected events around the 2004 Sumatra event. The red dots mark the templateevents. Blue and cyan lines denote cumulative number for all events and events with magnitude larger than Mc, respectively. (b) A zoom-in plot showing thedetections 50 hr before and 100 hr after the Sumatra main shock. (c) Detection results around the 2005 Nias event near Zhongba. Symbols and notations arethe same as in (a). d) A zoom-in plotting showing 50 hr before and 100 hr after the Nias event.


625

630

635

640

645

650

655

660

665

670

675

680

687

692

697

702

707

712

717

722

727

732

737

742


Figure 5. Detection results near Gaize. All symbols and notations are the same as in Fig. 4.

are found following both main shocks (Figs 5b and d). The seismic-ity rate increase was steady immediately following surface wavesof the Sumatra main shock and lasted nearly 2 d. The seismicityrate suddenly dropped below the pre-main shock level afterwards(Fig. 5b). In comparison, the increase following the Nias main shockwas dominated by two bursts, one right after the surface waves andanother one ∼2 hr later. To avoid potential bias from the muchlower Mc near Gaize, we also used the same Mc as in Zhongba, andthe rate changes remain essentially unchanged (Fig. S6). We alsocomputed the β-values, and the results suggest significant seismic-ity rate increase in most time windows after both events (Fig. S4).We noted clear differences in the β-value results between the 2004Sumatra and 2005 Nias events. The β-value generally increasedwith time following the Sumatra event, but decreased followingthe Nias event. In addition, we performed a sliding-window β-value analysis in a much longer time window (Meng & Peng 2014)(Fig. S7). In both cases, we found a moderate seismicity rate de-crease following a significant seismicity rate increase for the Suma-tra and Nias events.

5 D I S C U S S I O N S

In this study, we documented to our knowledge the first obser-vation of dynamically triggered seismicity within Tibet Plateau.Events triggered by the 2005 Nias earthquake mainly occurred nearGaize in the Qiangtang terrain. On the other hand, the 2004 Suma-tra earthquake triggered earthquakes at other places outside of theGaize region (Figs S1 and S5). In addition, the temporal increase ofseismicity lasted for at least 2 d and a few hours for the Sumatra andNias main shocks, respectively. Both temporal increases of seismic-ity rate were followed by a moderate rate decrease (Fig. S7). Thistype of rate reduction following a significant rate increase is similarto recent observations of triggered tremor in southern Taiwan fol-lowing the 2011 Mw 9.1 Tohoku-Oki earthquake (Sun et al. 2015),and a global quiescence of M > 5.5 earthquake after a transient in-crease following the 2012 Mw 8.6 Indian Ocean earthquake (Pollitz

et al. 2014). In these cases, the reduction can be best explained as a‘dynamic shadow effect’, where most fault patches close to failureare triggered by a large teleseismic earthquake, resulting in a periodof seismic quiescence.

One would argue that the seismic quiescence could be artificialdue to the fact that NWMFT can only detect events around thosetemplates. To evaluate this further, we examined the matching pat-tern to evaluate whether rerupturing of the same fault patch occurredin the 3 months after the Sumatra main shock. As shown in Fig. S8,the events triggered by the Sumatra (0–100 hr) and Nias events(0–10 hr) do not overlap and hence may rupture different fault seg-ments. This is consistent with our inference that the fault segmentsthat ruptured immediately after the Sumatra event were not readyto be triggered/ruptured again when the Nias event occurred. Inaddition, the detected events immediately after the Sumatra eventare exclusively matched by templates before the Nias event, whilethe detected events after the Nias event are matched by templatesaround it (Fig. S9). These observations further corroborate our in-terpretation of dynamic stress shadow.

