Triggering of earthquakes during the 2000 Papua New Guinea earthquake sequence

Similar documents
Centroid moment-tensor analysis of the 2011 Tohoku earthquake. and its larger foreshocks and aftershocks

Centroid-moment-tensor analysis of the 2011 off the Pacific coast of Tohoku Earthquake and its larger foreshocks and aftershocks

COULOMB STRESS CHANGES DUE TO RECENT ACEH EARTHQUAKES

SOURCE MODELING OF RECENT LARGE INLAND CRUSTAL EARTHQUAKES IN JAPAN AND SOURCE CHARACTERIZATION FOR STRONG MOTION PREDICTION

Slip distributions of the 1944 Tonankai and 1946 Nankai earthquakes including the horizontal movement effect on tsunami generation

RELOCATION OF THE MACHAZE AND LACERDA EARTHQUAKES IN MOZAMBIQUE AND THE RUPTURE PROCESS OF THE 2006 Mw7.0 MACHAZE EARTHQUAKE

Analysis of Seismological and Tsunami Data from the 1993 Guam Earthquake

Earthquake Doublet Sequences: Evidence of Static Triggering in the Strong Convergent Zones of Taiwan

Source of the July 2006 West Java tsunami estimated from tide gauge records

Source rupture process of the 2003 Tokachi-oki earthquake determined by joint inversion of teleseismic body wave and strong ground motion data

Coulomb stress change for the normal-fault aftershocks triggered near the Japan Trench by the 2011 M w 9.0 Tohoku-Oki earthquake

Widespread Ground Motion Distribution Caused by Rupture Directivity during the 2015 Gorkha, Nepal Earthquake

Synthetic Seismicity Models of Multiple Interacting Faults

Empirical Green s Function Analysis of the Wells, Nevada, Earthquake Source

Source Characteristics of Large Outer Rise Earthquakes in the Pacific Plate

Seismic Activity near the Sunda and Andaman Trenches in the Sumatra Subduction Zone

Coulomb stress changes due to Queensland earthquakes and the implications for seismic risk assessment

Rupture Characteristics of Major and Great (M w 7.0) Megathrust Earthquakes from : 1. Source Parameter Scaling Relationships

Spatial and Temporal Distribution of Slip for the 1999 Chi-Chi, Taiwan, Earthquake

Scaling relations of seismic moment, rupture area, average slip, and asperity size for M~9 subduction-zone earthquakes

Coseismic slip distribution of the 1946 Nankai earthquake and aseismic slips caused by the earthquake

Earth Tides Can Trigger Shallow Thrust Fault Earthquakes

Slip Partition of the 26 December 2006 Pingtung, Taiwan (M 6.9, M 6.8) Earthquake Doublet Determined from Teleseismic Waveforms

EARTHQUAKE SOURCE PARAMETERS FOR SUBDUCTION ZONE EVENTS CAUSING TSUNAMIS IN AND AROUND THE PHILIPPINES

Multi-planar structures in the aftershock distribution of the Mid Niigata prefecture Earthquake in 2004

revised October 30, 2001 Carlos Mendoza

RELATION BETWEEN RAYLEIGH WAVES AND UPLIFT OF THE SEABED DUE TO SEISMIC FAULTING

Moment tensor inversion of near source seismograms

SUPPLEMENTARY INFORMATION

Onto what planes should Coulomb stress perturbations be resolved?

Magnitude 7.5 NEW BRITAIN REGION, PAPUA NEW GUINEA

GEOPHYSICAL RESEARCH LETTERS, VOL. 31, L19604, doi: /2004gl020366, 2004

RELOCATION OF LARGE EARTHQUAKES ALONG THE PHILIPPINE FAULT ZONE AND THEIR FAULT PLANES

ON NEAR-FIELD GROUND MOTIONS OF NORMAL AND REVERSE FAULTS FROM VIEWPOINT OF DYNAMIC RUPTURE MODEL

Rapid source characterization of the 2011 M w 9.0 off the Pacific coast of Tohoku Earthquake

Seismic Source Mechanism

Effect of an outer-rise earthquake on seismic cycle of large interplate earthquakes estimated from an instability model based on friction mechanics

Estimation of S-wave scattering coefficient in the mantle from envelope characteristics before and after the ScS arrival

Scaling of apparent stress from broadband radiated energy catalogue and seismic moment catalogue and its focal mechanism dependence

Chapter 2. Earthquake and Damage

The Size and Duration of the Sumatra-Andaman Earthquake from Far-Field Static Offsets

Short Note Source Mechanism and Rupture Directivity of the 18 May 2009 M W 4.6 Inglewood, California, Earthquake

Source processes of the 2009 Irian Jaya, Indonesia, earthquake doublet

A GLOBAL MODEL FOR AFTERSHOCK BEHAVIOUR

LETTER Earth Planets Space, 56, , 2004

SUPPLEMENTARY INFORMATION

Source characteristics of large deep earthquakes: Constraint on the faulting mechanism at great depths

TSUNAMI CHARACTERISTICS OF OUTER-RISE EARTHQUAKES ALONG THE PACIFIC COAST OF NICARAGUA - A CASE STUDY FOR THE 2016 NICARAGUA EVENT-

Tsunami waveform analyses of the 2006 underthrust and 2007 outer-rise Kurile earthquakes

overlie the seismogenic zone offshore Costa Rica, making the margin particularly well suited for combined land and ocean geophysical studies (Figure

Tohoku-oki event: Tectonic setting

Magnitude 7.1 NEAR THE EAST COAST OF HONSHU, JAPAN

Aftershocks are well aligned with the background stress field, contradicting the hypothesis of highly heterogeneous crustal stress

A search for seismic radiation from late slip for the December 26, 2004 Sumatra-Andaman (M w = 9.15) earthquake

Magnitude 7.9 SE of KODIAK, ALASKA

An intermediate deep earthquake rupturing on a dip-bending fault: Waveform analysis of the 2003 Miyagi-ken Oki earthquake

Magnitude 7.6 & 7.4 SOLOMON ISLANDS

Hitoshi Hirose (1), and Kazuro Hirahara (2) Abstract. Introduction

Rupture complexity of the M w 8.3 sea of okhotsk earthquake: Rapid triggering of complementary earthquakes?

