HR: 1330h
AN: S22B-0443    [PDF]
TI: Towards the next generation of global 3D upper mantle Q models
AU: * Gung, Y
EM: gung@seismo.berkeley.edu
AF: Berkeley Seismological Laboratory,, 215 McCone Hall, Univ. of California, Berkeley, Berkeley, ca 94720 United States
AU: Romanowicz, B
EM: barbara@seismo.berkeley.edu
AF: Berkeley Seismological Laboratory,, 215 McCone Hall, Univ. of California, Berkeley, Berkeley, ca 94720 United States
AU: Capdeville, Y
EM: yann@seismo.berkeley.edu
AF: Institut de Physique du globe de Paris, 209 McCone Hall, Univ. of California, Berkeley, berkeley, ca 94720 United States
AB: Global anelastic tomography can bring important constraints on the thermal structure of the mantle and therefore is dynamics, complementing those provided by elastic tomography. Progress in anelastic tomography has been slow, because of the inherent technical difficulties encountered in discriminating anelastic signal from elastic effects on amplitude data. It has been shown that while the elastic focusing/defocusing effects are not significant at low degrees ($\sim$ 8) (e.g. Selby and Woodhouse, 2002; Gung and Romanowicz, 2003), they need to be included to achieve a higher resolution Q model. Ideally, one would use an exact method, such as the Spectral Element Method (SEM) for predicting the focusing effects. SEM is however very heavy computationally. We present a procedure to better constrain the 3D upper mantle Q from 3 component long-period seismic waveforms. In this procedure, the amplitude and phase perturbations due to the 3D elastic structure are corrected for using higher order normal mode asymptotic theory, and applying it to current elastic models. We first evaluate the normal mode asymptotic approach by comparing the corresponding 3D synthetics with those computed using the coupled spectral element/normal mode method (CSEM). 3 normal mode based asymptotic approaches are compared: path average approximation (PAVA), non-linear asymptotic coupling theory (NACT) and NACT+F, an extension of NACT with focusing terms computed using higher order asymptotic theory. Systematic waveform comparison and inversion experiments are implemented. We find that (1) when the anomaly lies on the source-receiver great circle path, the 3 techniques are fairly accurate for fundamental mode surface waves, but NACT and NACT+F provide much better fit for overtone phases and are therefore more powerful in resolving 3D structure in the mid and lower mantle; and (2) the off-great-circle effects, which result in focusing/defocusing and not seen by PAVA or NACT, are well explained by NACT+F, at distances $30^o$ or more away from the source or its antipode. We discuss a preliminary 3D upper mantle Q model obtained by waveform inversion by this new procedure.
DE: 7255 Surface waves and free oscillations
DE: 7260 Theory and modeling
SC: Seismology [S]
MN: 2003 Fall Meeting