HR: 1330h
AN: S22B-0444    [PDF]
TI: Surface Wave Constraints on Q in the Upper Mantle: Isolating the Signal of Attenuation
AU: * Dalton, C
EM: cdalton@fas.harvard.edu
AF: Harvard University, 20 Oxford St, Cambridge, MA 02138 United States
AU: Ekstr\"om, G
EM: ekstrom@seismology.harvard.edu
AF: Harvard University, 20 Oxford St, Cambridge, MA 02138 United States
AB: We use more than 60,000 surface wave amplitude measurements in the period range 150-300 seconds (Ekstr\"om et al., 1997) to construct maps of attenuation, or 1/Q, in the upper mantle. We initially calculate Q by constructing a datum that uses four consecutive wave trains to desensitize the amplitudes to effects from the source, instrument, and elastic structure. These Q measurements are inverted for maps of even-degree attenuation structure, and the results show variations of approximately 50% from PREM. When the Q measurements are averaged for nearly coincident great-circle paths, the resulting attenuation maps are nearly identical to the original ones, confirming that, despite extraneous effects, there is a robust signal in the amplitudes. Using the method of Selby and Woodhouse (2002), we invert minor- and major-arc Rayleigh and Love wave amplitudes for even- and odd-degree Q structure. When we assume that the amplitude anomaly is due entirely to intrinsic attenuation, the ability of the models obtained through this process to fit the data is poor. We next include terms in the inversion that allow the source moment and instrument gain to be corrected. These corrections greatly improve the fit of the data by the models. The path integral approximation to the amplitude anomaly (Woodhouse and Wong, 1986) is used both to predict the effect of focusing from existing phase velocity maps and to jointly invert for attenuation and phase velocity. We also perform a pure-path regionalized inversion using a six-tectonic-region model of the Earth, GTR1 (Jordan, 1981). On the most simplistic level, the results show that oceans are more strongly attenuating than continents at all periods, and that the surface wave attenuation values of PREM fall in between those of continents and oceans. The results using the great-circle Q measurements show many familiar patterns, in particular that young oceans are more highly attenuating than older oceans. Although the data set of minor- and major-arc amplitudes is somewhat noisy, its ability to match the results of the great-circle Q regionalization is a good measure of the usefulness of the amplitude corrections described above.
DE: 7218 Lithosphere and upper mantle
DE: 7255 Surface waves and free oscillations
SC: Seismology [S]
MN: 2003 Fall Meeting