X-2 HIKIMA AND KOKETSU: THE 2004 CHUETSU, JAPAN, EARTHQUAKE We relocated the hypocenters of the 2004 Chuetsu earthquake sequence, Niigata, Japan, usin

RELOCATION OF LARGE EARTHQUAKES ALONG THE SUMATRAN FAULT AND THEIR FAULT PLANES

Teleseismic waveform modelling of the 2008 Leonidio event

Case Study 2: 2014 Iquique Sequence

Materials and Methods The deformation within the process zone of a propagating fault can be modeled using an elastic approximation.

SOURCE PROCESS OF THE 2003 PUERTO PLATA EARTHQUAKE USING TELESEISMIC DATA AND STRONG GROUND MOTION SIMULATION

Rapid magnitude determination from peak amplitudes at local stations

Supplementary Materials for

Determination of fault planes in a complex aftershock sequence using two-dimensional slip inversion

FOCAL MECHANISM DETERMINATION OF LOCAL EARTHQUAKES IN MALAY PENINSULA

1.3 Short Review: Preliminary results and observations of the December 2004 Great Sumatra Earthquake Kenji Hirata

Effect of the Emperor seamounts on trans-oceanic propagation of the 2006 Kuril Island earthquake tsunami

Comment on A new estimate for present-day Cocos-Caribbean plate motion: Implications

Spatio-temporal variation in slip rate on the plate boundary off Sanriku, northeastern Japan, estimated from small repeating earthquakes

AVERAGE AND VARIATION OF FOCAL MECHANISM AROUND TOHOKU SUBDUCTION ZONE

Crustal deformation by the Southeast-off Kii Peninsula Earthquake

DETERMINATION OF SLIP DISTRIBUTION OF THE 28 MARCH 2005 NIAS EARTHQUAKE USING JOINT INVERSION OF TSUNAMI WAVEFORM AND GPS DATA

LETTER Earth Planets Space, 57, , 2005

FOCAL MECHANISMS OF SUBDUCTION ZONE EARTHQUAKES ALONG THE JAVA TRENCH: PRELIMINARY STUDY FOR THE PSHA FOR YOGYAKARTA REGION, INDONESIA

SOURCE INVERSION AND INUNDATION MODELING TECHNOLOGIES FOR TSUNAMI HAZARD ASSESSMENT, CASE STUDY: 2001 PERU TSUNAMI

Three Dimensional Simulations of Tsunami Generation and Propagation

Magnitude 7.5 NEW BRITAIN REGION, PAPUA NEW GUINEA

Crustal deformation in Taiwan: Results from finite source inversions of six M w > 5.8 Chi-Chi aftershocks

Lecture 20: Slow Slip Events and Stress Transfer. GEOS 655 Tectonic Geodesy Jeff Freymueller

RECIPE FOR PREDICTING STRONG GROUND MOTIONS FROM FUTURE LARGE INTRASLAB EARTHQUAKES

Static stress transfer from the May 20, 2012, M 6.1 Emilia-Romagna (northern Italy) earthquake using a co-seismic slip distribution model

FOCAL MECHANISM DETERMINATION USING WAVEFORM DATA FROM A BROADBAND STATION IN THE PHILIPPINES

JCR (2 ), JGR- (1 ) (4 ) 11, EPSL GRL BSSA

27th Seismic Research Review: Ground-Based Nuclear Explosion Monitoring Technologies

Earthquake Focal Mechanisms and Waveform Modeling

Lesvos June 12, 2017, Mw 6.3 event, a quick study of the source

Partial Breaking of a Mature Seismic Gap: The 1987 Earthquakes in New Britain

Slab pull, slab weakening, and their relation to deep intra-slab seismicity

Routine Estimation of Earthquake Source Complexity: the 18 October 1992 Colombian Earthquake

The Mw 6.2 Leonidio, southern Greece earthquake of January 6, 2008: Preliminary identification of the fault plane.

Tsunami Simulation of 2009 Dusky Sound Earthquake in New Zealand

Rotation of the Principal Stress Directions Due to Earthquake Faulting and Its Seismological Implications

Earthquake Source. Kazuki Koketsu. Special Session: Great East Japan (Tohoku) Earthquake. Earthquake Research Institute, University of Tokyo

Preliminary slip model of M9 Tohoku earthquake from strongmotion stations in Japan - an extreme application of ISOLA code.

Rupture Process of the Great 2004 Sumatra-Andaman Earthquake

Rupture process of the 2005 West Off Fukuoka Prefecture, Japan, earthquake

Transcription:

