Modeling the Migration of Fluids in Subduction Zones

Fluids play a major role in the formation of arc volcanism and the generation of continental crust. Progressive dehydration reactions in the downgoing slab release fluids to the hot overlying mantle wedge, causing flux melting and the migration of melts to the volcanic front. While the qualitative concept is well established, the quantitative details of fluid release and especially that of fluid migration and generation of hydrous melting in the wedge is still poorly understood. Here we present new models of the fluid migration through the mantle wedge for subduction zones. We use an existing set of high resolution metamorphic models (van Keken et al, 2010) to predict the regions of water release from the sediments, upper and lower crust, and upper most mantle. We use this water flux as input for the fluid migration calculation based on new finite element models built on advanced computational libraries (FEniCS/PETSc) for efficient and flexible solution of coupled multi-physics problems. The first generation of one-way coupled models solves for the evolution of porosity and fluid-pressure/flux throughout the slab and wedge given solid flow, viscosity and thermal fields from separate solutions to the incompressible Stokes and energy equations in the mantle wedge. These solutions are verified by comparing to previous benchmark studies (van Keken et al, 2008) and global suites of thermal subduction models (Syracuse et al, 2010). Fluid flow depends on both permeability and the rheology of the slab-wedge system as interaction with rheological variability can induce additional pressure gradients that affect the fluid flow pathways. These non-linearities have been shown to explain laboratory-scale observations of melt band orientation in labratory experiments and numerical simulations of melt localization in shear bands (Katz et al 2006). Our second generation of models dispense with the pre-calculation of incompressible mantle flow and fully couple the now compressible system of mantle and fluid flow equations, introducing complex feedbacks between the rheology, temperature, permeability, strain rate and porosity. Using idealized subduction zone geometries we investigate the effects of this non-linearity and explore the sensitivity of fluid flow paths for a range of fluid flow parameters with emphasis on variability of the location of the volcanic arc with respect to flow paths. We also estimate the expected degrees of hydrous melting using a variety of wet-melting parameterizations (e.g., Katz et al, 2003, Kelley et al, 2010). The current models only include dehydration reactions but work continues on the next generation of models which will include both dehydration and hydration reactions as well as parameterized flux melting in a consistent reactive-flow framework.