Imaging the Plate Interface in the Cascadia Seismogenic Zone: New Constraints from Offshore Receiver Functions

The Cascadia subduction zone, where the Juan de Fuca (JdF) plate subducts beneath North America, has paleoseismic evidence ofMw ∼9:0 megathrust earthquakes (Nelson et al., 1995; Goldfinger et al., 2003). However, there are virtually no instrumentally recorded thrust-zone earthquakes, hence the location and behavior of the seismogenic zone is known only indirectly. Temperature has been proposed to control seismogenesis with depth, assuming that the locked zone extends from the trench or the 150°C isotherm down-dip to the 350°C isotherm, with a transition zone extending to 450°C (Hyndman andWang, 1993; Oleskevich et al., 1999; Cozzens and Spinelli, 2012). These models generally place the down-dip edge of the locked zone near the coastline. Inversions of onshore Global Positioning System data also can be used to determine the locking behavior and place the locked zone offshore (e.g., McCaffrey et al., 2013). Onshore receiver function (RF) studies have imaged an eastward-dipping low-velocity zone (LVZ) with high VP=V S between the coastline and depths of 45 km (Rondenay et al., 2001; Nicholson et al., 2005; Abers et al., 2009; Audet et al., 2009). This structure has been interpreted as overpressured pore fluids, metamorphosed sediments, or a combination thereof at or just above the top of the subducting oceanic crust (Abers et al., 2009; Hansen et al., 2012). Because of this uncertainty, it is unclear if an LVZ should continue up-dip through the locked zone, since fluid pressure and metamorphism should vary differently with depth (e.g., Hacker et al., 2003; Liu and Rice, 2007; Saffer and Tobin, 2011). However, existing RF images only sample the plate boundary deeper than the locked zone because past broadband arrays are on land. Brillon et al. (2013) analyze RFs from two ocean-bottom seismometers (OBSs) offshore of Vancouver Island, but poor data quality at these stations contributes to large uncertainties. Receiver funtions are difficult to calculate from OBS instruments, because water column multiples interfere with other arrivals and noise is high particularly on horizontal components (Leahy et al., 2010; Bostock and Trehu, 2012; Ball et al., 2014). The Cascadia Initiative (CI) is a prime opportunity to revisit this challenge (Toomey et al., 2014). In particular, the new trawlresistant-mount (TRM) OBS design not only allows the instruments to be deployed in shallow water, but also greatly reduces horizontal-component noise (Webb et al., 2013). In this article, we evaluate the ability of all sites of the CI array to calculate RFs, and we focus on results from the 19 OBSs deployed off the coast of Grays Harbor,Washington. These extend the onshore Cascadia Arrays for Earthscope (CAFE) broadband array (Abers et al., 2009) offshore, allowing direct comparisons.