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 112,, doi:10.1029/2006jb004480, 2007 Triggering of earthquakes during the 2000 Papua New Guinea earthquake sequence Sun-Cheon Park 1 and Jim Mori 1 Received 3 May 2006; revised 20 September 2006; accepted 17 October 2006; published 8 March 2007. [1] A sequence of large earthquakes occurred in the New Britain/New Ireland region of Papua New Guinea in late 2000. The sequence started with a M w 6.8 earthquake along the New Britain Trench on 29 October. About 20 days later a M w 8.2 earthquake occurred in the New Ireland region on 16 November and produced large strike-slip surface displacements and tsunamis. Following the M w 8.2 event, two large earthquakes (M w 7.5) occurred along the nearby New Britain Trench on 16 and 17 November. Furthermore, small triggered events were observed over a wide area outside of the rupture zones of the large earthquakes, with different focal mechanisms from those of the major events. There is likely some mechanism(s) that triggered this remarkable sequence, and we document the details of the spatial and temporal patterns of the events. We investigated if static stress changes can explain the initiation of the large earthquakes and also some groups of triggered smaller events. There are mixed results. The occurrences of the two M 7.5 major events may be explained by the static stress changes; however, there are also some earthquakes that are not consistent with triggering by static stress changes. There may be multiple mechanisms, including static and dynamic stress changes, that are needed to explain the complicated sequence of earthquakes. Citation: Park, S.-C., and J. Mori (2007), Triggering of earthquakes during the 2000 Papua New Guinea earthquake sequence, J. Geophys. Res., 112,, doi:10.1029/2006jb004480. 1. Introduction [2] Papua New Guinea is one of the most seismically active regions in the world. Earthquakes, including many with magnitudes greater than 7, occur along the New Britain Trench, where the Solomon Sea Plate is subducting beneath the South Bismarck and Pacific plates (Figure 1). There is a triple junction south of New Ireland with the Bismarck Sea Seismic Lineation, which is a transform boundary between the South Bismarck Plate and the Pacific Plate (or North Bismarck Plate). In the studied area bounded by 150 E to 156 E and 2 S to 7 S, 17 earthquakes with magnitude greater than 7 occurred since 1970, including the remarkable sequence in late 2000 that included M w 8.2, M w 7.5, and M w 7.4 events. The latter two events are not aftershocks in the standard sense since they occurred well outside of the rupture area of the M w 8.2 earthquake in a different tectonic setting with different focal mechanisms. Stars in Figure 2 indicate the locations of the large earthquakes from the International Seismological Centre (ISC) and National Earthquake Information Center (NEIC) of the U.S. Geological Survey (USGS) catalogues. [3] This sequence is one of the clearest examples of the apparent triggering of very large earthquakes and we present details of the timing and spatial extents of the ruptures for 1 Disaster Prevention Research Institute, Kyoto University, Kyoto, Japan. Copyright 2007 by the American Geophysical Union. 0148-0227/07/2006JB004480$09.00 the major events so that the triggering mechanisms can be evaluated. Over the past decade, many studies have addressed the occurrence of aftershocks and triggered earthquakes and there has been significant progress in clarifying some of the explanations for the induced seismicity [Steacy et al., 2005]. Mechanisms have included static stress triggering [e.g., King et al., 1994; Toda et al., 1998; Díez et al., 2005], dynamic triggering [e.g., Gomberg, 2001; Gomberg et al., 2003], rate-and-state friction [e.g., Dieterich, 1994; Parsons, 2005], and the effects of fluid flow [e.g., Gavrilenko, 2005]. Among these proposed mechanisms and processes, static stress triggering is one common way to evaluate the influence of large earthquakes on the surrounding area. Therefore we investigate the effects of static stress changes from the large earthquakes and evaluate their effectiveness in explaining the triggering of following earthquakes in the sequence. Geist and Parsons [2005] also studied the static stress triggering from the M w 8.2 16 November event on the following M w 7.5 event. Using the fault models derived in this study, we extend this analysis to the 17 November event and other smaller earthquakes that are triggered over a wide region. Our waveform modeling of the larger events produces good constraints on the depths and faulting geometries. [4] A strong sequence of earthquakes occurred in Papua New Guinea in late 2000 (Figure 3). A M w 6.8 earthquake started the sequence along the New Britain Trench at 0837:08 UTC on 29 October 2000. Normal aftershocks followed this event and there were also earthquakes during the following 20 days in the New Ireland and New Britain 1of14

Figure 1. Tectonic setting of Papua New Guinea after Tregoning et al. [1998]. In the studied area, indicated by the dotted box, the Solomon Sea plate is subducting beneath the South Bismarck and Pacific plates. There is a triple junction with a transform fault, Bismarck Sea Seismic Lineation, south of New Ireland. region, which were few hundred kilometers from the 29 October event and near the locations of subsequent large earthquakes (Figure 3a). [5] At 0454:56 UTC on 16 November 2000, a great earthquake (M w 8.2) occurred in the New Ireland region. The epicenter was located in the St. George Channel between New Ireland and New Britain Islands. The aftershock distribution crosses southern New Ireland with locations that are coincident with the Weitin-Kamdaru fault (Figure 3b). There were observed strike-slip surface displacements of up to 5 m along the fault on the east side of New Ireland (I. Itikarai, personal communication, 2001). GPS observation also showed left-lateral movement up to 8 m along the Weitin-Kamdaru fault [Tregoning et al., 2005]. Tsunamis with heights up to 3 m were reported along the coasts of New Ireland, New Britain, and Bougainville (ISC catalogue). We name this earthquake the 16 November A event. [6] Another large earthquake (M w 7.5) occurred about 3 hours after the New Ireland event at 0742:16 UTC on 16 November along the New Britain Trench. We name this earthquake the 16 November B event. The following seismicity (Figure 3c) stretches about 100 km south of the rupture areas of these two major earthquakes. [7] A third large earthquake (M w 7.4) occurred along the New Britain Trench about 40 hours after the New Ireland event at 2101:56 UTC on 17 November. Figure 3d shows that following these major earthquakes, there are events in the Bismarck Sea west of New Ireland and continued seismicity south of the 16 November B event, which are outside of the immediate aftershock zones. These large earthquakes appear to strongly affect the stress field in this region and trigger many events over a wide area. Increased seismicity is noticeable here compared to the background seismicity (Figure 2) for 6 months before the sequence. Furthermore, the occurrences of the major earthquakes are probably related to the previously occurring seismicity. 2of14

Figure 2. Large earthquakes and background seismicity in studied area. Stars indicate large earthquakes (M > 7) since 1970: 1, 14 July 1971 shallow M7.9; 2, 26 July 1971 shallow M7.9; 3, 20 July 1975 shallow M7.9; 4, 18 March 1983 intermediate M7.9; 5, 10 May 1985 shallow M7.3; 6, 3 July 1985 shallow M7.4; 7, 30 December 1990 intermediate M7.5; 8, 28 September 1991 shallow M7.1; 9, 16 August 1995 shallow M7.8; 10, 16 August 1995 shallow M7.2; 11, 29 April 1996 shallow M7.5; 12, 10 May 1999 intermediate M7.1; 13, 16 May 1999 intermediate M7.1; 14, 16 November 2000 shallow M8.2; 15, 16 November 2000 shallow M7.8; 16, 17 November 2000 shallow M8.2; 17, 9 September 2005 intermediate M7.7. Circles and triangles indicate shallow seismicity from 1 May 2000 before the 29 October event for about 6 months. Earthquake data are from ISC and USGS catalogues. [8] Table 1 lists the source parameters for the studied events taken from the USGS catalogue. Hereafter event identifications shown in Table 1 will be used in the text. 2. Slip Distributions 2.1. Data and Methods [9] To determine the slip distributions for the three major earthquakes, we carried out waveform inversions using teleseismic P waveforms. We obtained digital P wave data for stations at distance ranges of about 30 and 90 from the Incorporated Research Institutions for Seismology (IRIS) data center. Original data were transformed to displacement using the station instrument files provided by IRIS, and high-passed filtered at 0.01 Hz. We used as many stations as possible (12 to 15) for the analyses; however, the azimuthal coverage is not optimum because there are far fewer station toward the south and east. Figure 4 shows the station distributions used for teleseismic waveform inversions for the three earthquakes. [10] To estimate the slip distributions, we used a multiple time window inversion [Hartzell and Heaton, 1983]. In this method, the fault is divided into smaller subfaults and Green s functions are calculated for the propagation response from each subfault to each station. We calculated Green s functions using a program by Langston and Helmberger [1975]. A least squares inversion was then used to determine the slip on each subfault by minimizing the fit between calculated waveforms and the observed data. Green s functions for slip both parallel and perpendicular to the strike of the fault are used, so that the slip direction of each subfault is determined. A smoothing matrix was added to the inversion to increase the stability of the solutions spatially and temporally. After testing various values, we found values of the smoothing that give a reasonable trade-off between the smoothness of the solution and a good spatial resolution. 2.2. The 16 November A Event [11] For setting up the inversion, the fault plane was divided into 20 4 subfaults. It has a length of 200 km and a downdip width of 42.6 km that reaches a depth of 40 km from the surface. The hypocenter is located at the northwestern end of the fault at a depth of 15 km. By trial and error we tried numerous models with a range of strike and dip angles for the fault plane, including multiple seg- 3of14