The seismicity triggered by the Sumatra main shock did notfollow the Omori-law type decay (e.g. Brodsky 2006). Instead itshowed a steady rate increase for ∼50 hr, and then a sharp ratereduction (Fig. 5). A similar pattern of triggered deep tremor wasidentified along the central San Andreas Fault and was explainedas secondary triggering by aseismic fault slip (Shelly et al. 2011).On the other hand, the M∼4 event triggered near Gaize by the Niasevent was followed by an Omori-like aftershock sequence (Fig. 5d).Another similar Omori-like sequence occurred ∼2 hr after the Niasevent, but no clear main shock was identified. The different be-haviours near Gaize following the Sumatra and Nias events couldbe attributed to the fact that most of triggered earthquakes dur-ing the Nias surface waves are clustered, while the Sumatra mainshock triggered events on different sites (Fig. S1). The mean peakground velocities (PGVs) across all stations are 0.93 ± 0.25 and0.26 ± 0.04 cm s−1 during the Sumatra and Nias main shocks,respectively. These PGVs correspond to a factor of 4 differencesin dynamic stresses (93 and 25 kPa), assuming a nominal phase


749

754

759

764

769

774

779

784

789

794

799

804

811

816

821

826

831

836

841

846

851

856

861

866


velocity of 3.5 km s−1 and crustal rigidity of 35 GPa (Aiken & Peng2014). In comparison, the cumulative energy density (Brodsky &Prejean 2005) for the Sumatra main shock is about 15 times morethan that for the Nias main shock (Fig. S10), likely due to its longrupture duration (Lay et al. 2005). Hence, we suggested that long-duration surface waves of the 2004 Sumatra main shock, togetherwith its high PGVs, help to excite more regions in South-CentralTibet and last longer than during the Nias main shock. The detectionresults for the entire region (Fig. 1c) also support this conclusion(Fig. S11).

The M∼6 Zhongba sequences are ∼300 km away from the Gaizeregion. Unfortunately, neither the 2004 nor the 2008 event wasrecorded by the Hi-CLIMB network. The predicted dynamic stressfrom the 2005 event is ∼75 kPa. However, we did not observeclear seismicity rate change near Gaize around the 2005 Zhongbaevent (Fig. 5c). One possible reason could be due to the ‘dy-namic shadow effect’ after the 2005 Nias main shock. The criticallystressed patches near Gaize ruptured when the surface waves ofthe 2005 Nias earthquake passed by and were not ready when the2005 Zhongba event occurred. Alternatively, triggering in this re-gion could be frequency dependent (e.g. Brodsky & Prejean 2005),such that only long-period surface waves from very large and distantearthquakes are capable of triggering seismicity.

As mentioned before, a Mw 6.3 local event occurred in Zhongba∼10 d after the 2005 Mw 8.6 Nias earthquake, suggesting a possibletriggering relationship (Ryder & Burgmann 2011). However, ourdetection results did not show any clear increase in seismicity ratefollowing the Nias event that could support this hypothesis (Fig. 4).We noticed that the Mc for the detected catalogues near Zhongba (1.1to 1.7) is about one magnitude larger than near Gaize (−0.6 to −0.4),mainly due to the differences in epicentral distances. Hence, it ispossible that smaller-magnitude events (e.g. <1) may be triggerednear Zhongba during the Nias earthquake, but were not detected byour NWMFT. Another possibility is that the Nias earthquake mainlytriggered aseismic slip in the Zhongba region, which could not bedetected by the seismic method. In any case, while we could notcompletely rule out a causal relationship between the Mw 8.6 Niasand Mw 6.3 Zhongba earthquakes, we did not find any clear changein local seismic activity that can be used to link these two events(e.g. Anderson et al. 1994; Papadopoulos 2002; van der Elst et al.2013).

