Résumé
Lateral limits of tectonic plates are well defined by a high concentration of seismicity and volcanism, but the bottom plate boundary, i.e. the lithosphere-asthenosphere boundary (LAB) remains elusive. The LAB occurs within the peridotitic mantle. It is not associated with neither a chemical nor a mineralogical discontinuity. The LAB is defined as a rheological transition between a high strength lithospheric plate, which displays both brittle and ductile behaviors, and a lower strength ductile asthenosphere. Yet at the LAB upper mantle rocks show clearly a ductile behavior, deforming by an association of dislocation creep and diffusion creep. Since both processes depend exponentially on temperature, the LAB is commonly defined as an isotherm varying from 1100 to 1400 Celsius degree. The effect of strain rate on viscosity, which should be significant if deformation by dislocation creep dominates, is usually neglected. Numerical experiments of deformation at the crystal scale have shown that deformation of olivine, at temperatures that characterize the lower part of the lithospheric plates and the asthenosphere, occurs through the same dislocation dynamics in olivine (glide and climb), the main upper mantle mineral. Based on these results, we parameterized an effective viscosity depending on temperature and strain rate for mantle rocks. We implemented this new rheology in numerical models of plate-driven mantle flow, in which the LAB naturally arises in response to the temperature and strain rate fields. We find that the temperature of the rheological transition between lithosphere and asthenophere is not constant, but depends on the strain rate. These results suggest a new definition for the LAB, which location partly depends on the velocity gradient between surface plates and asthenospheric mantle flow. We also test the new rheology in subduction models : the single parameterisation for both low- and high-temperature upper mantle mechanical behavior accounts well for both slab deformation and asthenophere dynamics.