I use global and regional mantle convection models, constrained by data assimilation, to connect deep-mantle processes with what we observe at the surface. A common thread runs through these projects: reconstructing how subducted slabs sink, stagnate, and interact with plumes — and how that history is written into lithospheric structure, magmatism, seismic anisotropy, and topography.
I build global thermal–chemical convection models that assimilate plate-motion histories, so the simulated present-day mantle reproduces seismic tomography. This lets me follow the evolution of large low-shear-velocity provinces (LLSVPs), the interplay between rising plumes and sinking slabs, and the surface topography they generate — each tested against independent geological and geophysical observations.
I identified a continental-scale flat Izanagi slab beneath East Asia during the Late Cretaceous. A single, long-lived episode of flat subduction ties together a set of otherwise puzzling regional features — the lithospheric structure, the magmatic record, and the pattern of surface uplift and subsidence — and most likely formed through prolonged, shallow-angle subduction.
The Pacific and Philippine Sea slabs stagnate within the mantle transition zone beneath East Asia. I proposed that a regional, pressure-driven westward "mantle wind" sustains this stagnation, and the seismic anisotropy predicted by the modeled flow matches observations. I also wrote about it for the EGU Geodynamics blog: Why do some slabs stagnate?
The Tonga–Kermadec zone has the fastest trench retreat on Earth, which conventional models struggle to reproduce. Using a nonlinear-rheology boundary condition, I recovered both the observed slab geometry and the opening of the Lau back-arc basin, linking the basin's formation to eastward migration of the Lau Ridge.
Each project above links to its key paper. For full method details and further work — across these and other topics — see my complete publication list or my Google Scholar profile.