The study region near Gaize is mainly characterized by northeast-ern trending left-lateral strike slip faults and northern–northwesterntrending normal faults (Kapp et al. 2005; Ryder et al. 2010), whichagrees with normal-faulting mechanisms of recent earthquakes(Fig. 1) and is similar to the Zhongba region further south. Thisis consistent with current observations that dynamically triggeredmicroearthquakes mostly occur in extensional or trans-extensionalregions (Hill & Prejean 2015). However, it is still not clear why bothmain shocks trigger near Gaize, but not near Zhongba, despite thefact that the Zhongba region is slightly closer to the main shocksand hence would receive higher dynamic stress perturbations. Onemajor factor is that there was another Mw 6.2 (Ms 6.9) earthquakenear Zhongba on 2004 July 11, while no major earthquakes oc-curred near Gaize in 2004–2005. Hence, it is possible that the 2004Zhongba earthquake and its aftershocks released most of the ac-cumulated tectonic stresses such that the region is no longer in acritical state (e.g. Brodsky & van der Elst 2014). However, this is in-consistent with the fact that another Mw 6.3 earthquake occurred inZhongba 10 d after the Nias main shock and previous observationsthat dynamic triggering preferentially occurs in aftershock regionsof previous large earthquakes (Hough et al. 2003; Jiang et al. 2010).

We note that the Zhongba region has higher background seismic-ity rate than the Gaize region (i.e. larger a value in Fig. S3), but mosttriggered activity was found in the Gaize region. This observation isalso inconsistent with recent observations that regions with higherbackground rate (i.e. geothermal and/or aftershock regions of re-cent/historic large events) are more susceptible to dynamic trigger-ing (Hough et al. 2003; Jiang et al. 2010; Aiken & Peng 2014). Theheat flow map in this region shows a higher flux near Zhongba in theLhasa Terrane (95–105 mW m−2), as compared to 70–80 mW m−2

near Gaize in the Qiangtang Terrane (Tao & Shen 2008), which doesnot support the observation that geothermal/volcanic regions withhigh heat flows favor triggering of microearthquakes (Aiken & Peng2014; Hill & Prejean 2015). In addition, previous studies revealedthat the Qiangtang Terrane has a thicker seismogenic zone thanLhasa Terrane (Wei et al. 2010). However, it is not clear how thiswould affect the triggering behaviour. Further studies are needed tobetter understand the differences in triggering behaviours in theseregions. This will be a subject of future research. Nevertheless, ourobservations presented here, along with recent studies (e.g. Hill &Prejean 2015), clearly demonstrate that large earthquakes are capa-ble of dynamically triggering microearthquakes up to magnitude 4in active intraplate regions.

A C K N OW L E D G E M E N T S

The seismic data used in this study is obtained from the IRISDMC (http://www.iris.edu/mda/XF?timewindow=2002-2005) andprocessed with Antelope (http://www.brtt.com/), GMT and SACtools. We thank Jacob Walter for helping to build the Antelopedatabase. We also thank Chastity Aiken, Xiaofeng Liang andChunquan Wu for valuable comments. This work is supported byNSF CAREER award EAR-095605.

R E F E R E N C E S

Aiken, C. & Peng, Z., 2014. Dynamic triggering of microearthquakes inthree geothermal/volcanic regions of California, J. geophys. Res., 119(9),6992–7009.

Anderson, J.G., Brune, J.N., Louie, J.N., Zeng, Y., Savage, M., Yu, G., Chen,Q. & dePolo, D., 1994. Seismicity in the Western Great basin apparentlytriggered by the Landers, California, Earthquake, 28 June 1992, Bull.seism. Soc. Am., 84(3), 863–891.

Aron, A. & Hardebeck, J.L., 2009. Seismicity rate changes along the centralCalifornia coast due to stress changes from the 2003 M6.5 San Simeonand 2004 M6.0 Parkfield earthquakes, Bull. seism. Soc. Am., 99(4),2280–2292.

Brodsky, E.E., 2006. Long-range triggered earthquakes that continue afterthe wave train passes, Geophys. Res. Lett., 33, L15313.

Brodsky, E.E. & Prejean, S.G., 2005. New constraints on mechanisms ofremotely triggered seismicity at Long Valley Caldera, J. geophys. Res.,110, B04302, doi:10.1029/2004JB003211.