Figure 3. The 2000 earthquake sequence in Papua New Guinea. (a) Time period from the M w 6.8 event on 29 October 2000 up to the time immediately before the 16 November A event. (b) Time from the 16 November A event (M w 8.2) at 0454:56 UTC up to the time prior to the 16 November B event. (c) Time from the 16 November B event (M w 7.5) at 0742:16 UTC up to the time prior to the 17 November event. (d) Time from the 17 November event (M w 7.4) at 2101:56 UTC until 30 November. Open stars and solid rectangles indicate the epicenters and fault planes of the large earthquakes, respectively, occurred in the time interval. Gray solid stars and dotted rectangles indicate the previously occurring epicenters and fault planes. Circles and triangles are used to distinguish the hypocentral depth with gray for earthquakes occurred in the previous time interval. Sizes of the symbols are dependent on magnitude. Earthquake data are from the USGS catalogue. ments with different strikes, and found that a model with strike of 320 and dip of 70 provided the best fit to the data. Figure 5 shows the slip distribution (Figure 5a) and waveform fits (Figure 5b) for this event. The vectors show the direction and amount of displacement of the hanging wall at each subfault in Figure 5a. In Figure 5b, solid lines are observed waveforms and dotted lines are the synthetic seismograms. Synthetics in northeastern stations do not fit well the observed waveforms; however, they reproduce the overall amplitudes. The average value of rakes for all subfaults was 58.3 (Figure 5c). The source time function (Figure 5d) shows a long moment release with a total duration of 88 s. We tested various values of a fixed rupture velocity, which ranged from 2.0 to 3.0 km/s, and found the best solution to be 2.4 km/s. The largest slips are about 10 to 11.5 m. The total moment obtained is 2.2 10 28 dyn cm, which is equivalent to M w 8.2. Table 1. Source Parameters of Earthquakes Studied a Event Origin Time, UTC Latitude Longitude Depth, km Magnitude M w 29 Oct 0837:08 4.77 153.95 50 6.8 16 Nov A 0454:56 3.98 152.16 13 7.6 16 Nov B 0742:16 5.23 153.10 30 7.3 17 Nov 2101:56 5.49 151.78 37 7.4 a Data from USGS. 4of14

Figure 4. Station distribution used for the waveform inversions for the (a) 16 November A, (b) 16 November B, and (c) 17 November events. [12] This event shows a complicated slip distribution having both strike-slip and dip-slip components. Strike-slip motions are seen in the upper portion of the fault about 40 to 110 km southeast of the hypocenter, consistent with the mostly strike-slip surface fault displacements seen on New Ireland. Especially those of the uppermost subfaults around 100 km from the hypocenter correspond to the large (about 5 m) surface displacement that was observed along the Weitin-Kamdaru fault on the southeastern coast of New Ireland (I. Itikarai, personal communication, 2001). Farther to the southeast, the slips gradually rotate to dip slip. The large shallow dip-slip movements on the southeastern 5of14

Figure 5. Slip distribution of the 16 November A event. (a) Slip distribution. Star indicates the hypocenter, and vectors show the direction and amount of displacement. Slip motions gradually change from strike slip to dip slip for subfaults from the northwest to southeast. (b) Fit of synthetic (dotted) and observed (solid) waveforms. Scales are the same for all stations. (c) Focal mechanism; strike 320, dip 70, average rake 58. (d) Source time function (horizontal scale in s, vertical scale in dyn cm/s). portion may provide an explanation for the observed tsunamis on the southeast coast of New Ireland. 2.3. The 16 November B Event [13] For this event and the 17 November event, we assumed a fault plane that is consistent with shallow thrust events on the New Britain Trench. Almost all moderate and large earthquakes located near the subduction interface with this typical mechanism are observed to occur on the shallowly dipping plane. Also relocated aftershocks by Tregoning et al. [2005] showed that both the 16 November B and 17 November events were plate boundary thrust events on the New Britain subduction zone defined by 20 dips on the Wadati-Benioff zones. [14] The fault plane of this event was divided into 10 7 subfaults. Figure 6 shows the slip distribution (Figure 6a) and waveforms (Figure 6b). The length and width of the fault plane that we inverted for are 100 km and 95.8 km, respectively, and the fault reaches 36 km from 8 km depth. The hypocenter is located close to the center of the fault at a depth of 26 km. These fault dimensions are determined by trial and errors so that the area is large enough that there are 6of14

Figure 6. Slip distribution of the 16 November B event. (a) Slip distribution. Star indicates the hypocenter. (b) Fit of synthetic (dotted) and observed (solid) waveforms. Scales are the same for all stations. (c) Focal mechanism; strike 240, dip 17, average rake 70. (d) Source time function (horizontal scale in s, vertical scale in dyn cm/s). no large values of slip on the edge, yet small enough so that there are not large areas of zero slip which can give unstable solutions. The strike and dip angles were determined as 240 and 17 and the average value of rakes for all subfaults was 69.8 (Figure 6c). According to the source time function (Figure 6d), seismic moment was released mainly around the hypocenter for totally 35 s. We found the best solution to be for a rupture velocity of 2.7 km/s. [15] The slip distribution is smaller and simpler than for the 16 November A event. There is an area of large slip very close to the hypocenter and the largest slip is about 8.5 m. 7of14

