HR: 1340h
AN: NG43B-0579    [Abstracts]
TI: Time Reversal Imaging of Seismic Sources by the Spectral Element Method.
AU: * Larmat, C
EM: carene@gps.caltech.edu
AF: Seismological Laboratory, California Institute of Technology, 1200E. California Blvd, Pasadena, CA CA91125 United States
AU: Montagner, J
EM: jpm@ipgp.jussieu.fr
AF: Dept. Sismologie, IPG, 4 Place Jussieu, Paris, 75252 France
AU: Fink, M
EM: Mathias.Fink@espci.fr
AF: Laboratoire Ondes et Acoustique, ESPCI, rue Vauquelin, Paris, 75013 France
AU: Capdeville, Y
EM: capdevil@ipgp.jussieu.fr
AF: Dept. Sismologie, IPG, 4 Place Jussieu, Paris, 75252 France
AU: Cl‚v‚d‚, E
EM: clevede@ipgp.jussieu.fr
AF: Dept. Sismologie, IPG, 4 Place Jussieu, Paris, 75252 France
AU: Tourin, A
EM: arnaud.tourin@espci.fr
AF: Laboratoire Ondes et Acoustique, ESPCI, rue Vauquelin, Paris, 75013 France
AB: The increasing power of computers and numerical methods (such as spectral elements methods) makes it possible to simulate more and more accurately the propagation of seismic waves in heterogeneous media and even to conceive new applications such as time reversal experiments within the three--dimensional Earth. These latter use the time reversal invariance and the spatial reciprocity of the wave equation. The idea is to construct a reverse movie of the propagation by sending the time--reversed recorded signals back from the receivers. The energy refocuses back at the location and the time of the original source. The concept of time-reversal has previously been successfully applied for acoustic waves in many fields such as medical imaging, oceanography and non destructive testing. For simulating the propagation of waves in the Earth as well as their time-reversed propagation, we used 2 different techniques, the normal mode summation technique (Gilbert and Dziewonski, 1975) and the spectral element method coupled with the modal solution (Capdeville et al., 2003). The first method is very accurate for 1D-earth models such as PREM whereas the second method is required for general heterogeneous 3D-models. For the first time, we have performed several synthetic and real data time-reversal experiments for seismic waves until the time of focalisation at the source. These tests show that sources are successfully localized in time and in space (though less accurately at depth), especially at very long period (> 200s) where the seismic properties of the Earth are well constrained. The corresponding movies are visible at the following address: http://www.gps.caltech.edu/~carene. We collect and send back the seismograms of the Global network of broadband seismic stations of the Federation of Digital Seismic Network (FDSN). We first consider a moderately large earthquake which can be considered as a point source in both time and space (Peru, June 23, 2001, Mw = 8.4). The coordinates of the earthquake and the focusing in space and time is sharp. For the recent Great Sumatra- Andaman earthquake (Dec 26, 2004, Mw = 9.3), where the point source approximation is no longer valid, we still refocus at the location and time of the centroid and the migration of the rupture from the south towards the north is visible. So far, our time-reversal technique is limited to long period recordings in order to be able to perform many tests in a reasonable computing time. However, it will be possible in the future to improve these results by broadening the frequency range of seismic data in heterogeneous earth models.
DE: 3285 Wave propagation (0689, 2487, 4275, 4455, 6934)
DE: 7209 Earthquake dynamics (1242)
DE: 7290 Computational seismology
SC: Nonlinear Geophysics [NG]
MN: Fall Meeting 2005