U22A-01
Mapping The Ruptures of the Great Sumatra-Andaman Earthquakes
We use Rayleigh waves with periods between 100 and 500 seconds to map seismic moment distribution during the 26 December, 2004 Great Sumatra-Andaman earthquake along the northern Indonesian subduction zone and the associated 28 March, 2005 event. To isolate source effects from the observed waveforms we deconvolved normal-mode synthetic seismograms computed using PREM and the Harvard CMT mechanism located at the USGS-NEIC epicenter. We obtain about 200 useable R1 source time functions with a good azimuthal distribution, admittedly more densely sampled to the north. Our results suggest that the major normal seismic moment release began about 40-60 seconds after the initial slip (identified by the USGS-NEIC origin time) and then continued for about 400-450 additional seconds. The 28 March event included two regions of large slip, one beneath Nias Island. For both events, the surface-wave time-function based models are consistent with seismogram modeling efforts using similarly parameterized faults. We also demonstrate the utility of R1 observations perpendicular to large, low-angle thrust events to construct simplified, smooth images of the rupture suitable for estimating tsunami potential of large earthquakes. An important lingering question regarding the 26 December rupture is the timing of the slow (postseismic) slip observed in GPS observations completed several weeks later, but not contained in seismogram-based models. We have examined the later, low-frequency signals of the surface-wave time functions during the hours immediately following the onset of slip. We find little coherent motion in the long-period seismic band during the first few hours after the event, which suggests very slow motion with a long time history, perhaps taking hours (or longer) to accumulate the meters of offset observed in the GPS observations on the northern third of the aftershock zone.
U22A-02
Constraining Slip Histories of Recent Great Earthquakes Using Broadband Waveforms of Global Seismic Network
We have developed a new inverse method which constrains the slip history of large earthquakes by combined inverting teleseismic body waves and long period surface waves. These two datasets are mutually complementary in the studies of earthquake sources. The long period surface waves could robustly constrain the general parameters of ruptures, such as seismic moment, but are lack of resolution for the detailed rupture process. In contrast, the relative high frequency teleseismic body waves are sensitive to the detailed rupture process but could not well resolve the total seismic moment when the rupture duration is long, presumably due to the interference with the free surface. Combining these two datasets together could improve the data coverage in frequency domain. By carefully choosing time windows, we have also been able to use more seismic body phases in source study, which improves the spatial data coverage. Here, the along-strike and downdip resolutions of this new method are evaluated using a check-box model similar to the fault geometry of the 2004 great Sumatran-Andanman island earthquake. Seven great earthquakes (Mw>8) since 1990 will be revisited too.
U22A-03
The Spatio-temporal Context of the December 26, 2004 Aceh-Andaman Earthquake
The available seismic history of the Andaman-Nicobar region indicates that in terms of size and fault length (1300 km), the M 9.3 December 26, 2004 earthquake is unprecedented in the region. A tsunamigenic earthquake (M ~7.5) is known to have occurred off Car Nicobar in 1881, reportedly defining a 300- km-long segment. Another event occurred in 1941 ruptured 300-400 km of the subduction interface, extending from middle to south Andamans. Available tide gauge data and eyewitness accounts suggest that the 1941 event, unlike the 1881 earthquake, may not have generated any tsunami. An earlier one occurred in the Andaman offshore on 28 January 1679 with its far-field felt area similar to the 1941 event. Historical data and estimations based on convergence rate from the Andaman-Nicobar region indicate a recurrence period of 200+-50 yr for the large earthquakes (M 7.5-8.0). Our field observations on coseismic uplift at various locations of the Andaman-Nicobar arc and the sites on the east coast of India, however, argue for a longer recurrence interval for the giant tsunamigenic earthquakes. The available age data of the coral terraces indicate that two major onshore uplift events (about 1 m) around 1000 and 3000 yr-BP. Historical information and stratigraphic context of suspected tsunami deposits from an archaeological site (7-12th century A.D) located in the east coast of India hint at the possible occurrence of a great tsunami around 900 A.D. From the available data, it appears that the 2004-type, megathrust events may be rare, aperiodical events. The temporal irregularity of such events vis-a-vis the more frequent segment-specific large events (e.g. 1679, 1881, and 1941), may be characteristic specific to the subduction zone. The northern segment between the North Andamans and the Burmese Coast (the Irrawady delta) is the most potential segment that remains to be broken. Historically this segment is quiescent, but we observed indications of previous liquefaction in the geological record in that region, whose size and timing need to be worked out.
U22A-04
Rupture process of the 2004 great Sumatra-Andaman earthquake estimated from tsunami waveforms
Rupture process of the 2004 Sumatra-Andaman earthquake is estimated using five tsunami waveforms observed at tide gauges (Sibolga Belawan, Colombo, Vishakhapatnam, and Prot Blair) and tsunami height data obtained from two satellite altimetry data, _gJason-1_h and _gTOPEX/Poseidon_h. The coseismic vertical deformation surveyed along the coast of Sumatra Island, Nicobar Islands, and Andaman Islands, are also used to constrain the fault model. The average rupture speed of the 2004 Sumatra-Andaman earthquake is estimated to be about 2 km/s from tsunami waveform analysis. The rupture extends about 1200 km toward north-Northwest along the Andaman trough. The large slip of more than 20m is estimated on the plate interface off the northwest coast in the Aceh province in Sumatra. Another large slip of more than 20m is also estimated on the plate interface beneath the north of Simeulue Island in Indonesia. The other large slip of 10-15m is estimated on the plate interface near Little Andaman and Car Nicobar Inlands. The slip amount beneath North and Middle Andaman Islands are very small, about 1m. The total seismic moment is calculated to be 7.8 x 1022 Nm (Mw 9.2) which is similar to the other studies using seismic waves (Park et al., 2005, Ammon et al., 2005). Our estimated slip amount off Sumatra Island is larger than the slip amounts estimated by the other studies, such as Ammon et al (2005). This large slip should be responsible for large tsunami run-Up heights of about 35m surveyed along the northwest coast of Ache province in Sumatra Island.