Figure 7. Slip distribution of the 17 November event. (a) Slip distribution. Star indicates the hypocenter. (b) Fit of synthetic (dotted) and observed (solid) waveforms. Scales are the same for all stations. (c) Focal mechanism; strike 230, dip 25, average rake 57. (d) Source time function (horizontal scale in s, vertical scale in dyn cm/s). The total moment obtained is 2.5 10 27 dyn cm, which is equivalent to M w 7.5. 2.4. The 17 November Event [16] We divided the fault plane into 10 7 subfaults for the inversion. Figure 7 shows the slip distribution (Figure 7a) and waveforms (Figure 7b). The inverted fault plane has a length of 100 km and width of 82.8 km. It is located deeper than the 16 November B event and extends from a depth of 20 km to 55 km. The hypocenter is located southeast from the center of the fault and at a depth of 36 km. The strike and dip angles were determined as 230 8of14

Table 2. Focal Mechanisms and Magnitudes of the Three Major Earthquakes Obtained in This Study Comparing to Those From USGS MT and Harvard CMT (HRV) Solutions Source (Strike, Dip, Rake) a Depth, km M w 16 November A Event This study (320, 70, 58) 15 8.2 USGS (180, 14, 143), (306, 82, 79) 13 7.6 HRV (328, 43, 3), (236, 88, 133) 24 8.0 16 November B Event This study (240, 17, 70) 26 7.5 USGS (288, 24, 122), (73, 70, 76) 30 7.3 HRV (253, 15, 93), (70, 75, 89) 31 7.8 17 November Event This study (230, 25, 57) 36 7.4 USGS (281, 32, 94), (97, 58, 88) 37 7.4 HRV (230, 24, 64), (78, 68, 101) 17 7.8 a Rake angles of this study are average values. and 25, respectively, and the average value of rakes was 56.5 (Figure 7c). The source time function (Figure 7d) has mainly one pulse and its total duration is 49 s. The best solution was found to be for a rupture velocity of 2.5 km/s. [17] The slip distribution is fairly simple showing an area of large slip including the hypocenter. The largest slips are about 4 m. The total moment obtained is 1.4 10 27 dyn cm, which is equivalent to M w 7.4. 2.5. Comparison With Other Studies [18] We compare our results from the teleseismic waveform inversions with those from USGS moment tensor (MT) and Harvard centroid moment tensor (CMT) solutions [Dziewonski et al., 1981] in Table 2. Focal mechanisms obtained in this study are more similar to those of the Harvard CMT than the USGS MT solutions, although our moment magnitudes (M w ) are smaller than those of the Harvard solutions except for the 16 November A event. [19] The Earthquake Research Institute, University of Tokyo, also provided body waveform inversions of the 16 November A event (EIC note 94, 2000). Their results showed that the earthquake is left-lateral strike-slip event with focal mechanism of (strike, dip, rake) = (143, 85, 2) and moment magnitude (M w ) is 8.0. Their slip distribution has the largest slip at about 120 km to the southeast from the hypocenter. This slip distribution is somewhat different from our result which has the largest slip at about 80 km far, probably due to the difference of preferred rupture velocity. They also cited the possibility of a dip-slip subevent about 2 min after the initial rupture, which may be consistent with the dip-slip motion at southeast portion of our slip distribution. [20] Rham and Das [2003] relocated aftershocks of the 16 November A earthquake, including the 16 November B and 17 November events. They noted that the 16 November A event is a shallow left-lateral strike-slip event with a fault plane having strike 320 and vertical dip, and the rupture propagated unilaterally from northwest to southeast for about 120 km. This is generally consistent with our results if we consider only the strike-slip portions of our model. They also reported that most of the larger aftershocks (M w > 6) including the 16 November B and 17 November events have thrust mechanisms, which are consistent with our values. 3. Static Stress Changes [21] To evaluate the effects of static stress triggering, we calculated the Coulomb failure stress changes (Ds f ) that resulted from the larger earthquakes (the 29 October and 16 November A and B) in the region of the following earthquakes. This method for calculating a static failure potential has been used recently for many earthquakes [e.g., King et al., 1994; Stein, 1999]: Ds f ¼ Dt þ m 0 Ds n where Dt is the change in shear stress, Ds n is the change in normal stress, and m 0 is the value of the apparent coefficient of friction. Using the slip distributions obtained in section 2, we can calculate the shear and normal stress changes for all locations in the surrounding area. To calculate these stresses, we used the program written by Okada [1992] and calculated the Coulomb failure stress changes using equation (1). Positive values of the Coulomb failure stress changes (Ds f ) mean that static stress increases and brings the specified fault closer to failure. On the other hand, the negative values mean that static stress decreases and tends to inhibit failure on the fault. [22] We calculated the Coulomb failure stress changes at the hypocenter of the triggered event because once a rupture begins, the dynamic stress changes associated with the rupture front are much larger than the static stress changes. In this point of view, triggering of both large and small earthquakes may be treated similarly, that is, we may not need to distinguish different effects of static stress triggering depending on earthquake size. The mechanism of rupture growth to a large earthquake after initiation is another problem not addressed in this paper. [23] We first calculated the effects of the static stress changes from the large earthquakes on the following large earthquakes, starting with the M w 6.8 29 October event and continuing for the 16 November A, 16 November B, and 17 November events. We also estimated the Coulomb failure stress changes from the 16 November A and B events in the region of the smaller triggered earthquakes. Triggered earthquakes are identified as seismicity outside of the immediate rupture areas of the major events having different focal mechanisms. Figure 8 shows the focal mechanisms of the major earthquakes, their aftershocks and three groups of triggered events that occurred between 16 and 30 November. Solid rectangles are the horizontal projections of the fault planes of the three major earthquakes obtained in section 3. For aftershocks and triggered events, we used the hypocenters determined by USGS and focal mechanisms both of the USGS and Harvard solutions. 3.1. Triggering of Major Earthquakes [24] For the fault geometry of the 29 October event (M w 6.8), we used (strike, dip, rake) = (322, 63, 90) obtained from the Harvard CMT solutions and focal depth of 50 km from the USGS catalogue. Then we calculated the regional static stress changes caused by this earthquake. The static stress perturbation patterns showed stress ð1þ 9of14

