The last two decades have produced a large collection of tomographic models of the mantle on a global scale, with the scale length of the structure resolved increasing with time. Significant improvements to the quality of images of the deep mantle have been the result of efforts aimed at deriving a theoretical formalism for body waves as accurate and computationally efficient as possible. The emergence of time-domain waveform tomography suited to the modeling of body waves (Li & Romanowicz, 1995) has allowed to treat the data in a normal mode theoretical framework 'Nonlinear Asymptotic Coupling Theory' (NACT) that accommodates the existence of Fresnel zones, and allows the modeling of all the energy on the seismic trace, providing excellent sampling of mantle structure. A waveform extraction scheme in which each energy arrival in the seismic trace is selected separately has proved a powerful tool to resolve lowermost mantle structure (Mégnin & Romanowicz, 1999). A model of the shear velocity structure of the mantle expanded laterally up to spherical harmonic degree 12 and radially in Legendre polynomials was derived with this methodology from the inversion of transverse component waveforms of body and surface waves for events of magnitude greater than 5.5 for the period from 1977 to 1994 recorded at IRIS and GEOSCOPE stations (Li & Romanowicz, 1996). In the present study, we used the same methodology to derive a higher resolution model 'SAW24B16' expanded laterally up to degree 24 and radially in cubic b-splines. The adoption of local basis functions in the radial parametrization allows to accommodate the varying ray sampling of mantle structure with depth, which is precluded by the use of global basis functions. Taking advantage of recent deployments of broadband seismograms, we expanded the existing collection of seismograms with new body and fundamental surface wave information, and introduced higher mode waveforms into the data set. The present model was derived from the inversion of 31,000 body waves, 9,300 fundamental and 1,400 overtone surface waves.
Our model shows in general fast anomalies associated with subduction that agree with results from many regional studies. In figure 23.1, we show such an example for the West Pacific. The top of the figure shows the model at 175 km depth in that region. Fast anomalies are associated with the Australian and Asian continents. Slow anomalies offshore correlate with mid-ocean ridges as well as back-arc spreading.
Figure 23.1 (A) is a cross section of the model across the southern Kurile Islands region, displaying a subhorizontal slab sinking to the 670 km discontinuity without entering the lower mantle. This observation is consistent with Vp models derived from ISC data (e.g. Fukao et al., 1992).
The cross-section of figure 23.1 (B) is taken across the Japan Trench. The slab appears to `rest' horizontally on the discontinuity, in agreement with (e.g. van der Hilst et al., 1991). The anomaly penetrates into the lower mantle at its eastern end. The amplitude of this feature is nevertheless weak and may not be well resolved but it agrees qualitatively with the results of Tajima et al., 1998.
The cross-section of figure 23.1 (C) is taken across the Mariana trench region and suggests that slab structure is present in the lower mantle, down to a depth of about 900 km.
In the region across the Solomon trench (figure 23.1 (D)), the slab appears to be confined to the transition zone and the uppermost part of the lower mantle. In contrast, the section across the Tonga-Kermadec region suggests that the slab penetrates into the lower mantle (figure 23.1 (E)). Continuation of the Tonga slab into the lower mantle has been observed previously (e.g. Fisher et al., 1991).
In figure 23.1 (F), we show a cross-section across the Java subduction zone. The image suggests the presence of slab in the deep mantle, in agreement with Fukao et al., 1992. Although this figure suggests that the slab reaches the CMB, the amplitudes below 1,500 km are low, indicating that the depth extent of this fast anomaly may not be reliably addressed with the present model.
The model presents three large scale slow anomalies that extend continuously from the bottom of the mantle into the uppermost mantle, one under Africa and two in the Pacific. In figure 23.2, we show an image of the upwelling beneath the African continent.
Figure 23.2 (top) shows a lateral section of the upper mantle at 175 km depth under Africa and the Indian Ocean, displaying fast anomalies beneath the African and Indian plates and slow anomalies near the East African rift zone, the Red Sea and along the mid-Indian ridge. The East African, West African and Congo cratons are isolated by the presence of slow regions surrounding them.
Figure 23.2 (A), shows a SW-NE cross-section across the continent. The slow domain under Africa extends vertically from the base of the mantle into the upper mantle and is subsequently shifted to the northeast. The upper mantle image is consistent with the closure of the Tethys by the African continent throughout the Cretaceous: in our tomographic model, the upper part of the low velocity anomaly is offset in the direction of motion of the African plate. Figure 23.2 (B) shows an East-West cross-section. The low velocity zone is deflected to the east by the Nubian plate where it connects with the mid-Indian ridge. In the lower mantle, the shape of the anomaly in SAW24B16 is in general agreement with the model M2 of Ritsema et al, 1998, suggesting an upwelling shifting in the northeast direction with increasing radius. However, in contrast with their study, we find continuity of the slow domain into the upper mantle.
This study shows that, with a relatively restrained data set consisting of waveforms of all the phases present on a seismogram, it is possible to derive global images of the mantle on scale lengths of , consistent with results of regional studies. This is related to the adoption of a waveform methodology to image deep Earth structure. We use a theoretical formalism valid both for body and surface waves that allows 1) the modeling of finite frequency effects on the seismogram and 2) the use of all the information on the seismic trace. In particular we make use of the S phase diffracted on the core-mantle boundary which cannot be modeled with a ray-theoretic approach and which provides the bulk of lower mantle information; we also use simultaneous phase arrivals which are precluded by travel time studies as they rely on the the determination of an onset time for each datum. This methodology is well suited to the imaging of both fast and slow domains in the earth. The present model suggests the existence of three large scale plume-like slow anomalies in the mantle, one under Africa, and two in the Pacific.
Fukao, Y., M. Obayashi, H. Inoue, and M. Nenbai, Subducting slabs stagnant in the mantle transition zone, J. Geophys. Res., 97, 4,809-4,822, 1992.
Fischer, K. M., K. C. Creager, and T. H Jordan, Mapping the Tonga slab, J. Geophys. Res., 96, 14,403-14,427, 1991.
Li, X.D. and B. Romanowicz, Comparison of global waveform inversions with and without considering cross-branch modal coupling, Geophys. J. Int., 121, 695-709, 1995.
Li, X.D. and B. Romanowicz, Global mantle shear-velocity model developed using nonlinear asymptotic coupling theory, Geophys. J. R. Astr. Soc., 101, 22,245-22,272, 1996.
Mégnin, C. and B. Romanowicz, The effects of the theoretical formalism and data selection on mantle models derived from waveform tomography, Geophys. J. Int., 138, 366-380, 1999.
Ritsema, J., S. Ni, D. V. Helmberger, and H. P Crotwell, Evidence for strong shear velocity reductions and velocity gradients in the lower mantle beneath Africa, Geophys. Res. Lett., 25, No 23, 4,245-4,248, 1998.
Tajima, F., Y. Fukao, M. Obayashi, and T. Sakurai, Evaluation of slab images in the Northwestern Pacific, Earth Planets Space, 50, 953-964, 1998.
van der Hilst, R. D., E. R. Engdahl, W. Spakman, and G. Nolet, Tomographic imaging of subducted lithosphere below northwest pacific island arcs, Nature, 353, 37-42, 1991.