Physical model of source region of subduction zone volcanics

The thermal structure of a generic subduction zone is investigated to elucidate the source region of subduction zone volcanics. The steady state thermal field is evaluated for a model subduction zone where the plates are prescribed by kinematic boundary conditions, such that the subducting slab induces a flow in the mantle wedge. The resulting model suggests that the oceanic crust of the downgoing slab is not melted extensively, if at all, and hence is not the source of subduction zone magmatism (with the possible exception of the special case of very young oceanic crust). The temperature in the mantle wedge is high enough to produce melting at the amphibole‐buffered peridotite solidus. It is proposed that the combination of vertical motion of water as a free phase and the transport of hydrous phases (e.g., amphiboles) by the slab‐induced mantle wedge flow lead to the net transport of water being horizontal, across the mantle wedge from the slab. Provided the subducting oceanic crust enters the asthenosphere at a velocity > 6(±2) cm/yr, the mantle wedge will be hot enough at the limit of the lateral water transport mechanism to generate melting at the amphibole‐buffered solidus. The model was then extended to include the effect of localized sources of buoyancy (melt, residue, etc.) as a stationary body force, to investigate the possibility of reversing the slab‐induced flow. Best estimates of the buoyancy sources and appropriate viscosity in the wedge suggest that there is likely to be only a weak modulation of the slab‐induced flow unless the slab and wedge are locally decoupled, for instance by shear heating, the presence of water, or dehydration/hydration reactions. If there is decoupling, then it is possible for there to be an appreciable reversal of the slab‐induced flow. Such an appreciable reversal of flow, if it persists, leads to cooling of the mantle wedge. Hence flow reversal cannot be a steady state mechanism. Instead it would lead to a cycle in the melting with a period of O(1 m.y.). The time dependence of a model with appreciable flow reversal would be reinforced by the need to clear the wedge of infertile material.