Figure 8. Focal mechanisms of the large studied earthquakes, aftershocks, and triggered events occurred until 30 November. Stars indicate the epicenters of the large earthquakes, and rectangles are horizontal projection of fault planes of the three major earthquakes. Focal mechanisms of the three major earthquakes are obtained in this study, and those of the 29 October event, aftershocks and triggered earthquakes are from Harvard CMT solutions. Groups 1 to 3, which are assumed to be triggered earthquakes, are identified by focal mechanisms, which are different from those of the three major earthquakes. Earthquake parameters of events 1 to 6 are given in Table 4. increases around the hypocenters of the 16 November A and 17 November events and decreases around the 16 November B event, but all the changes were very small (order of 10 4 MPa). For a plate velocity of 15 cm/yr [Tregoning et al., 1998], the accumulating stress is about 0.3 MPa/yr, so that the static stress changes are equivalent to the tectonic stress that accumulates in just a few hours. Therefore static stress changes from the 29 October event can have little effect in the regions of the following large earthquakes. [25] We next estimated the Coulomb failure stress changes from the larger events in the sequence. Figure 9 shows the effects from the M w 8.2 16 November A event for the region of the 16 November B event. The Coulomb failure stress changes are calculated for a depth of 26 km and m 0 = 0.6. We tested various values for the apparent coefficient of friction from 0.1 to 0.7. The stress changes patterns did not show significant differences from that for m 0 = 0.6, and the values of the stress increases around the hypocenter were nearly the same. Note that the hypocenter and almost all of the rupture area of the 16 November B event are located within a positive area of Figure 9. This result indicates that the event occurred in an area of static stress increases, promoting to faulting. The value of the stress increases around the hypocenter is about 0.29 MPa. We also calculated the static stress changes at different depths from 10 to 34 km for average rake angles and found that most of the slip area falls within a region with fairly large stress changes (more than about 0.1 MPa). [26] Figure 10 shows the combined Coulomb failure stress changes from the 16 November A and B events, estimated for the fault plane orientation of the 17 November event. These calculations are done for a depth of 36 km and m 0 = 0.6. The summed values of the Coulomb failure stress changes around the hypocenter of the 17 November event are about 0.03 MPa. There is an increase of the static stress around the hypocenter produced mainly by the 16 November A event and the effect of the 16 November B event, which is negative, is relatively small. Table 3 summarizes the results of the static stress changes for the major earthquakes. 10 of 14

Figure 9. Static stress changes from the 16 November A event for the fault plane of the 16 November B event. Star indicates the epicenter of the 16 November B event, and the rectangle is the fault plane of the 16 November A event. Static stress changes show strong increase around the hypocenter of the 16 November B event. 3.2. Triggering of Smaller Earthquakes [27] If static stress changes are effective in triggering the large earthquakes, they should also trigger smaller earthquakes. Using hypocentral data from the USGS, we identified three groups of triggered events that occurred outside of the regions of the fault planes of the three major earthquakes from 16 November through 30 November, as shown in Figure 8. Group 1 had 9 events (compared to 7 events that occurred in the same region over the past year) with oblique strike-slip mechanisms that were located along the Bismarck Sea Seismic Lineation (Figure 1). Group 2 had 70 events (4 events over the past year) with normal fault mechanisms that are outer rise events along the New Britain Trench. Group 3 had 37 events (2 events over the past year) Figure 10. Combined static stress changes from the 16 November A and B events for the fault plane of the 17 November event. Star indicates the epicenter of the 17 November event, and the rectangles are the fault planes of the 16 November A and B events. Static stress changes show increases around the hypocenter of the 17 November event. 11 of 14

Table 3. Summary of Static Stress Change Estimates for the Major Earthquakes a Receiver Plane Source 29 Oct 16 Nov A 16 Nov B 16 Nov A and B Dominant Event 16 Nov A + 16 Nov B + 17 Nov + + + 16 Nov A a A plus indicates that the Ds f was conducive to faulting while a minus indicates that Ds f was inconsistent with the faulting. See the text for the details. that occur within the New Britain subduction zone with strike-slip mechanisms. We calculated the Coulomb failure stress changes from the 16 November A and B events for the appropriate focal mechanism in the region of each group, and the results with source parameters are listed in Table 4. The static stress changes from the 16 November B event are relatively small and do not affect the results for the groups 1 and 2. We tested for both nodal planes of the USGS and Harvard focal mechanisms. [28] As shown in Table 4, earthquakes in the Bismarck Sea in group 1 have positive values of the Coulomb failure stress changes from both of the 16 November A and B events. Outer rise earthquakes in group 2, which are normal fault events, experienced static stress decreases from the 16 November A and B events, though static stress changes from the 16 November B event can be either increases or decreases depending on the details of the assumed mechanism. For the strike-slip events in group 3, a static stress decrease was observed. However, the hypocenter is located at almost the boundary between static stress increase Table 4. Results of Static Stress Change Estimates for Smaller Earthquakes a Date Origin Time, UTC Location (Latitude, Longitude) Depth, Focal km M w Mechanism Group 1, Event 1 21 Nov 2000 2003:48 3.617, 150.823 20 5.8 (26, 77, 177) +0.0028 A (117, 87, 13) +0.0024 A (4, 58, 163) +0.0137 A (103,76, 33) +0.0139 A (26, 77, 177) +0.0002 B (117, 87, 13) +0.0002 B (4, 58, 163) +0.0006 B (103, 76, 33) +0.0006 B Group 1, Event 2 23 Nov 2000 0656:05 3.720, 151.580 15 5.2 (2, 42, 166) +0.0490 A (102, 81, 49) +0.0495 A (2, 42, 166) +0.0017 B (102, 81, 49) +0.0017 B Group 2, Event 3 17 Nov 2000 0422:56 6.268, 153.366 15 5.7 (139, 38, 71) 0.0053 A (296,54, 104) 0.0057 A (130,44, 106) 0.0176 A (333,48, 75) 0.0182 A (139,38, 71) 0.0032 B (296,54, 104) 0.0029 B (130,44, 106) +0.0099 B (333,48, 75) +0.0101 B Group 2, Event 4 22 Nov 2000 0627:14 6.378, 153.340 16 5.5 (183, 53, 73) 0.0173 A (337, 41, 111) 0.0174 A (137, 48, 89) 0.0125 A (315, 42, 91) 0.0123 A (183, 53, 73) 0.0006 B (337, 41, 111) 0.0006 B (137, 48, 89) +0.0040 B (315, 42, 91) +0.0041 B Group 2, Event 5 26 Nov 2000 0523:12 6.44, 153.42 32.1 5.4 (134, 45, 85) 0.0145 A (307, 45, 95) 0.0147 A (134, 45, 85) +0.0065 B (307, 45, 95) +0.0065 B Group 3, Event 6 26 Nov 2000 0214:08 5.69, 153.63 51 5.2 (171, 66, 2) +0.023 +0.034 A (262, 88, 156) +0.023 +0.034 A (171, 66, 2) 0.040 0.036 B (262, 88, 156) 0.039 0.035 B a Origin time, location, depth, and M w are from USGS catalogue, and those from Harvard solutions are used when they were not determined by USGS. Focal mechanisms tested are two failure planes from USGS and Harvard solutions, but only Harvard ones are given when they were unavailable in USGS solutions. CFF denotes calculated Coulomb failure stress changes. A and B in source event column denotes the 16 November A and B events, respectively. CFF, MPa Source Event 12 of 14

