"The Earth mantle: lumpy 'plum pudding' or streaky 'marble cake'?"
PhD project title:
The Seismic Signals of the Heterogeneous Earth Mantle.
When applied to the deep mantle, traditional geophysical imaging methods face unavoidable limitations, especially in terms of spatial resolution. Geochemical data provide even lower resolution power. Thus, Earth's deep interior remains poorly constrained, limiting our understanding of the planet's composition and long-term evolution. In this context, two main models for the mantle appear equally viable. One is a "marble cake"-like regime, where subducted oceanic lithosphere is recycled in the mantle and stretched into narrow streaks by convection. The other is a "plum pudding" mantle, in which coherent blobs of primordial material (possibly enriched in the mineral bridgmanite) may resist convective entrainment thanks to their intrinsic strength.
In this project, we will predict the seismic signatures of these two models and compare them to observations. Thanks to numerical modeling methods, we will simulate the thermochemical evolution of the mantle to build a database of 2D "marble cakes" and "plum puddings". From this database, seismic velocity maps will be obtained via thermodynamic modeling tools, allowing us to generate an ensemble of synthetic seismograms. At this point, we will be able to identify model-specific features in the waveforms and, with the help of a neural network, isolate them in seismograms from a real dataset. This approach will provide us with a novel, quantitative method to test geodynamic model predictions.
Our goal is to reduce the uncertainty surrounding the compositional structure of the present-day, lower mantle, thus advancing our understanding of Earth's long-term dynamics.