Brodsky, E.E. & van der Elst, N.J., 2014. The use of dynamic earthquaketriggering, Annu. Rev. Earth planet. Sci., 42, 317–339.

Chao, K., Peng, Z., Wu, C., Tang, C.-C. & Lin, C.-H., 2012. Remote trig-gering of non-volcanic tremor around Taiwan, Geophys. J. Int., 188,301–324.

Ghosh, A., Vidale, J.E., Peng, Z., Creager, K.C. & Houston, H., 2009.Complex nonvolcanic tremor near Parkfield, California, triggered bythe great 2004 Sumatra earthquake, J. geophys. Res., 114, B00A15,doi:10.1029/2008JB006062.

Gomberg, J., 2010. Lessons from (triggered) tremor, J. geophys. Res., 115,B10302, doi:10.1029/2009JB007011. Q5

http://www.iris.edu/mda/XF?timewindow=2002-2005

http://www.brtt.com/


873

878

883

888

893

898

903

908

913

918

923

928

935

940

945

950

955

960

965

970

975

980

985

990


Gomberg, J., Bodin, P., Larson, K. & Dragert, H., 2004. Earthquake nucle-ation by transient deformations caused by the M = 7.9 Denali, Alaska,earthquake, Nature, 427, 621–624.

Hill, D.P., 2012. Dynamic stresses, Coulomb failure, and remote triggering-corrected, Bull. seism. Soc. Am., 102, 2313–2336.

Hill, D.P. & Prejean, S., 2015. Dynamic triggering, in Treatise onGeophysics, Vol. 4, 2nd edn, ed. Kanamori, H., Elsevier.

Hough, S.E., Seeber, L. & Armbruster, J.G., 2003. Intraplate triggeredearthquakes: observations and interpretation, Bull. seism. Soc. Am., 93,2212–2221.

Jiang, T., Peng, Z., Wang, W. & Chen, Q.-F., 2010. Remotely triggeredseismicity in continental China following the 2008 Mw 7.9 Wenchuanearthquake, Bull. seism. Soc. Am., 100(5B), 2574–2589.

Kapp, P., Yin, A., Harrison, T.M. & Ding, L., 2005. Cretaceous-Tertiaryshortening, basin development, and volcanism in central Tibet, Geol.Soc. Am. Bull., 117, 865–878.

Kato, A. & Nakagawa, S., 2014. Multiple slow-slip events during a foreshocksequence of the 2014 Iquique, Chile Mw 8.1 earthquake, Geophys. Res.Lett., 41(15), 5420–5427.

Kato, A., Obara, K., Igarashi, T., Tsuruoka, H., Nakagawa, S. & Hirata, N.,2012. Propagation of slow slip leading up to the 2011 Mw 9.0 Tohoku-Okiearthquake, Science, 335, 705–708.

Kilb, D., Gomberg, J. & Bodin, P., 2002. Aftershock triggering by completeCoulomb stress changes, J. geophys. Res., 107(B4), ESE 2-1–ESE 2-14.Q6

Lay et al., 2005. The great Sumatra-Andaman earthquake of 26 December2004, Science, 308(5725), 1127–1133.

Lengline, O., Enescu, B., Peng, Z. & Shiomi, K., 2012. Decay and expansionof the early aftershock activity following the 2011, Mw 9.0 Tohoku earth-quake, Geophys. Res. Lett., 39, L18309, doi:10.1029/2012GL052797.

Matthews, M.V. & Reasenberg, P.A., 1988. Statistical methods for inves-tigating quiescence and other temporal seismicity patterns, Pure appl.Geophys., 126, 357–372.

Meng, X. & Peng, Z., 2014. Seismicity rate changes in the San Jacinto FaultZone and the Salton Sea Geothermal Field following the 2010 Mw 7.2 ElMayor-Cucapah earthquake, Geophys. J. Int., 197(3), 1750–1762.

