| We present evidence that seismic imaging based on waveform inversion is capable of resolving the spectral signature of a convecting mantle, by computing tomographic images of a 3-D spherical convection simulation. The input model includes the thermal effects on the mantle's convection planform of 20% bottom heating, two phase transitions at 410 km and 670 km, a depth-dependent viscosity profile, and the history of plate motions starting in the mid-Mesozoic. The comparison between images of this model ‘seen’ by the tomography and a model of shear velocity in the mantle indicates that the part of the temperature field corresponding to cold mantle heterogeneity in the convection simulation agrees well with the fast velocities derived from tomography, but that current convection computations are lacking some key ingredients necessary to reproduce the character of upwellings correctly. After quantifying the distortive effect of the aliasing of small scale heterogeneity into long wavelength models, and of the use of the asymptotic expansion of a Born seismogram in the waveform modeling, we derive a model of 3-dimensional seismic shear velocity in the mantle by inverting a dataset consisting of body, surface, and higher mode waveforms.; Our model shows the presence of three low velocity zones continuous throughout the depth of the mantle, one under the African Plate and two in the Pacific. These ‘megaplumes’ connect at the surface with the mid-Indian ridge and the complex of ridges constituted by the East-Pacific Rise, the Chile Ridge and the Pacific-Antarctic Ridge, respectively. Our model suggests the presence in the upper mantle of isolated low velocity anomalies that coincide with the surface location of many hotspots, in particular in the Pacific region. No significant accumulation of low velocity anomalies is observed in the upper part of the lower mantle, suggesting that the 670 km discontinuity does not act as a significant barrier to mantle flow. The spectral analysis of the model suggests a narrowing of the slow anomalies in the lower mantle with increasing radius. Fast anomalies coinciding with subduction regions suggest that, except for parts of the Northwestern Pacific, slabs penetrate into the lower mantle. This provides further evidence that the phase boundary may not impede flow between the upper and the lower mantle. |
