Thermo-chemical plume interactions with mantle viscosity layering and phase transformations: comparisons with seismic imaging

We model thermo-chemical mantle plumes, interacting with the Earth's mantle mineralogical and viscosity stratification and also with a mobile lithosphere, following Farnetani and Samuel (2005).

We compare our models to recent seismic imaging results that suggest a possible mantle interface/transition at ~1000 km-depth, in which: (a) low-velocity anomalies (hot-upwelling mantle plumes?) express quasi-vertically from the core-mantle boundary (CMB) up to ~1000 km-depth (e.g. French and Romanowicz , 2015; Nelson and Grand , 2018), above which they are significantly deflected (advection by upper-mantle flow?), and (b) some cold-downwelling slabs appear to pond at about ~1000 km depth (e.g. Fukao and Obayashi, 2013), for subduction zones of the Pacific and Indian oceans. Modeling work combining long-wavelength geoid and mineral physics ( Rudolph et al. , 2015) suggested that these observations could be due to a viscosity increase with depth in the interval 800-1200 km.

Mantle plumes may provide strong upward buoyancy forces within the convecting mantle. We pay attention to the transport capabilities of a generic upwelling plume, when interacting with a stratified ambient mantle which in turn contains chemical heterogeneities: We track material lying at the base of the mantle and featuring a compositional/mineralogical density anomaly, as it is carried upward by the upwelling thermal plume - the thermochemical problem, addressed numerically by Dannberg and Sobolev (2015), Samuel and Bercovici (2006), Lin and van Keken (2002); and in laboratory experiments by Davaille (1999). In turn, this upwelling material is subject to advection by the circulating mantle (due largely to plate motions), interacts with a viscosity interface at 1000 km-depth, and with an endothermic phase change at 660 km-depth.

To study this problem, we are conducting 3D, time-dependent numerical simulations with the software ASPECT.