Originally published as:
Sørensen, M., Atakan, K. (2008): Continued earthquake hazard in Northern Sumatra: Potential effects of a future earthquake. - EOS, Transactions, American Geophysical Union, 89, 14, 133-134.
EOS
EOS, TRANSACTIONS, AMERICAN GEOPHYSICAL UNION
VOLUME 89, NUMBER 14, APRIL 2008, PAGES 133-134
Continued Earthquake Hazard in Northern Sumatra BY M. B. SØRENSEN AND K. ATAKAN The occurrence of two large earthquakes (Mw = 8,4 and Mw = 7.9) along the Sumatran West coast on 12 September 2007 as well as an Mw = 7.4 event on 20 February 2008 have again put the high earthquake hazard of this region into focus. These events are the most recent in a series of major subduction zone earthquakes that began with the great Mw = 9.3 event of 26 December 2004 followed by an Mw = 8.7 event on 28 March 2005 [Bilham, 2005; Lay et al., 2005; Stein and Okal, 2005]. The major subduction zone earthquakes have been propagating southward along the Sunda trench, and the remaining stress is expected to be released along the subduction zone in a long stretch from the Andaman Sea in the north to the southernmost extension of the recent ruptures, especially in the southernmost part close to the Sunda Strait (Figure 1). Also, there is an additional and significant hazard due to potential earthquakes along the Great Sumatran Fault (GSF), a major right-lateral strike-slip fault parallel to the Western coast of Sumatra. The GSF accommodates the component of plate convergence parallel to the trench, where strain partitioning is a result of the oblique collision along the Sunda trench. During the past two centuries, there has been no significant earthquake along the northern part of the GSF, which is therefore considered a seismic gap (i.e., a region within a seismically active area with a longer-lasting low level of activity) [Bellier et al., 1997]. As stresses build up continually as a consequence of plate convergence, the probability of an earthquake increases with the time elapsed since the previous event. In the case of the northern GSF, a large amount of stress is expected to have accumulated along this part of the fault, which will be released in a future earthquake. Furthermore, occurrences of subduction earthquakes along the Sunda trench have brought the structure closer to rupture [Nalbant ei al., 2005]. The GSF is assumed to be capable of producing earthquakes with magnitudes of up to M = 7.9 [Petersen et al., 2004], and the largest known event occurred in 1892 with M = 7.7 near the city of Sibolga, approximately 570 kilometers southeast of Banda Aceh (see Figure 1). It is therefore urgent that the seismic hazard in Banda Aceh and its surroundings be reevaluated in light of the recent earthquakes. This article illustrates how simulation of ground motions for a likely worst-case scenario earthquake along the northernmost segment of GSF can provide valuable information about the current hazard level in the region. Simulation results indicate that the occurrence of an Mw = 7.7 event with rupture propagating toward Banda Aceh would have severe consequences for the region in terms of loss of life and
Figure 1. Tectonic overview of the study area. Major faults are shown as gray lines, and the scenario fault plane is highlighted with a red line and a star marking the rupture initiation point. Selected focal mechanisms only for strike-slip earthquakes in the Global Centroid Moment Tensor (CMT) database are shown. Outlines of important thrust earthquakes are indicated with yellow lines, and outlines of strike-slip earthquakes are indicated with white lines. The M = 7.4 thrust earthquake on 20 February 2008 is shown as a yellow star. Only the southernmost part of the 26 December 2004 rupture is included. The block arrow indicates the direction of plate convergence in the region.
