Of Mantle Plumes, Their Existence, and Their Nature: Insights from Whole Mantle SEM-Based Seismic Waveform Tomography

Abstract:

Many questions remain on the detailed morphology of mantle convection patterns. While high resolution P wave studies show a variety of subducted slab behaviors, some stagnating in the transition zone, others penetrating into the lower mantle (e.g. Fukao & Obayashi, 2013), low velocity structures – the upwelling part of flow – are more difficult to resolve at the same scale. Indeed, depth extent and morphology of the low velocity roots of hotspot volcanoes is still debated, along with the existence of "mantle plumes".

Using spectral element waveform tomography, we previously constructed a global, radially anisotropic, upper mantle Vs model (SEMum2, French et al., 2013) and have now extended it to the whole mantle by adding shorter period waveform data (SEMUCB-WM1, French & Romanowicz, GJI, in revision). This model shows long wavelength structure in good agreement with other recent global Vs models derived under stronger approximations (Ritsema et al. 2011; Kustowski, et al. 2008), but exhibits better focused, finer scale structure throughout the mantle.

SEMUCB-WM1 confirms the presence in all major ocean basins of the quasi-periodic, upper mantle low velocity anomalies, previously seen in SEMum2. At the same time, lower mantle low velocity structure is dominated by a small number (~15 globally) of quasi-vertical anomalies forming discrete "column"" rooted at the base of the mantle. Most columns are positioned near major hotspots, as defined by buoyancy flux, and are wider (~800-1000 km diameter) than expected from the thermal plume model – suggestive of thermo-chemical plumes, which may be stable for long times compared to purely thermal ones.

Some columns reach the upper mantle, while others deflect horizontally near 1000 km – the same depth where many slabs appear to stagnate. As they reach the transition zone, the wide columnar structure can be lost, as these "plumes" appear to meander through the upper mantle, perhaps entrained by more vigorous, lower viscosity, convection. Most "plumes" in the Pacific LLSVP region appear as isolated columns rising from the CMB, such as beneath Hawaii (rooted near a known ultra low velocity zone, Cottaar & Romanowicz, 2012). Conversely, the African LLSVP region appears more massive up to mid-mantle depths, with isolated "plumes" at its borders, including that beneath Iceland.