Fluid and mass transfer into the cold mantle wedge of subduction zones: budgets and seismic constraints

Tuesday, 15 December 2015: 10:20
303 (Moscone South)
Geoffrey A Abers, Cornell University, Ithaca, NY, United States, Bradley R Hacker, University of California Santa Barbara, Santa Barbara, CA, United States, Peter E Van Keken, University of Michigan Ann Arbor, Ann Arbor, MI, United States, Junichi Nakajima, Tokyo Institute of Technology, Tokyo, Japan and Saeko Kita, National Research Institute for Earth Science, Tsukuba, Japan
Dehydration of subducting plates should hydrate the shallow overlying mantle wedge where mantle is cold. In the shallow mantle wedge hydrous phases, notably serpentines, chlorite, brucite and talc should be stable to form a significant reservoir for H2O. Beneath this cold nose thermal models suggest only limited slab dehydration occurs at depths less than ca. 80 km except in warm subduction zones, but fluids may flow updip from deeper within the subducting plate to hydrate the shallow mantle. We estimate the total water storage capacity in cold noses, at temperatures where hydrous phases are stable, to be roughly 2-3% the mass of the global ocean. At modern subduction flux rates its full hydration could be achieved in 50-100 Ma if all subducting water devolatilized in the upper 100 km flows into the wedge; these estimates have at least a factor of two uncertainty.

To investigate the extent to which wedge hydration actually occurs we compile and generate seismic images of forearc mantle regions. The compilation includes P- and S-velocity images with good sampling below the Moho and above the downgoing slab in forearcs, from active-source imaging, local earthquake tomography and receiver functions, while avoiding areas of complex tectonics. Well-resolved images exist for Cascadia, Alaska, the Andes, Central America, North Island New Zealand, and Japan. We compare the observed velocities to those predicted from thermal-petrologic models. Among these forearcs, Cascadia stands out as having upper-mantle seismic velocities lower than overriding crust, consistent with high (>50%) hydration. Most other forearcs show Vp close to 8.0 km/s and Vp/Vs of 1.73-1.80. We compare these observations to velocities predicted from thermal-mineralogical models. Velocities are slightly slower than expected for dry peridotite and allow 10-20% hydration, but also could also be explained as relict accreted rock, or delaminated, relaminated, or offscraped crustal material mixed with mantle. The absence of wholesale hydration of forearcs globally can be taken as evidence that most forearcs are too young to be substantially hydrated, that most subducted water bypasses the forearc and is released deeper, or that most fluid passing through the mantle nose does not react with the mantle.