Resolving global 3-D anelastic structure in the mantle is important for at least two reasons: (1) The quality factor Q of the earth is considerably more sensitive to temperature than elastic velocity, which implies that, in principle, Q tomography should be able to resolve thermal structure better than elastic tomography, and help us constrain the distribution of chemical versus thermal heterogeneity in the mantle. (2) The dispersion effects in velocity due to attenuation need to be taken into account when interpreting seismic velocity models in different seismic frequency bands.
In this study, we experiment with a waveform formalism, based on mode summation by path average method. Starting with our most recent 
model, SAW24B16 (Mégnin and Romanowicz, 2001) and a degree 16 
model developed with our new data set, we invert iteratively for Q structure of upper mantle. The Q model is parameterized radially in terms of splines, and laterally in spherical harmonics up to degree 8.
One of the common features among existing 1-D Q models is that the highest attenuating region of mantle exists in shallow mantle (e.g., Dziewonski and Anderson, 1981; Durek and Ekstrom, 1996). Various regional studies also show that the lateral variations of Q in the crust and upper mantle can be of an order of magnitude larger than observed lateral variations in velocity. To model the Q structure on the global scale progressively, we first aim at the upper mantle in this study. The strong sensitivity of surface waves to upper mantle makes them the best candidate for this purpose.
Our three component surface data, i.e. the transverse component (Love waves), vertical and longitudinal components (Rayleigh waves), were recorded at IRIS from 249 events for the period 1995-1999, and Geoscope from 440 events for the time period 1993-1999. The dataset consists of first and second orbit fundamental and overtone surface wavetrains, low-pass filtered with a cutoff frequency of 1/60 
and a corner frequency of 1/80
The data were collected using an auto-picking algorithm which compares the raw data with synthetics by mode summation from PREM . Noisy and corrupted traces are eliminated. On a packet-by-packet basis, the residuals, correlation between data and reference synthetics, and amplitude ratio are used as criteria for data elimination. All selected wave-packets are then visually inspected to insure data quality as well.
 |
Images of our degree 8 Q model in upper mantle at three depths are shown in Figure 27.1. The main features in this model are summarized: (1)The pattern in the uppermost mantle is well correlated with tectonic features, with high attenuation below active regions, such as ridges and back arcs, and low attenuation below continental shields and older oceanic regions. (2) As the depth increases, the high attenuation regions are mainly focused below Southern Pacific near French Polynesia and Africa, strikingly correlated in position with the regions of two identified superswells (e.g., McNutt and Judge, 1990 ; Nyblade and Robinson, 1994) and with the two low velocity minima at the base of the mantle, as known from seismic tomography. Figure 27.2 shows the lateral variation and cross section of the upper mantle in Pacific Ocean. The cross section displays a strong low Q column under the Southern Pacific, which may be associated with the high temperature of the rising plume, which exists under the Southern Pacific superswell.
 |
Durek, J. J. and G. Ekström, A radial model of anelasticity consistent with long period surface wave attenuation, Bull. Seism. Soc. Am., 86 144-158, 1996
Dziewonski, A. M., and D. L. Anderson, Preliminary Reference Earth Model, Phys. Earth Planet. Int. 25 297-356, 1981.
Mégnin, C., and B. Romanowicz, The 3-D shear velocity structure of the mantle from the inversion of body, surface, and higher mode waveforms, Geophys. J. Inter., 143 709-728, 2001
McNutt, M. K., and A. V. Judge, The superswell and mantle dynamics beneath the South Pacific, Science, 248 969-975, 1990.
Nyblade, A. A., and S. W. Robinson, The African superswell, Geophys. Res. Lett., 21 765-768, 1994.