Meng, X., Peng, Z. & Hardebeck, J.L., 2013. Seismicity around Parkfieldcorrelates with static shear stress changes following the 2003 Mw 6.5 SanSimeon earthquake, J. geophys. Res., 118, 3576–3591.

Miyazawa, M. & Mori, J., 2006. Evidence suggesting fluid flow beneathJapan due to periodic seismic triggering from the 2004 Sumatra-Andamanearthquake, Geophys. Res. Lett., 33, L05303.

Nabelek et al., 2009. Underplating in the Himalaya-Tibet Collison ZoneRevealed by the Hi-CLIMB Experiment, Science, 325, 1371–1374.

Papadopoulos, G.A., 2002. The Athens, Greece, earthquake (Ms 5.9) of 7September 1999: an event triggered by the Izmit, Turkey, 17 August 1999earthquake? Bull. seism. Soc. Am., 92(1), 312–321.

Parsons, T., 2005. A hypothesis for delayed dynamic earthquake triggering,Geophys. Res. Lett., 32, L04302, doi:10.1029/2004GL021811.

Peng, Z. & Gomberg, J., 2010. An integrated perspective of the continuumbetween earthquakes and slow-slip phenomena, Nat. Geosci., 3, 599–607.

Peng, Z. & Zhao, P., 2009. Migration of early aftershocks following the 2004Parkfield earthquake, Nat. Geosci., 2, 877–881.

Peng, Z., Vidale, J.E., Ishii, M. & Helmstetter, A., 2007. Seismic-ity rate immediately before and after main shock rupture fromhigh-frequency waveforms in Japan, J. geophys. Res., 112, B03306,doi:10.1029/2006JB004386.

Peng, Z., Vidale, J.E., Wech, A.G., Nadeau, R.M. & Creager, K.C., 2009.Remote triggering of tremor along the San Andreas Fault in centralCalifornia, J. geophys. Res., 114, B00A06, doi:10.1029/2008JB006049.

Peng, Z., Wang, W., Chen, Q.-F. & Jiang, T., 2010. Remotely triggered seis-micity in north China following the 2008 Mw 7.9 Wenchuan earthquake,Earth Planets Space, 62, 893–898.Q7

Pollitz, F.F., Burgmann, R., Stein, R.S. & Sevilgen, V., 2014. The profoundreach of the 11 April 2012 M 8.6 Indian Ocean earthquake: short-termglobal triggering followed by a longer-term global shadow, Bull. seism.Soc. Am., 104(2), 972–984.

Royden, L.H., Burchfiel, B.C. & van der Hilst, R.D., 2008. The geologicalevolution of the Tibetan Plateau, Science, 321, 1054–1058.

Ryder, I. & Burgmann, R., 2011. Communication between multiple largeearthquakes at different spatial scales across and beyond the TibetanPlateau, in American Geophysical Union, Fall Meeting 2011, Abstract#S21D-07.

Ryder, I., Burgmann, R. & Sun, J., 2010. Tandem afterslip on connected faultplanes following the 2008 Nima-Gaize (Tibet) earthquake, J. geophys.Res., 115, B03404, doi:10.1029/2009JB006423.

Ryder, I., Burgmann, R. & Fielding, E., 2012. Static stress interactions inextensional earthquake sequences: an example from the South LunggarRift, Tibet, J. geophys. Res., 117, B09405, doi:10.1029/2012JB009365.

Shelly, D.R., Beroza, G.C. & Ide, S., 2007. Non-volcanic tremor and low-frequency earthquake swarms, Nature, 446, 305–307.

Shelly, D.R., Peng, Z., Hill, D.P. & Aiken, C., 2011. Triggered creep as apossible mechanism for delayed dynamic triggering of tremor and earth-quakes, Nat. Geosci., 4, 384–388.

