Weakening of the subduction interface and its effects on surface heat flow, slab dehydration, and mantle wedge serpentinization

Ikuko Wada, Kelin Wang, Jiangheng He, Roy D. Hyndman

Research output: Contribution to journalArticle

150 Scopus citations

Abstract

The shallow part of the interface between the subducting slab and the overriding mantle wedge is evidently weakened by the presence of hydrous minerals and high fluid pressure. We use a two-dimensional finite element model, with a thin layer of uniform viscosity along the slab surface to represent the strength of the interface and a dislocation-creep rheology for the mantle wedge, to investigate the effect of this interface "decoupling." Decoupling occurs when the temperature-dependent viscous strength of the mantle wedge is greater than that of the interface layer. We find that the maximum depth of decoupling is the key to most primary thermal and petrological processes in subduction zone forearcs. The forearc mantle wedge above a weakened subduction interface always becomes stagnant (<0.2% slab velocity), providing a stable thermal environment for the formation of serpentinite. The degree of mantle wedge serpentinization depends on the availability of aqueous fluids from slab dehydration. A very young and warm slab releases most of its bound H2O in the forearc, leading to a high degree of mantle wedge serpentinization. A very old and cold slab retains most of its H2O until farther landward, leading to a lower degree of serpentinization. Our preferred model for northern Cascadia has a maximum decoupling depth of about 70-80 km, which provides a good fit to surface heat flow data, predicts conditions for a high degree of serpentinization of the forearc mantle wedge, and is consistent with the observed shallow intraslab seismicity and low volume of arc volcanism.

Original languageEnglish (US)
Article numberB04402
JournalJournal of Geophysical Research: Solid Earth
Volume113
Issue number4
DOIs
StatePublished - Apr 4 2008
Externally publishedYes

    Fingerprint

Cite this