and decrease. Therefore the result will strongly depend on the details of the hypocentral determinations and focal mechanisms. 3.3. Summary of Static Stress Changes [29] We summarize the results of our analyses of static stress changes and triggered earthquakes as follows: (1) Effect of the M w 6.8 29 October event in the region of the following major earthquakes was too small to be significant, (2) effect of the M w 8.2 16 November A event had a positive stress change of about 0.29 MPa in the region of the M w 7.5 16 November B event (similar to conclusion of Geist and Parsons [2005]), (3) combined effects of the 16 November A and B events had a positive stress change of about 0.03 MPa in the region of the M w 7.4 17 November event, (4) combined effects of the 16 November A and B events had a positive stress change of about 0.002 to 0.05 MPa for the strike-slip events in the Bismarck Sea group, (5) combined effects of the 16 November A and B events have a negative stress change of about 0.008 MPa for the outer rise normal-faulting events, and (6) combined effects of the 16 November A and B events have a mixed stress change, depending on the faulting geometries for the strike-slip events in subduction zone. 4. Discussion [30] For the five cases of triggered events summarized above, where the static stress changes are large enough to potentially be the triggering mechanism, three examples have positive results. With this mixed result and small sample, it is difficult to make a definitive statement about the validity of static stress triggering. Taking into account the fact that for this sequence the areas of the positive triggering are larger than the areas of negative triggering (Figures 9 and 10), we can say that the hypothesis of static stress triggering does not make a large improvement over random chance for explaining the locations of triggered earthquakes. [31] It seems unlikely that this sequence is a coincidence of large events, and we think there is some triggering mechanism(s) that relates the M w 8.2, M w 7.5, and M w 7.4 earthquakes, along with the smaller clusters that occurred in the wide surrounding area. If static stress triggering does not provide a good general explanation for all of the triggered events, what is the mechanism? We do not have any strong evidence for other types of mechanisms, but the possibility of dynamic triggering [e.g., Gomberg et al., 2003] from the seismic waves has been proposed as a mechanism that might be able to explain all the triggered events. Furthermore, the study of Pollitz and Johnston [2006] suggests that even a large percentage of ordinary aftershocks may be due to dynamic triggering. [32] It is intriguing that there was also a M6 event on 7 November, near the epicenter (within 25 km) of the subsequent M w 7.4 17 November event (Figure 3). One might speculate that the region of the 17 November event was close to failure so that it could have been triggered by the M6 event. There have been some similar examples in regions of triggered seismicity where smaller foreshocks led to a larger triggered earthquake within hours to days. These include the M6.5 Big Bear event following the M7.4 Landers earthquake [King et al., 1994], the M5.5 event following the M6.6 2000 western Tottori, Japan, earthquake [Ohmi et al., 2002], the M7.4 event following the M7.1 2004 Kii Peninsula, Japan, earthquake [Park and Mori, 2005]. [33] The triggered normal faulting earthquakes in group 2 are tensional earthquakes near the outer rise. As described by Christensen and Ruff [1988], the tensional stresses perpendicular to the arc in this region could be increased by large slip downdip on the subduction interface. Another example of such triggering of tensional earthquakes occurred after the 1990 M w 7.0 event in Costa Rica [Protti et al., 1995]. [34] Lay and Kanamori [1980] point out that the occurrences of large earthquakes in this region are often characterized by doublets, so that triggering of large earthquakes has often observed in the past. The example involving the largest earthquakes is for the two M8 events on 14 July and 26 July 1971 (Figure 2) on adjacent segments of the New Britain Trench. Using slip distributions calculated by Park and Mori [2007], we estimated the static stress changes from the first event in the region of the hypocenter of the second event. We found that the 26 July event occurred in a positive area with static stress changes of about 0.03 to 0.06 MPa, with the location very close to the boundary between stress increase and decrease. So the evaluations of the effectiveness of static stress changes depend on the details of the fault and slip geometry. [35] Our evaluation of static stress changes as a mechanism for the triggering of earthquakes in the New Britain/ New Ireland region has mixed results. The complicated sequences of interrelated earthquakes cannot all be explained by static stress changes, although the occurrences of some large and small earthquakes are consistent with changes in the static stress. At this stage it is difficult to judge if static stress changes are part of the explanation for the triggered earthquakes, or if the dominant mechanism is completely different. 5. Conclusions [36] We used teleseismic P wave data to obtain slip distributions for the M w 8.2 New Ireland earthquake in 2000 and the two following large earthquakes (M w 7.5 on 16 November and M w 7.4 on 17 November). The slip distribution for the New Ireland event showed large amounts of strike-slip and dip-slip displacements along a rupture length of about 200 km. The slip motions changed gradually from strike slip in the northwest to dip slip in the southeast. The two large following events to the south along the New Britain Trench had smaller and simpler slip distributions. For both events, the slip mainly occurred in a region close to the hypocenter. [37] We calculated the Coulomb failure stress changes to evaluate the effects of the static stress triggering on the initiation of the large earthquakes and groups of smaller triggered events. We found that the occurrences of the two M 7.5 earthquakes were consistent with the static changes; however, also examining other smaller earthquakes, static stress triggering does not provide a general explanation for all of the triggered events in the sequence. It is likely that dynamic stress triggering or dynamic and static stress triggering play an important role in explaining the 13 of 14