material damage. Furthermore, we emphasize that smaller earthquakes also can cause widespread damage in the region, especially considering possible amplification effects due to local geological conditions. Earthquake Activity in the Sumatra Region The occurrence of earthquakes in the Sumatra region is controlled mainly by the north-northeast oriented convergence of the Indian-Australian lithospheric plate toward the Sunda plate The rate of convergence is approximately 6 centimeters per year [McCaffrey et al., 2000], and it gives rise to an oblique collision resulting in the partitioning of the accumulated strain. The component of this motion perpendicular to the trench is accommodated by pure thrust earthquakes along the subduction zone. The trench-parallel component of the plate motion (of the order of 3.8 centimeters per year [McCaffrey et al., 2000]), on the other hand, is accommodated by large strike-slip earthquakes that occur mainly along the GSF. Unfortunately, the record of historical earthquakes along the GSF is not complete prior to the nineteenth century. Since 1822, there have been more than 20 earthquakes with M > 6, with eight of these at M > 7. These events were spread across the GSF but occurred mostly in the central and southern parts. Recurrence times for major earthquakes are estimated to be of the order of 200 years in the northern part
EOS
VOLUME 89, NUMBER 14, APRIL 2008, PAGES 133-134
of the GSF, increasing to 400 years for the southernmost part of the fault (approximately south of Bengkulu (Figure 1)) [Bellier et al., 1997]. The mentioned lack of large earthquakes in the northern part of the GSF during the past two centuries, in combination with the estimated approximately 200-year recurrence time, indicates that the probability of a future earthquake here is high. Apart from the GSF, another area with increased earthquake probability is the southernmost part of the Sumatra subduction zone, south of the recent thrust earthquakes. Continued southeastward migration of the ruptures along the plate interface may have significant consequences for population centers such as Jakarta. Effect of Fault Interactions on Seismic Hazard Following the 26 December 2004 earthquake, there was much focus on the possible implications of static stress transfer on neighboring segments of the trench Coulomb stress transfer modeling (i.e., modeling of changes in the stress field induced by an earthquake) [McCloskey et al., 2005] estimated a stress increase along the trench south of the 2004 rupture. These estimates were then manifested by the 28 March 2005 earthquake that occurred south of the 2004 rupture, and again by the 12 September 2007 earthquakes, which were in the region of increased loading following the 2005 event. The most recent earthquake, on 20 February 2008, on the other hand, occurred in the region between the 2004 and 2005 events and thereby most probably represents the release of remaining stresses in this area. In addition to the loading along the trench, a positive loading of up to 9 bars was reported for an approximately 300-kilometer-long segment along the GSF near Banda Aceh [McCloskey et al., 2005] after the 2001 earthquake. The occurrence of the 2005 event was shown to have extended this loading along the GSF farther south [Nalbant et al., 2005]. It is expected that the occurrence of the most recent events (2007 and 2008) has further increased the loading on the GSF, leaving little doubt that the northern segment of GSF has been brought closer to failure. Estimated Ground Motions Due to a Scenario Earthquake Along the GSF
Figure 2. Distribution of bedrock ground motions obtained for the scenario earthquake. (a) Peak ground velocity (b) Peak ground acceleration. The gray line indicates the location of the rupturing fault; the star is the rupture initiation point.
Broadband-frequency (0.1-10 hertz) ground motions due to a scenario earthquake on the GSF were calculated using a hybrid procedure combining a deterministic calculation at low frequencies with a semi-stochastic simulation at high frequencies. This methodology was recently applied to the 26 December 2004 earthquake [Sørensen et al., 2007a]. As input for the modeling, the earthquake source is defined in terms of the location and geometry of the rupturing fault and its asperities (i.e., regions of high slip), together with rupture parameters such as rise time, rupture velocity, stress drop, and seismic moment. The seismic waves generated from this complex finite fault model are propagated through the sur-
rounding crust, which is represented by the seismic velocity structure and attenuation characteristics. Using this approach, we have simulated the bedrock ground motions due to an Mw = 7.7 scenario earthquake rupturing the 190-kilometer-long Aceh segment of the GSF (Figure 1). We have chosen to simulate the rupture of this particular segment because it is where the largest increase in Coulomb stress is observed. Furthermore, the segment is located at the shortest distance from Banda Aceh and therefore poses the greatest hazard. The earthquake hazard in the region is expressed by maps of peak ground acceleration (PGA) and peak ground velocity
