DI34A-01
Thermo-chemical plumes rooted in the deep mantle beneath major hotspots: implications for mantle dynamics

Wednesday, 16 December 2015: 16:00
301 (Moscone South)
Barbara A Romanowicz, Institut de Physique du Globe, Paris, France; Berkeley Seismological Lab, Berkeley, CA, United States and Scott French, NERSC, Oakland, CA, United States
Abstract:
The existence of mantle plumes as a possible origin for hotspots has been the subject of debate for the last 30 years. Many seismic tomographic studies have hinted at the presence of plume-like features in the lower mantle, but resolution of narrow low velocity features is difficult, and ambiguity remains as to the vertical continuity of these features and how distinct they are from other low velocity blobs.

We present robust evidence for significant, vertically continuous, low velocity columns in the lower mantle beneath prominent hotspots located within the footprint of the large low shear velocity provinces (LLSVPs), from a recent global, radially anisotropic whole mantle shear-wave velocity (Vs) model, SEMUCB-WM1 (French and Romanowicz, 2014, 2015). This model was constructed by inversion of a large dataset of long period three-component seismograms down to 32s period. Because it includes surface-wave overtones, S-diffracted waves and multiply reflected waves between the surface and the CMB, this dataset provides considerably better illumination of the whole mantle volume than can be obtained with a standard set of travel times alone. In addition, accurate numerical computation of the forward wavefield using the spectral element method at each iteration of the model construction, allows us to better resolve regions of lower than average Vs.

The imaged plumes have several common characteristics: they are rooted in patches of very low Vs near the core mantle boundary, some of which contain documented ULVZs, and extend vertically through the lower mantle up to ~1000 km depth, where some are deflected horizontally, or give rise to somewhat thinner conduits that meander through the upper mantle in the vicinity of the target hotpots. Combined with evidence for slab stagnation at ~1000 km depth, this suggests a change in rheology between 660 and 1000 km depth, very high viscosity throughout the bulk of the lower mantle, and lower viscosity plumes, only mildly deflected by secondary scale flow in the upper mantle. The lower mantle plume conduits must be at least 500-600 km wide: their morphology resembles that of thermo-chemical plumes rooted in a denser than average basal layer. No such fat plumes are found under hotspots located above the ring of fast velocities surrounding the LLSVPs.