HR: 09:15h
AN: U21C-05    [Abstracts]
TI: Very broad-band modeling of the Dec.~26th, 2004, Sumatra-Andaman Earthquake
AU: * Hjorleifsdottir, V
EM: vala@gps.caltech.edu
AF: SeismoLab, MS 252-21, Caltech, Pasadena, CA 91125 United States
AU: Ji, C
EM: ji@geol.ucsb.edu
AF: Department of Earth Science, UCSB, Santa Barbara, CA 93106 United States
AU: Song, T A
EM: alex@gps.caltech.edu
AF: SeismoLab, MS 252-21, Caltech, Pasadena, CA 91125 United States
AU: Liu, Q
EM: lqy@gps.caltech.edu
AF: SeismoLab, MS 252-21, Caltech, Pasadena, CA 91125 United States
AU: Tromp, J
EM: jtromp@gps.caltech.edu
AF: SeismoLab, MS 252-21, Caltech, Pasadena, CA 91125 United States
AU: Kanamori, H
EM: hiroo@gps.caltech.edu
AF: SeismoLab, MS 252-21, Caltech, Pasadena, CA 91125 United States
AB: The December 26th, 2004 Sumatra-Andaman earthquake ruptured along a 1200 km long stretch of the plate boundary. The resulting displacements of the solid Earth were observed both by the global seismic network and by GPS stations around the source area. We present a model of co-seismic slip, based on inversion of teleseismic body waves as well as regional and global seismic waveforms. This model has slip distributed from Simuele Island in the south to the Andaman Islands in the north, with the largest slip in the south. The average rupture velocity over the whole rupture is 2.4 km/s, but less than 2 km/s during the first 200 seconds. Local rise times are up to 50 seconds, and the total moment is 6.5*10²²~Nm. We show that this model can explain the observations over the seismic frequency band. Furthermore, it can predict the measurements at GPS stations in the far-field as well as displacements observed with a continuous GPS array in Malaysia and Thailand to within 20~% (one day solutions). These data were not used to invert for the model. The only measurements that our model is not consistent with are the near-field campaign GPS, as measured in the Andaman and Nicobar Islands. These measurements reflect displacements on a longer time scale, up to a few weeks, and are therefore not necessarily representative of the co-seismic displacements from which the model was constructed. These results indicate that the same model can explain the displacements measured in different seismic frequency bands, as well as far-field GPS, without invoking large slip over large areas at timescales between 10 minutes and a day. We are working on adjoint simulations of the earthquake, in which the waveforms observed at seismic stations are back-propagated, such that they converge on the source, illuminating the source region.
DE: 7209 Earthquake dynamics (1242)
DE: 7215 Earthquake source observations (1240)
DE: 7230 Seismicity and tectonics (1207, 1217, 1240, 1242)
DE: 7240 Subduction zones (1207, 1219, 1240)
DE: 7290 Computational seismology
SC: Union [U]
MN: Fall Meeting 2005