EOS (PGV) at bedrock level, calculated based on the simulated seismograms (Figure 2). The distributions of the ground motion values are dominated by the direction of rupture propagation toward the northwest. Maximum values of PGV (170 centimeters per second) and PGA (650 centimeters per second squared) are concentrated around the location of the proposed asperities. Considering the simulation results for a site located in Banda Aceh, it is clear that strong ground shaking can be expected here with bedrock accelerations reaching up to 200 centimeters per second squared and velocities of up to 90 centimeters per second. The duration of simulated strong ground shaking is 20-30 seconds, which is important for the damage capability of the event. The simulation results show that strong ground shaking will affect the northern Sumatra region in the case of a large earthquake along the GSF. The frequency content of simulated ground motion indicates the strongest shaking effects at frequencies of about 0.3-0.4 Hertz, which is expected mainly to have an effect on the fall buildings in Banda Aceh. However, strong accelerations, as indicated from the simulated waveforms, also emphasize higher-frequency ground motions affecting more common low-and intermediateheight buildings. Another important aspect related to the expected damage is the ongoing reconstruction in the region alter the 2004 earthquake and tsunami. This work still is far from finalized, and therefore many buildings may experience increased vulnerability due to already existing damage, especially when the expected duration of the strong shaking is taken into account. The current study does not account for local site effects. The amplification of ground motion due to local unconsolidated sediments is likely to take place in the region, and this would lead to much higher ground motions than predicted here. The effects of local site amplification thus should be considered before the ground motion simulation results can be used in any ongoing risk mitigation efforts. The earthquake scenario presented here is not the only possible scenario for stress release along the northern GSF. Another likely scenario is that the northernmost part of the GSF ruptures in shorter segments, giving rise to smaller-size events and hence smaller ground motions. Therefore, the scenario of an M = 7.7 earthquake represents the worst case and should be considered as a conservative estimate of the seismic hazard. However, considering the potential local site effects, even smaller earthquakes with different rupture propagation directions could cause severe damage. In addition to the uncertainties associated with the segmentation of the GSF, the model parameters of the earthquake scenario (such as rise time, rupture velocity, stress drop, and so forth) are based on a number of assumptions. These uncertainties can be reduced when the earthquake source characteristics
VOLUME 89, NUMBER 14, APRIL 2008, PAGES 133-134 are better understood [Sørensen et al., 2007b]. On the basis of the recent sequence of large thrust earthquakes since 2004 along the subduction interface, it is estimated that a large earthquake is likely to break along the GSF. Although the timing of the event is unknown, it is definitely brought closer to rupture due to static stress transfer from the large thrust earthquakes. Considering that the last large known earthquake along the GSF occurred in 1892 (M = 7.7) along the central part of the fault, the earthquake hazard in northern Sumatra cannot be neglected. This underlines the urgent need for risk mitigation efforts in the affected areas. Acknowledgments We thank Jens Havskov and Gottfried Grünthal for constructive comments and Mohammad Raeesi for contributions to Figure 1. References Bellier, O., M. Sébrier, S. Pramumijoyo, T. Beaudouin, H. Harjono,1. Bahar, and 0. Fomi (1997), Paleoseimicity and seismic hazard along the Ornat Sumatran Fault (Indonesia),J. Geodyn., 24, 169-183. Bilham, R. (2005). A flying start, then a slow Slip, Science, 308, 1126-1127. Lay, T, et al. (2005), The great Sumatra-Andaman earthquake of December 26, M, Saenre, 30.4,1127-1133. McCaffrey, R., P. C. Zwick, Y. Bock, L. Prawiradirdjo, 1. F. Genrich, C. W. Stevens, S. S. 0. Puntodewo, and C. Suharya (2000), Strain partitioning during oblique plate convergence in northern Sumatra: Geodetic and seismologic constraints and numerical modeling. J. Geophys. Res., 105(B12), 28,363-28,376. McCloskey, J., S. S. Nalbant, and S. Stacy (2005), Earthquake risk from coseismic stress, Nature, 434, 291. Nalbant, S. S., S. Steacy, K. Sieh, D. Natawidiaja, and J. McCloskey (2005), Earthquake risk on the Sunda trench, Nature, 435, 756. Petersen, M. D., J. Dewey, S. Hartzell, C. Mueller, S. Hansen, A. D. Frankel, and K. Rukstales (2001), Probabilistic seismic hazard analysis for Sumatra, Indonesia and across the southern Malaysian peninsula, Tectonophysics, 390, 141-158. Sørensen, M. B., K. Atakan, and N. Pulido (2007a), Simulated strong ground motion for the great M 9.3 Sumatra-Andaman earthquake of 26 December 2001, Bull. Seismol. Soc. Am., .97, S139-S151. Sørensen, M. B., N. Pulido, and K. Atakan (2007b), Sensitivity of groundmotion simulations to earthquake source parameters. A case study for Istanbul, Turkey, Bull. Seismol. Soc. Am., 97, 881-900. Stein, S., and E. A. Okal (2005), Speed and size of the Sumatra earthquake, Nature, 434, 581-582.
Author Information Mathilde B. Sørensen, GeoForschungsZentrum Potsdam, Germany; E-mail:
[email protected], and Kuvvet Atakan, Department of Earth Science, University of Bergen, Norway