interrelated occurrences of the large and small earthquakes of this sequence. [38] Acknowledgments. We thank Marino Protti, the Associate Editor, and an anonymous reviewer for their critical comments. We used digital waveform data from the Incorporated Research Institutions for Seismology data center. Some of figures in this paper were made using the General Mapping Tool (GMT) software [Wessel and Smith, 1998]. References Christensen, D. H., and L. J. Ruff (1988), Seismic coupling and outer rise earthquakes, J. Geophys. Res., 93, 13,421 13,444. Dieterich, J. (1994), A constitutive law for rate of earthquake production and its application to earthquake clustering, J. Geophys. Res., 99, 2601 2618. Díez, M., P. C. La Femina, C. B. Connor, W. Strauch, and V. Tenorio (2005), Evidence for static stress changes triggering the 1999 eruption of Cerro Negro Volcano, Nicaragua and regional aftershock sequences, Geophys. Res. Lett., 32, L04309, doi:10.1029/2004gl021788. Dziewonski, A. M., T.-A. Chou, and J. H. Woodhouse (1981), Determination of earthquake source parameters from waveform data for studies of global and regional seismicity, J. Geophys. Res., 86, 2825 2852. Gavrilenko, P. (2005), Hydromechanical coupling in response to earthquakes: On the possible consequences for aftershocks, Geophys. J. Int., 161, 113 129. Geist, E. L., and T. Parsons (2005), Triggering of tsunamigenic aftershocks from large strike-slip earthquakes: Analysis of the November 2000 New Ireland earthquake sequence, Geochem. Geophys. Geosyst., 6, Q10005, doi:10.1029/2005gc000935. Gomberg, J. (2001), The failure of earthquake failure models, J. Geophys. Res., 106, 16,253 16,263. Gomberg, J., P. Bodin, and P. A. Reasenberg (2003), Observing earthquakes triggering in the near field by dynamic deformations, Bull. Seismol. Soc. Am., 93, 118 138. Hartzell, S. H., and T. H. Heaton (1983), Inversion of strong ground motion and teleseismic waveform data for the fault rupture history of the 1979 Imperial Valley, California earthquake, Bull. Seismol. Soc. Am., 73, 1553 1583. King, G. C. P., R. S. Stein, and J. Lin (1994), Static stress changes and the triggering of earthquakes, Bull. Seismol. Soc. Am., 84, 935 953. Langston, C. A., and D. V. Helmberger (1975), A procedure for modeling shallow dislocation sources, Geophys. J. R. Astron. Soc., 42, 117 130. Lay, T., and H. Kanamori (1980), Earthquake doublets in the Solomon Islands, Phys. Earth Planet. Inter., 21, 283 304. Ohmi, S., K. Watanabe, T. Shibutani, N. Hirano, and S. Nakao (2002), The 2000 western Tottori earthquake Seismic activity revealed by the regional seismic networks, Earth Planets Space, 54, 819 830. Okada, Y. (1992), Internal deformation due to shear and tensile faults in a half-space, Bull. Seismol. Soc. Am., 82, 1018 1040. Park, S.-C., and J. Mori (2005), The 2004 sequence of triggered earthquakes off the Kii peninsula, Japan, Earth Planets Space, 57, 315 320. Park, S.-C., and J. Mori (2007), Are asperity patterns persistent? Implication from large earthquakes in Papua New Guinea, J. Geophys. Res., 112, B03303, doi:10.1029/2006jb004481. Parsons, T. (2005), A hypothesis for delayed dynamic earthquake triggering, Geophys. Res. Lett., 32, L04302, doi:10.1029/2004gl021811. Pollitz, F. F., and M. J. S. Johnston (2006), Direct test of static stress versus dynamic stress triggering of aftershocks, Geophys. Res. Lett., 33, L15318, doi:10.1029/2006gl026764. Protti, M., et al. (1995), The March 25, 1990 (M w = 7.0, M L = 6.8), earthquake at the entrance of the Nicoya Gulf, Costa Rica: Its prior activity, foreshocks, aftershocks, and triggered seismicity, J. Geophys. Res., 100, 20,345 20,358. Rham, D. J., and S. Das (2003), Triggering of thrust aftershocks on the New Britain Trench by the November 16, 2000 M w 8.0 New Ireland strike-slip earthquake, paper presented at the General Assembly, Int. Union of Geol. and Geophys., Sapporo, Japan. Steacy, S., J. Gomberg, and M. Cocco (2005), Introduction to special section: Stress transfer, earthquake triggering, and time-dependent seismic hazard, J. Geophys. Res., 110, B05S01, doi:10.1029/2005jb003692. Stein, R. S. (1999), The role of stress transfer in earthquake occurrence, Nature, 402, 605 609. Toda, S., R. S. Stein, P. A. Reasenberg, J. H. Dieterich, and A. Yoshida (1998), Stress transferred by the 1995 M w = 6.9 Kobe, Japan, shock: Effect on aftershocks and future earthquake probabilities, J. Geophys. Res., 103, 24,543 24,565. Tregoning, P., et al. (1998), Estimation of current plate motions in Papua New Guinea from Global Positioning System observations, J. Geophys. Res., 103, 12,181 12,203. Tregoning, P., M. Sambridge, H. McQueen, S. Toulmin, and T. Nicholson (2005), Tectonic interpretation of aftershock relocations in eastern Papua New Guinea using teleseismic data and the arrival pattern method, Geophys. J. Int., 160, 1103 1111. Wessel, P., and W. H. F. Smith (1998), New, improved version of the Generic Mapping Tools released, Eos Trans. AGU, 79(47), 579. J. Mori and S.-C. Park, Disaster Prevention Research Institute, Kyoto University, Gokasho, Uji, Kyoto, 611-0011, Japan. (suncheon@eqh.dpri. kyoto-u.ac.jp) 14 of 14