U53A-01 INVITED
A 700-year-long paleoseismic context for the Sumatran megathrust earthquakes of 2007
The Mw8.4 Sumatran earthquake of 2007 resulted from a 275-km rupture within the 700-km-long Mentawai section of the Sunda megathrust. Paleoseismic and paleogeodetic patterns of the past 700 years show that it was the largest rupture within this section in nearly the last two centuries and imply that it is the beginning of a series of gap-filling ruptures. We have characterized previous great ruptures using the growth patterns of coral microatolls on the Mentawai island reefs above the megathrust. These natural instruments have allowed us to map ancient coseismic vertical deformations in enough detail to make useful comparisons to the recent event. The last two giant ruptures occurred in 1797 (Mw 8.5–8.7) and in 1833 (Mw 8.6–8.9). Rupture in 1797 produced uplift above a 350-km long section of the megathrust from 0.5 to 3.2˚S. Rupture in 1833 produced uplift for about 275 km, from 2.2 to about 4.2˚S. Uplift in 2007 extended from 2.5 to 4.5˚S, nearly coincident with that of 1833. In all three events, the islands tilted toward the mainland, away from the trench. In the South Pagai Island (from 2.8˚ to 3.2˚S) in both the 1797 and 2007 events, uplift ranged as high as 80 cm. Uplift in 1833 was about 4 times greater there. The close timing of the 1797 and 1833 ruptures and their partial overlap indicates that a single great megathrust rupture need not relieve all the strain that has accumulated across the megathrust. Comparison of strain accumulated in the past half-century with strain relieved during the 2007 event shows that, indeed, the 2007 break involved only the southern half of the Mentawai locked patch and that even along that section, strain relief was only partial. Coral microatolls also reveal two earlier rupture sequences, in the mid- to late 1300s and in the late 1500s to early 1600s. Since the 1300s, intervals between great-earthquake sequences have been about 200 years. Thus, it was not a surprise that the first in what is likely to be a sequence of great earthquakes occurred this year. We fear that the later great earthquake of the recent sequence is likely to be greater in magnitude than that of the first, as much more strain is still unreleased. Although some emergency preparations have been undertaken in communities at risk from the next great rupture of the Mentawai patch, it is not at all clear that enough will have been done to secure the lives, livelihoods and well-being of most of the million or so people who live along that part of the Sumatran coast.
U53A-02 INVITED
THE 2007 BENGKULU EARTHQUAKE SERIES IN THE CONTEXT OF MEGA-RUPTURES OFF SOUTH SUMATRA
We present a seismological and tectonic study of the 2007 Bengkulu series, including an evaluation of the source characteristics of the main three events based on quasi-real time estimators, including the Energy-to-Moment parameter THETA, which suggests a trend towards slowness for the main shock at 11:10 UTC. While the earthquakes took place within the fault area of the great 1833 earthquake, in the probable context of Coulomb stress transfer from the 2004 and 2005 events, the mainshock and its immediate aftershocks define a faulting area significantly smaller than that of the 1833 event, and a preliminary review of plate kinematic models also suggests that the 2007 shock did not release the full strain accumulated since 1833. This is also supported by the seismic moment of the Bengkulu event (5E28 dyn*cm), and by preliminary modeling of the moderate near field tsunami (see Borrero et al., this session). In this context, we address the question of how much slip remains to be released along the 1833 rupture zone and to the South of it, and more generally of the ability for a large, but not gigantic event, to act as a barrier to the propagation of rupture during a future mega-earthquake. Namely, do the 2007 events preclude the future occurrence of a mega-earthquake whose rupture would over-ride the Bengkulu fault area? Historical records in Central and South Peru, as well as the over-riding by the 2004 Sumatra rupture of the 1881 Car Nicobar fault zone, would suggest a "cautiously negative" answer, casting a pessimistic note on far-field tsunami risk for the Southwestern Indian Ocean. ~
U53A-03 INVITED
Lessons from 2007 Mentawai Islands Earthquake
The 12/09/07 Mentawai Islands, M8.4 earthquake may, by its waywardness, contribute greatly both to our understanding of the occurrence of great subduction zone events and to their tsunamigenic potential. The earthquake nucleated off-shore Bengkulu in a region which likely experienced up to 18m of slip in the 1833 earthquake; at current rates of convergence only some 10m of this slip has been replaced and assessments of plate coupling in the epicentral region would suggest some of this has been relaxed aseismically. The megathrust north of Enggano Island has been remarkably quite during the instrumental period. Studies based on accelerating moment release have assigned low probabilities of occurrence for a great earthquake on this segment. Other studies employing interaction stresses as well as paleogeodesy and contemporary convergence estimates indicated that the most likely hypocenter for a Mentawai event was under Siberut Island; the hypocenter of the September event experienced no interaction stress from the 2004 or 2005 earthquakes. Preliminary slip inversions indicate that the maximum slip on the event was just 4 meters and, while the aftershock distribution would suggest that the rupture occupied much of the 1833 patch, it appears that much of the area experienced such small slip that the possibility of re-rupture in a future great earthquake cannot be discounted. Such a possibility has the potential to advance our understanding of the recurrence behaviour of subduction zone earthquakes. Finally, the earthquake generated a tsunami whose height and geographic distribution might contribute to our understanding of the relationship between earthquake slip and tsunami energy.
