Subduction zone anisotropy beneath Corvallis, Oregon: A serpentinite skid mark of trench-parallel terrane migration?

[1] We studied the back azimuth dependence of Ps converted phases at GSN station COR (Corvallis, Oregon) using broadband P receiver functions computed from 602 teleseismic earthquakes. The amplitudes and polarities of the transverse Ps phases are largely two-lobed, which indicates anisotropy with a tilted symmetry axis. A double-peaked Ps conversion at 4.5–6.5 s delay has the moveout of a dipping slab but is not consistent with simple deflection of the Ps converted phase by the dipping interface. A polarity flip on the transverse receiver function near north-south back azimuth indicates an anisotropic symmetry axis aligned north-south, far from the convergence direction (N68 E). The Ps phase is modeled using reflectivity synthetics with a highly anisotropic layer of depressed wave speed (VP 6.0–6.5 km/s) near 40 km depth, at the slab interface with the overriding North American plate. One-dimensional modeling suggests 10% anisotropy in the supraslab layer, with a slow symmetry axis oriented N5 W at a 60 tilt from the vertical. Adjustments for the effect of slab interface dip on Ps amplitude suggest a somewhat lower 7% anisotropy. We infer a thin ( 7 km) anisotropic detachment zone for the northward slippage of the Siletz forearc terrane along the top of the descending slab, a motion consistent with GPS measurements and models of regional lithospheric dynamics. Serpentinite is a likely constituent for the deep anisotropic layer, owing to its reduced VP and high Poisson ratio. The ductile rheology and hydrated composition of serpentinite make it a plausible lithology for a mechanical detachment zone.