Global seismic tomography aims at improving our understanding of mantle dynamics by providing constraints on three-dimensional (3D) temperature and composition, using elastic velocities as proxies. Much progress has been made in recent years in resolving increasingly finer details in the 3D distribution of elastic velocities from the inversion of seismic phase and travel time data. However, the detailed morphology and role of upwellings, as manifested by two prominent zones of lower than average velocity, commonly referred to as superplumes, in the lowermost mantle, is not yet clear. Their location, under the south-central Pacific and under Africa, correlates with the global distribution of hotspots, as well as two major geoid highs. Recent tomographic S wave velocity models suggest that the superplumes rise high above the core-mantle boundary (CMB)Mégnin and Romanowicz, 2000; Ritsema et al., 1999, and joint seismic/geodynamic studies imply that they may be active upwellings (Forte and Mitrovica,2001). However, finer scale resolution is still lacking.
To obtain additional constraints on hotter than average features, we consider amplitudes of seismic waves, which are sensitive to 3D anelastic structure. Owing to the exponential dependence on temperature of attenuation, which we shall express in terms of , where is the quality factor, we expect anelastic tomography to highlight hotter than average regions better than standard elastic tomographic approaches.
We have developed a waveform tomographic inversion method, originally aimed at constructing global 3D elastic models of the whole mantle (Li and Romanowicz, 1996), which now has been extended to iteratively solve for elastic and anelastic structure in the upper mantle, using three-component waveform data of fundamental and higher mode surface waves (Gung and Romanowicz,2002). While we do not directly account for elastic effects in the amplitudes, strict data selection criteria are designed to reject data most strongly contaminated by focusing. The first step (elastic inversion) allows us to align the phases in our waveforms, and we do so separately for SV sensitive (vertical and longitudinal S wave component) and SH sensitive (transverse S wave component) data, to account for anisotropy in the uppermost mantle. In the second step, the model, , is derived using all three component data.
In the top 250 km of the mantle, correlation of high Q regions with shields is seen systematically in North and South America, Eurasia, Australia and Antarctica, whereas mid-ocean ridges in the Pacific, Atlantic and Indian Ocean exhibit generally low Q values, as do western Pacific back-arc regions (Figure 33.1). This is similar to what is observed in elastic velocity models, with regions of high/low velocity correlated with regions of high/low Q. A notable exception is an elongated zone of high attenuation in the central Pacific, extending from south of the equator to Hawaii, not seen in velocity models at these depths. Below 250 km, this tectonics-related Q distribution is gradually replaced by a simpler pattern, with two strong attenuation maxima centered in the southern Pacific and under Africa, throughout the upper mantle transition-zone. At depths greater than 400 km, a majority of hotspots are located above regions of high attenuation.
In model , the high attenuation regions in the transition zone coincide in location with the minima in elastic velocity associated with the two superplumes in the lowermost mantle. Correlation between Q in the transition zone and velocity in the last 500 km of the mantle is particularly strong at degree 2 (Figure 33.2), but persist at shorter wavelengths. Cross-sections in the Pacific and under Africa (Figure 33.3) comparing upper mantle with lower mantle velocity distributions, emphasize the vertical correspondence of the lowermost mantle superplumes with transition zone low Q zones. Because our model does not extend to the lower mantle, and the low velocity zones are only expressed faintly in the upper half of the lower mantle, where they appear to be narrower and have a contorted shape, it is not possible to determine whether the superplumes are simply continuous across the 670 km discontinuity, or whether they induce rising hot currents in the upper mantle through thermal coupling processes. However, our results show that the superplumes must carry enough energy across the lower mantle to create coherent upwelling flow in the upper mantle transition zone, in agreement with mantle flow models. In contrast, ridges are shallow high attenuation features, mostly confined to the upper 200 km of the mantle.
The low Q zones in the transition zone connect with shallower ones whose positions are shifted horizontally, suggesting that the upwelling, plume related flow is deflected horizontally below the cold lithosphere, towards the Indian and Atlantic mid-ocean ridges under Africa, and in the Pacific, towards the East Pacific rise and the center of the Pacific plate. In the latter case, the flow is impeded on the west side by the presence of the Fiji-Tonga subduction zone. This deflection occurs at greater depths under the thicker continental lithosphere (, Fig. 3 b) than under the oceanic one (, Fig. 3a).
We thank the National Science Foundation for support of this research.
Forte, A. M. and J. X. Mitrovica, Nature, 410, 1049 (2001)
Gung, Y.C. and B. Romanowicz, Q tomography of the upper mantle using three component long period waveforms, submitted to J. Geophys. Res.
Li, X.D. and B. Romanowicz, Global mantle shear velocity model developed using nonlinear asymptotic coupling theory, J. Geophys. Res. 101 22,245 (1996).
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
Ritsema, J. , H. van Heijst and J. Woodhouse, Complex shear wave velocity structure imaged beneath Africa and Iceland, Science 286, 1925-1928, 1999.
Romanowicz, B. and Y. C. Gung, Superplumes from the core-mantle boundary to the base of the lithosphere, Science,296, 513-516, 2002.