U53A-04
The 2007 Sumatra seismic sequence revealed by a regional seismic network in Indonesia
On September 12, 2007, a great earthquake with Mw 8.4 occurred at 11:10 (UTC) off Bengkulu, Sumatra, Indonesia. This event was followed by a large earthquake with Mw 8.0, which occurred at 23:49 (UTC) on the same day in the northwest of the Mw 8.4 earthquake. Another earthquake with Mw 6.8 occurred off Padang, in the northwest of the second earthquake, at 03:35 (UTC) on the next day. These earthquakes caused dozens of casualties and damage to the buildings in Bengkulu and Padang areas. Tsunami with a hight of 1m was observed in Padang, but no severe damage by tsunami was reported. We have been developing an automated system for rapid source parameter determinations of earthquakes in Indonesia, using data obtained from a broadband seismic network in this country (JISNET). This network is operated by NIED and Indonesia Meteorological and Geophysical Agency (BMG). In our method, the moment function is estimated simultaneously with a centroid moment tensor (CMT) solution based on the waveform inversion carried out in the frequency domain. The source parameters for an earthquake greater than Mw 5 can be determined automatically by our system within 15 minutes after the occurrence of an earthquake. The automatic inversion result is then checked manually. The source parameters of earthquakes that are not determined by the automatic system are also determined manually. The source parameters of the 2007 Sumatra earthquakes were succesfully determined by our system. The estimated focal mechanisms show similar thrust-type faultings, suggesting that these earthquakes occurred in association with the subduction of the Indo-Australian Plate beneath Sumatra Island. Both the Mw 8.4 and 8.0 events occurred at a depth of about 20 km. The rupture durations estimated from the moment functions are 140 s and 108 s for the Mw 8.4 and Mw 8.0 events, respectively. These rupture durations are slightly longer than typical values for earthquakes of these magnitudes. The aftershocks are distributed in a region extending about 300 km along the subduction zone of the Indo-Australian Plate. The depths of aftershocks range from 15 to 60 km. In the off-Bengkulu region, the earthquakes with M8 or above occurred in 1381, 1608 and 1833, approximately every 230 years. Now 174 years have been passed since the last event in 1833. We did not experience M8 class earthquakes during the last 210 years in the off-Padang region, which is to the north of off-Bengkulu. The off-Padang region corresponds to a seismic gap between the source regions of the 2007 Sumatra seismic sequence and M8.7 Nias earthquake on March 2005. The sequence of the M8.4, M8.0, and M6.8 earthquakes moving towards the northwest from off-Bengkulu in the 2007 seismic activity suggests the imminency of a large earthquake off Padang. We have been deploying broadband seismograph networks in Indonesia, including JISNET, by an international cooperation among Indonesia, Germany, China, and Japan, aiming at improving the capabilities to monitor seismic activity and tsunami generation in Indonesia. The seismic networks are now in operation, and totally 150 seismic stations will be installed by the end of 2008. Seismic monitoring based on these regional networks would contribute to early notification of a large earthquake anticipated to occur in the off-Padang region.
