| Seismic imaging has led to spectacular advances in our knowledge and understanding of Earth's deep interior. Our previous work has shown, for example, that slabs of subducted lithosphere can sink to large depths into Earth's lower mantle, thus rendering untenable conventional models of convective layering at 660 km depth. Key issues pertaining to the nature and interpretation of seismic models remain enigmatic, however. Uneven data coverage and inaccurate theoretical approximations render non-unique, fuzzy images, with substantial lateral variations in reliability. This prevents tomography from reaching its full potential as a quantitative probe of Earth's deep interior. Confident that we can do better and learn more, and inspired by recent advances in computational seismology, computer technology, and wave propagation theory, we embark on an ambitious effort to generate new classes of tomographic model. PC clusters (BEOWULF) will be used for novel data processing, model construction, and validation. With better - but computationally 'expensive' - wave propagation theory we will infer spatial variations in (anisotropic) wave speed from very large amounts of automatically processed seismic data. The ability to compute accurate synthetic seismograms for realistic (heterogeneous, anisotropic) Earth models has begun to open exciting new avenues of research, and we aim to use this to test models against observed records and to search the data-space for observations that remain unexplained by 3D Earth models. The powerful new capabilities will enable a diverse range of applications but will initially be used to gain quantitative understanding of two parts of Earth's interior that contain critical clues to Earth's evolution: the bottom 1000 km of the global mantle (where we hope to discover robust evidence for compositional variations) and the continental mantle beneath Eurasia (to constrain composition, anisotropy, and discover how continental structure changes over time). Keywords: seismic tomography; new theoretical and computational concepts; mantle convection and composition; continental anisotropy and evolution; Eurasia. |
