A HOT MANTLE WEDGE AND BACKARC AT THE CASCADIA SUBDUCTION ZONE: NUMERICAL TESTS USING A MODEL OF SLAB-DRIVEN WEDGE FLOW

Heat flow, seismic velocities and other observations indicate that the mantle wedge above a subducting plate is anomalously hot, both beneath the volcanic arc and into the backarc, even where there is no extension. Temperatures of at least 1300°C are required below the arc for magma generation. For 100's of kilometres into the backarc, estimated temperatures are ~1200°C at 60 km depth.

We have used finite element thermal models to investigate the backarc advective-convective regime that maintains these high temperatures in spite of the cooling effect of a subducting slab. There are two types of flow models: flow driven by traction along the top of the downgoing plate and thermal buoyancy driven flow. In both cases, the observations require that flow must originate from great depths along the landward boundary of the backarc. In this study, we deal with traction-driven flow, using Cascadia subduction zone parameters.

For an isoviscous mantle, upward flow into the wedge can be generated by introducing a thick (>200 km) lithosphere at landward backarc boundary, consistent with the presence of a craton root. Flow originates from below the craton, producing a strong upward flow at the seaward edge of the craton. This results in nearly uniform temperatures, although the wedge temperatures are 100-200°C lower than inferred from observations.

With a stress- and temperature-dependent viscosity in the model, high velocity flow originates from great depths along the landward boundary, even without a thick craton. Flow is strongly focussed into the wedge corner, leading to much higher temperatures (>1250°C) beneath the arc, in better agreement with observations. However, the flow pattern results in extremely low backarc mantle temperatures and heat flow, inconsistent with observations.

Our results suggest that traction-driven flow alone is inefficient at transporting heat from depth to produce a uniformly hot mantle wedge. In our models, thermal convection has been neglected. Buoyancy-driven flow becomes significant if the wedge viscosity is low (<1020 Pa s), due to the presence of aqueous fluids and/or partial melt. Whereas the velocity of traction-driven wedge flow is limited by the subduction rate, thermal convection can produce very high flow rates, leading to much more efficient heat transfer.