"The long-lived paradox between geochemical arguments for two compositionally distinct mantle reservoirs and geophysical evidence for extensive mass exchange between upper and lower mantle has still not been resolved in a manner to reach consensus. Recently, finite frequency tomography has produced images of large lower mantle plumes, which potentially carry a significant amount of heat through the mantle. This raises the prospect that the dynamic role of these plumes is larger than inferred classically from observation of dynamic topography. A consistent evaluation of this question requires a combination of large-scale numerical forward simulations of mantle circulation, mineral physics and seismology, complementing the tomographic inversions. We shall wish to focus on the boundary between the upper and lower mantle (the 660 km discontinuity) using seismic data from French Polynesia (the PLUME experiment). The most important aspect will be to test whether the seismic velocity images should be interpreted as a thermal boundary layer at the discontinuity, as sometimes suggested, or if the anomalies reflect a multi-phase transition pattern. Thereby, we estimate the heat transported by plumes through the mantle and the probability of mass exchange across this boundary. Tools to answer this question are: high-performance computations based on 3D finite element as well as 3D spectral element techniques to model mantle circulation and 3D global wave propagation, respectively. In addition, a thermodynamically self-consistent mineralogical model based on Gibbs Free Energy minimization is used to relate temperatures to seismic velocities. First theoretical investigations based on the above approach have demonstrated its feasibility and potential for the proposed project (Schuberth et al. [2008a,b]). The results of this study will be relevant to geophysical as well as to geochemical deep Earth studies and will help in building conceptual models of the Earth."