U22A-05
Constraining the Rupture Length, Duration and Speed of the Great Sumatra-Andaman Earthquake Using Hydroacoustic Data
Hydroacoustic data from the International Monitoring System Diego Garcia South hydrophone station have been used to track the rupture of the Great Sumatra-Andaman Earthquake using the T-wave arrival. T-waves are generated by radiation of seismic energy at the crust-water interface, and they can travel great distances with relatively little loss in energy through a low-velocity waveguide known as the SOund Fixing and Ranging (SOFAR) channel. For this event the T-wave allows us to provide a relatively high-resolution picture of changes in the rupture speed and character. A single station approach allows topographic steering effects to be minimized. Results show that there were at least two large-scale phases of rupture with the first 180 seconds having an average speed of 2.8 km/s and the next 300 seconds having an average rupture speed of 2.1 km/s. Waveform amplitudes show two primary pulses of energy coincident with the two phases of rupture speed. Results also show that the full-length of the northern portion of the fault zone ruptured up to the latitude of the pole of rotation between the Burma and Indian Plates where subduction ends. Aftershock patterns mapped using the hydroacoustic data illustrate that in the first 30mins to few hours following the mainshock, aftershock numbers/size tracked the amplitude of waveform from the mainshock. T-wave data have the potential to address a number of fundamental questions of the rupture of this large event and also provide another tool for rapid assessment of the scale of a shallow submarine event and its tsunamigenic potential.
U22A-06
Sumatra-Andaman Megathrust Earthquake Slip: Insights From Mechanical Modeling of ICESat Surface Deformation Measurements
Modeling the kinematics of the 2004 Great Sumatra-Andaman earthquake is limited in the northern two-thirds of the rupture zone by a scarcity of near-rupture geodetic deformation measurements. Precisely repeated Ice, Cloud, and Land Elevation Satellite (ICESat) profiles across the Andaman and Nicobar Islands provide a means to more fully document the spatial pattern of surface vertical displacements and thus better constrain geomechanical modeling of the slip distribution. ICESat profiles that total ~45 km in length cross Car Nicobar, Kamorta, and Katchall in the Nicobar chain. Within the Andamans, the coverage includes ~350 km on North, Central, and South Andaman Islands along two NNE and NNW-trending profiles that provide elevations on both the east and west coasts of the island chain. Two profiles totaling ~80 km in length cross South Sentinel Island, and one profile ~10 km long crosses North Sentinel Island. With an average laser footprint spacing of 175 m, the total coverage provides over 2700 georeferenced surface elevations measurements for each operations period. Laser backscatter waveforms recorded for each footprint enable detection of forest canopy top and underlying ground elevations with decimeter vertical precision. Surface elevation change is determined from elevation profiles, acquired before and after the earthquake, that are repeated with a cross-track separation of less than 100 m by precision pointing of the ICESat spacecraft. Apparent elevation changes associated with cross-track offsets are corrected according to local slopes calculated from multiple post-earthquake repeated profiles. The surface deformation measurements recorded by ICESat are generally consistent with the spatial distribution of uplift predicted by a preliminary slip distribution model. To predict co-seismic surface deformation, we apply a slip distribution, derived from the released energy distribution computed by Ishii et al. (2005), as the displacement discontinuity boundary condition on the Sumatra-Andaman subduction interface fault. The direction of slip on the fault surface is derived from the slip directions computed by Tsai et al. (in review) for centroid moment tensor focal mechanisms spatially distributed along the rupture. The slip model will be refined to better correspond to the observed surface deformation as additional results from the ICESat profiles become available.
U22A-07
Inversion of satellite altimetry data to recover the Sumatra 2004 earthquake slip distribution
The great Sumatra tsunami of 26 December 2004 confirmed the possibilities of measuring the wave heights using satellite altimetry. Although these are not the first observations of the kind, they are the first quality data sets (signal to noise ratio ten times better than any previous observation) that can be used to study propagation effects and sources parameters without the need to take into account the strongly non-linear coastal effects. We focus on the earthquake source assuming linear earthquake-tsunami coupling and propagation. This linearization allows us to generate Green functions for different parts of the fault plane and to invert for their relative contribution to the altimetry waveforms. We estimate the validity of our approach through synthetic tests and check if the problem is well-posed. Our final solution is in very good agreement with recent GPS and seismological studies with two main asperities releasing energy: the strongest being localized at the latitude of Aceh and the second around the island of Car Nicobar. Finally, we compute a forward model of this solution and compare it with tide gauges records available around the Indian Ocean.