Numerical modeling of subduction zones with slab breakoff : application to adakite genesis
Résumé
Adakites are generally considered as geodynamical markers of partial melting of young subducted oceanic crust (<20 Ma). However, many adakites are now described in arcs below which the subducting oceanic lithosphere is rather old and cold. In some localities (Aleutian-Kamchatka junction, Costa Rica-Panama), adakites seem to be associated with the subduction of plate edges which may result from slab breakoff. Some authors argue that the mantle flow around slab edges heats the slab up to the melting temperature generating adakites. In this study, we use a numerical model to test if the opening of a slab window can generate important melting of slab edges. The numerical model uses a 2D finite element approximation for both temperature and velocity field. The geometry of the subduction zone is mapped using a Delaunay triangulation. At each time, the mesh is updated near the slab edges in order to provide an accurate estimate of the amount of melting in this zone. The 2D model is roughly 2000 km width and 1000 km high and is composed of three distinct domains: an upper fixed lithosphere, a convecting mantle and an oceanic lithosphere which is forced to subduct under the fixed lithosphere near the middle of the box. When the slab plunges into the mantle, partial melting occurs because the mantle is still very hot. As subduction carries on, the mantle wedge is slowly cooled. Finally, slab breakoff is imposed at a given depth. Important heating is observed in the slab edges shortly after breakoff (less than 100 000 yr) but melting is not always predicted to occur mainly because the mantle wedge has been cooled during the first stages of the subduction. Conditions required for melting are currently investigated.