Sun, W.-F., Peng, Z., Lin, C.-H. & Chao, K., 2015. Detecting deep tectonictremor in Taiwan using dense arrays, Bull. seism. Soc. Am., in press. Q8

Tang, C.-C., Peng, Z., Chao, K., Chen, C.-H. & Lin, C.-H., 2010. Detect-ing low-frequency earthquakes within non-volcanic tremor in southernTaiwan triggered by the 2005 Mw 8.6 Nias earthquake, Geophys. Res.Lett., 37, L16307, doi:10.1029/2010GL043918.

Tao, W. & Shen, Z., 2008. Heat flow distribution in Chinese continent andits adjacent areas, Prog. Natural Sci., 18, 843–849.

Taylor, M. & Yin, A., 2009. Active structures of the Himalayan-Tibetanorogen and their relationships to earthquake distribution, contemporarystrain field, and Cenozoic volcanism, Geosphere, 5(3), 199–214.

van der Elst, N.J., Brodsky, E.E. & Lay, T., 2013. Remote triggering notevident near the epicenters of impending great earthquakes, Bull. seism.Soc. Am., 103, 1522–1540.

Velasco, A.A, Hernandez, S., Parsons, T. & Pankow, K., 2008. Global ubiq-uity of dynamic earthquake triggering, Nat. Geosci., 1, 375–379.

Wang, W., Meng, X., Peng, Z., Chen, Q. & Liu, N., 2015. Background seismicactivities and dynamic triggering behaviors around Fangshan Pluton nearBeijing, China, J. geophys. Res., in press. Q9

Wei et al., 2010. Regional earthquakes in northern Tibetan Plateau: implica-tions for lithospheric strength in Tibet, Geophys. Res. Lett., 37, L19307,doi:10.1029/2010GL044800.

Wessel, P., McNutt, W.H.F., Scharroo, R., Luis, J. & Wobbe, F., 2013. GenericMapping Tools: improved version released, EOS, Trans. Am. geophys.Un., 94(45), 409–410. Q10

West, M., Sanchez, J.J. & McNutt, S.R., 2005. Periodically triggered seis-micity at Mount Wrangell, Alaska, after the Sumatra Earthquake, Science,308(5725), 1144–1146.

Wiemer, S., 2001. A software package to analyze seismicity: ZMAP, Seism.Res. Lett., 72(3), 373–382.

Wu, C., Peng, Z., Wang, W. & Chen, Q.-F., 2011. Dynamic triggering ofshallow earthquakes near Beijing, China, Geophys. J. Int., 185(3), 1321–1334.

Yukutake, Y., Miyazawa, M., Honda, R., Harada, M., Ito, H., Sakaue, M.,Koketsu, K. & Yoshida, A., 2013. Remotely triggered seismic activityin Hakone volcano during and after the passage of surface waves fromthe 2011 M9.0 Tohoku-Oki earthquake, Earth planet. Sci. Lett., 373,205–216.

Zhang, P., Shen, Z., Wang, M., Gan, W., Burgmann, R. & Molnar, P., 2004.Continuous deformation of the Tibetan Plateau from global positioningsystem data, Geology, 32, 809–812.

S U P P O RT I N G I N F O R M AT I O N

Additional Supporting Information may be found in the online ver-sion of this paper:

Figure S1. Red triangles are stations used in the XF network.The 2004, 2005 and 2008 Zhongba earthquakes, as well as 2008Gaize-Nima earthquake are marked with their focal mechanisms.Other beachballs are events listed in the CMT catalogue in the past30 yr. Different terranes are labelled, while pink lines indicate the