U53A-05
Near-field strong ground-motions during the September 12-13, 2007 Sumatran earthquakes
Three K2 triaxial accelerometers were operating at 200 samples per second within the region of strong shaking during the recent Mw 8.4, Mw 7.9 and Mw 7.0 Sumatran earthquakes and the many smaller events that followed. Presently, records from two stations, Pulau Sikuai (PSKI) and Silabu (SLBU) have been processed. Station PSKI is located 392 and 165 km from the epicenters of the two larger earthquakes, near Padang. Station Silabu (SLBU) is very near the northwestern limit of rupture during the Mw 8.4 and in the direction of rupture propagation. SLBU is up-dip and ~50 km distant from the hypocenter of the Mw 7.9 earthquake, providing a near-field constraint on indications from teleseismic data that this event nucleated sharply. Despite this, peak horizontal acceleration was approx. 300 cm/s/s and peak horizontal velocity was approx. 70 cm/sec. The SLBU station includes a GPS receiver that is part of the Sumatran GPS Array (SuGAr). The GPS instrument recorded the two larger earthquakes at a 1 sample per second data rate. Data from both the accelerometer and GPS instrument are expected to enable a useful comparison of the records from seismic inertial sensors (accelerometers) and GPS recordings of strong ground motions. A third station called Malakone (MLKN) on Enganno island lies about 100 km southeast of the epicenter of the Mw 8.4 and in the back-azimuth direction, away from propagation of that rupture. Because these three stations are located along-strike of the several large to great events in this sequence, and in one case even immediately up-dip, these recordings may be expected to provide data that are useful for refining attenuation relations that are used in engineering and hazard mapping. Strong-motion data are available through the National Center for Engineering Strong Motion Data (http://strongmotioncenter.org).
U53A-06
Analysis of High Rate GPS Data From the September 12 and 13 Strong Motion Events off Sumatra
The Sumatran Global Positioning System Array (SuGAr) consists of currently 29 continuous GPS stations that cover the forearc region of Sumatra. Four stations in the southern part of the array recorded high rate (1 sec. sampling rate) data during and surrounding the 3 major regional earthquakes (M8.4, M7.9, and M7.0) on September 12 and 13. Using the GAMIT/TRACK software, we compute 3-component time series of station displacements to investigate coseismic and postseismic kinematics. We present spectral characteristics of the recorded coseismic wave forms and compare them to model predictions. We analyze the observed early postseismic displacement in the context of frictional sliding models
U53A-07
Rupture kinematics of Mw8.4 South Pagai Earthquake, Sumatra, from joint inversion of seismic and geodetic data
On September 12, 2007, the Mw8.4 South Pagai earthquakes ruptured the subduction interface offshore southern Sumatra. The earthquake occurred within the area monitored by the Sumatra Geodetic Array (SuGAr). The cGPS stations on South Pagai Island show about 1.5 m of southwestward displacement, up to 0.6 m of uplift of the western coast of the island, and 0.1 m of subsidence of the eastern coast. Coastal stations near Bengkulu show about 70 cm southwest motion. We have obtained a finite-fault source model of the earthquake from the inversion of co-seismic displacements measured at 13 near-field cGPS stations from the SUGAR network together with the teleseismic waveforms measured at 35 stations. In addition, the model geometry was checked and adjusted by comparing observed and predicted long period surface waves and normal modes. The source is a northward unilateral rupture with two main asperities, a deeper one, extending relatively deep beneath Bengkulu coastal area and a shallower one beneath South Pagai. Co-seismic slip reached a maximum of about 5.5 meters under South Pagai. The earthquake was not very impulsive with rise times of the order 10 seconds and a rupture velocity of about 2 km/s. The rupture initiated at the southeastern edge of a patch of the subduction interface that had been shown to be strongly locked from geodetic and paleogeodetic interseismic measurements. The rupture propagated unilaterally to the north rupturing only a fraction of a strongly coupled fault patch. The previous large earthquake on that portion of the megathrust was a M8.6 event in 1833 which produced much larger co-seismic uplift. The 2007 event released only a small fraction of the deficit of moment that has accumulated since then due to interseismic locking of the subduction interface.