Open Session - SWISS GEOSCIENCE MEETINGs

18
Symposium 1: Structural Geology, Tectonics and Geodynamics
1.
Models of thermo-chemical convection: what is needed to fit probabilistic
tomography?
Deschamps Frédéric & Tackley Paul J.
Institute of Geophysics, ETH Zurich, Switzerland (deschamps@erdw.ethz.ch)
A growing amount of seismological observations indicate that strong lateral density anomalies, likely due to compositional
anomalies, are present in the deep mantle. We aim to identify models of thermo-chemical convection that can generate
strong thermo-chemical density anomalies in the lower mantle – particularly at its bottom –, and maintain them for a long
period of time. For this, we explore the model space of thermo-chemical convection, determine the thermal and chemical
density distributions predicted by these models, and compare their power spectra against those from probabilistic tomography.
We run 3D-Cartesian numerical experiments using STAG3D, in which we varied compositional (buoyancy ratio, fraction
of dense material), physical (Clapeyron slope of the phase change at 660 km, internal heating), and viscosity (thermal, radial,
and compositional viscosity contrasts) parameters, and identified five important ingredients for a successful (in the sense
that it fits seismological observations well) model of thermo-chemical convection. (1) A reasonable buoyancy ratio, between
0.15 and 0.25 (corresponding to chemical density contrasts in the range 60-100 kg/m 3 ). Larger density contrasts induce stable
layering for long period of time, rather than the strong topography required by seismic observations. (2) A moderate (typically
in the range 0.1-10), chemical viscosity contrast. Small chemical viscosity contrasts induce rapid mixing, whereas large
chemical viscosity contrasts lead to stable layering. (3) A large (≥ 10 4 ), thermal viscosity contrast. Temperarure-dependent
viscosity creates and maintains pools of dense material with large topography at the bottom of the mantle. (4) A 660-km
viscosity contrast around 30. (5) And a Clapeyron slope of the phase transition at 660-km around 1.5-3.0 MPa/K. These two last
ingredients help to maintain dense material in the lower mantle. Interestingly, they strongly inhibit thermal plumes, but
still allow the penetration of downwellings in the lower mantle. Finally, we test models that include various combinations
of the previous ingredients. The power spectra of these models are in excellent agreement with those from probabilistic tomography
except for the spectra of chemical anomalies in the layer located right below the 660-km boundary. This discrepancy
might be related to the stacking of slabs at these depths. Because our treatment of the chemical field does not specifically
account for the compositional differences between the descending slabs and the regular mantle, our models cannot
reproduce the chemical signal due to the slab stacking. This will be fixed in future works, which will also include calculations
in spherical geometry and the effects of the post-perovskite phase transition.
1.8
Direct numerical simulation of two-phase flow: homogenization and
collective behaviour in suspensions
Deubelbeiss Yolanda*,**, Kaus Boris J.P.**,***, Connolly James*
*Institute for Mineralogy and Petrology, ETH Zurich, Clausiusstr. 25, CH-8092 Zurich (yolanda.deubelbeiss@erdw.ethz.ch)
**Institute for Geophysics, ETH Zurich, Schafmattstr. 30, CH-8093 Zurich
***Department of Earth Sciences, University of Southern California, 3651 Trousdale Parkway, Los Angeles, CA 90089-740, USA.
Melt transport mechanism has received much attention recently, since melting and melt migration play dominant roles in
heat, composition and mass budget of the Earth. Whereas many aspects of melt migration processes are well known the
physics of the processes remain a matter of discussion.
Melt migration and partial melting processes are generally solved by coupling solid and fluid flow. The theory of compaction-driven
two-phase flow is based on macroscopic models considering the motion of a low-viscosity fluid moving through
a high-viscosity, permeable and deformable matrix (e.g. McKenzie 1984). Previous work concerning modeling of two-phase
flow has been performed by different authors. Nevertheless, models commonly are solved by neglecting and simplifying
parts of the whole complex dynamical system such as the mechanical deformation of the solid, melting processes or the
porosity treatment of the rock. Therefore many numerical codes use very simplified models. However, by reducing the number
of unknowns and assumptions it is possible to study the dynamics of partially molten systems.