Abstract

Subduction zones are regions where cool oceanic crust and lithosphere sink into the warm interior of the earth. It seems strange that we find melts above subduction zones when the mantle here is cooler than average. The current theory is that water is incorporated into the oceanic crust when it is formed at a mid-ocean ridge, and that this water is released into the over-lying mantle when the oceanic crust is subducted. This water then lowers the melting temperature of the mantle and allows melting to occur, even though temperatures are lower than average. In order to locate a possible source region for magmatism it is essential to model the thermal structure of a subduction zone and to understand the role of water. We have developed a two-dimensional kinematic model of subduction and solved for the temperature and velocity field. Using the ideas of Davies and Stevenson (that water is released when the oceanic crust dehydrates and that this water is then carried out into the mantle wedge by amphibole) we located a possible source region for magmatism. Using this standard model, we have investigated the effect on the thermal field and the location of the source region for: (i) stress dependent mantle rheology (Newtonian verses non-Newtonian), (ii) temperature dependent mantle viscosity, (iii) age of the subducting oceanic plate, (iv) thickness of the overriding plate, (v) frictional heat sources on the slab-mantle interface, (v) shape of the subducting slab and (vi) convergence velocity of the subduction zone. We find that the stress dependence of the mantle's rheology makes very little difference to the thermal field, suggesting that the forcing boundary conditions are more important than the rheology. In contrast, making the mantle viscosity temperature dependent has a dramatic effect on both the thermal field and the location of the source region. The age of the subducting slab has very little influence on the thermal field (apart from inside the slab itself). Increasing the thickness of the overriding slab reduces the size of the source region and leads to a cooler wedge. Adding frictional heat sources to the interface increases the interface temperatures, but not enough to melt the oceanic crust. Modelling a slab shape leads to interesting results including a possible explanation as to why we don't observe volcanism above the region of flat subduction in Peru. Reducing the convergence velocity leads to a cooler mantle wedge, as we would expect. Using the results from wet melting experiments and parameterization of melting we used the results from our models to predict the composition of magmas at the surface. We find that all the predicted primary magmas resulting from wet melting of the mantle wedge are basaltic.