UCLA

The dynamical and chemical processes of the Earth’s interior are the critical driver of the evolution of various “spheres”, like the hydrosphere, atmosphere, and biosphere, and shape the surface where we live. While the interior of the Earth is largely inaccessible, the surface volcanism provides a unique window into the planet’s physical and chemical evolution. The majority of the volcanism on Earth is generated by decompression melting at the mid- ocean ridges, with a smaller proportion due to dehydration melting at subduction zones, and melting at intraplate volcanoes. Intraplate volcanism away from ridges (like Hawaii) and excessive melting anomalies near or at ridges (like Iceland) are not readily explained by plate tectonics. Such phenomena are also known as hotspots.

Hotspots could be fed by active upwellings, or mantle plumes, a distinct scale of flow within mantle convection, separate from the plate-scale flow, and directly linked to the bot- tom thermal boundary layer. In particular, it has long been suggested that hotspots originate from the Large Low Shear Velocity Provinces (LLSVPs) at the core-mantle boundary (CMB), as imaged by seismic waves. However, the presence of mantle plumes under hotspot volcanoes extending to the CMB, and the possible link between LLSVPs and plumes have been the source of debate.

In this thesis, I aim to address the link between hotspots and LLSVPs and their relation to the bottom thermal boundary layer by analyzing surface observations and performing analog experiments. By employing a multi-disciplinary approach, this research facilitates a comprehensive exploration from the Earth’s surface to its deep interior. This holistic investigation helps bridge the gap in our understanding of the geophysical and geochemical signals seen at the top thermal boundary layer, and their correlation to the dynamics and entrainment phenomena at the bottom thermal boundary layer.

In Chapter 2, I use self-consistent thermodynamic calculations to infer the potential temperature beneath hotspots and mid-ocean ridges in the mantle from seismic tomography. By comparing the excess temperature of hotspots over ridges with buoyancy flux estimates and geochemical signals from hotspots, I have discovered not all hotspots are hot and fed by deep active upwellings; some are cold and require different dynamical mechanisms from those provided by classical plume theory. I have further explored the spatial pattern of global ridge temperature in Chapter 3, and showed the ridge temperature contains fingerprints of past mantle convection and plate tectonics, which can be used to predict the geographic distribution of ridges and disentangle the contribution of deep and shallow processes to temperature and geophysical, geochemical and geological variations observed at ridges.

Chapter 2 and 3 serve to provide an observational lens with which I investigate the dynamical, morphological, and chemical relationship between the LLSVPs and plumes with laboratory experiments in later chapters. The working fluid (corn syrup) is heated from below, and cooled from the top, with Earth-like convective vigor. Among different geneses, I focus on two end-members of the LLSVPs: purely thermal plume clusters and an undeformable, fixed pile, or essentially Rayleigh-Benard convection without or with a 3-D-printed obstacle. Utilizing 4-D velocity acquisition 4, Lagrangian analysis 5, and adjoint reconstructions 7, I have quantitatively analyzed the flow in unprecedented detail (Chapters 6 and 8). In the simple Newtonian working fluid we use for our experiments (corn syrup with only has temperature-dependent viscosity, I have identified a diverse array of plume behaviors that could explain a range of observations from neighboring hotspot tracks to the tree-structure of the LLSVPs and seismic velocity local minima in the lowermost mantle (cf. Chapter 6). Plumes do not seem to have site preference in the purely thermal case, while strong and stable plumes tend to rise around and from the rigid LLSVP (cf. Chapter 8). With a thorough material entrainment analysis of the two end-member experiments at different proximity to the LLSVP, the physical locations of reservoirs linked to the geochemical signals seen at hotspots are also inferred using the geographical (dis)correlation of these hotspots and the LLSVPs. This framework leverages the infinite resolution and unaltered physics of real-world experiments, offering a thorough workflow for 4-D viscous flow measurement that could be extended to other terrestrial planets and beyond. My results also open the door to quantitative comparison of laboratory experiments and numerical simulations of geodynamical time-dependent, multi-scale flow.