997

1002

1007

1012

1017

1022

1027

1032

1037

1042

1047

1052

1059

1064

1069

1074

1079

1084

1089

1094

1099

1104

1109

1114


boundaries between different terranes. Events occurred 5 hr after the2004 Sumatra and 2005 Nias events are shown with filled blue andgreen circles, respectively. Cyan filled circles are other templatesused in Fig. S11.Figure S2. Comparison of detection results with different windowlengths. (a) (CCC-mean CCC)/MAD versus time in both cases.(b) CCC versus time relative to the 2005 Nias main shock. In bothpanels, blue and red circles correspond to 5 s window and 20 swindow, respectively.Figure S3. Gutenberg–Richter (G–R) relationship at differentspace–time windows. Diamonds show the cumulative number ofearthquakes, black triangles are number of events for different mag-nitude bins. Top two panels show detected catalogues around the2004 Sumatra and 2005 Nias events near Zhongba, and bottomtwo are near Gaize. Red bold curves are maximum-likelihood G–Rfitting. The Mc and a value are labelled in each panel.Figure S4. β-Value for varying window length of 1–24 hr with 1-hrincrement. Top two panels show results near Zhongba, and bottomtwo show results near Gaize. The longest pre-main shock windowis used for calculating background seismicity rate (250 hr beforeSumatra and 180 hr before Nias). Blue horizontal line shows 95 percent confidence level.Figure S5. (a) 5-HZ high-passed vertical-component waveformsaligned with epicentral distances during the 2004 Mw 9.1 Sumatramain shock. The dashed red and blue lines mark the predicted Loveand Rayleigh wave arrival (with a nominal phase velocity of 4.1 and3.5 km s−1). (b) Log10 envelope function of 5-Hz high-passed seis-mograms at stations near Gaize (H1490 and H1510) and Zhongba(H1190 and H1200). The bottom three traces are broadband record-ings rotating to transverse, vertical and radial components. P, S,Love and Rayleigh phases are marked. Blue vertical dashed linesare detected events and cyan lines are templates used for detection.(c) Spectrogram of vertical component recorded at station H1490.Figure S6. Comparison of the seismicity rate change patternsnear Gaize with different Mc (blue shows values near Zhongba).

Left-hand panel is around the 2004 Sumatra event, while rightaround the 2005 Nias event.Figure S7. Sliding window β analysis after the 2004 Sumatra(a) and 2005 Nias events (b). A 40-hr window is used for Sumatracase and 2-hr window for Nias. The longest pre-main shock timewindow is used for computing background seismicity rate (samewith Fig. S4). Two horizontal lines mark value of 2 and –2.Figure S8. Spatial distribution of template events near Gaize. Blackopen circles are all templates used for detection, while green andblue ones are post-Sumatra (0–100 hr after the 2004 Sumatra) andpost-Nias (0–10 hr after the 2005 Nias) templates, respectively.Figure S9. (a) Templates used for detection near Gaize. Green andblue filled circles are templates that detect events during the shadedperiods in panel (b) and (c), respectively. (b) and (c) Detection re-sults around the 2004 Sumatra and 2005 Nias event, respectively.Open and filled grey circles denote all detections and events withmagnitude higher than Mc, respectively. Shaded periods are win-dows showing in Figs 5(b) and (d).Figure S10. (a) The velocity seismogram within 3000 s after the2004 Sumatra and 2005 Nias main shocks recorded by stationH1490. (b) Average cumulative energy density (ACED) within3000 s after the Sumatra and Nias main .shocks. The ACEDof the Sumatra event is 15.3 times the Nias event in first 3000s and the source duration lasts for longer time around Sumatraevent.Figure S11. (a) Detection results around the 2004 Sumatra eventin the separate analysis. Blue curve shows the cumulative numberof detected events, and the red curve is the result near Gaize inthe main text. (b) The result around the 2005 Nias event (http://gji.oxfordjournals.org/lookup/suppl/doi:10.1093/gji/ggv037/-/DC1).

Please note: Oxford University Press is not responsible for the con-tent or functionality of any supporting materials supplied by theauthors. Any queries (other than missing material) should be di-rected to the corresponding author for the paper.

http://gji.oxfordjournals.org/lookup/suppl/doi:10.1093/gji/ggv037/-/DC1

http://gji.oxfordjournals.org/lookup/suppl/doi:10.1093/gji/ggv037/-/DC1

express letter remotely triggered earthquakes in south...

Documents