We present new seismic tomography of the Washington forearc using a suite of manually picked regional earthquake phase arrivals. The recovery of similarly sampled P and S velocity permits the robust calculation of Poisson's ratio throughout the region. The seismological signature of Siletzia, an accreted oceanic plateau that crops out in Washington as the Crescent Formation, is evident in our models as a continuous high Poisson's ratio body that coincides with subsurface structures estimated from potential field maps. Relocated earthquakes preferentially locate in low Poisson's ratio regions in the forearc crust and, in particular, in a diffuse layer located at 15-25 km depth in the crust beneath relatively aseismic Siletzia. Our imaging of the Puget Sound is consistent with previous interpretations of the architecture of major faults and blocks, as evident by distinct Poisson's ratio signatures that distinguish sedimentary basins from mafic rocks of the Crescent Formation. We speculate that seismicity below the Puget Sound is promoted by slab-derived fluids that are localized beneath Siletzia as a result of intrinsic low vertical permeability.

The northern third of the Cascadia subduction zone is marked by a change in margin orientation near 48°N in Washington state that extends to the northern half of Vancouver Island. Here, the Juan de Fuca (JdF) and Explorer plates converge with North America at rates near 45 and 20 mm/yr, respectively (DeMets et al., 2010;McCaffrey et al., 2007). Convergence is manifest by seismicity both within the descending oceanic plate(s) and overriding crust, as well as near the JdF plate interface in the form of low frequency earthquakes (LFEs) and tectonic tremor (Bostock et al., 2019;Wech, 2010). Megathrust events (M > 9) originate along the shallow (<20 km) segment of the JdF plate interface but have not been recorded by modern instruments in Cascadia. Several lines of evidence suggest that they occur with an average repeat time around 600 years, with the most recent one having occurred

We investigate an isolated cluster of temporally persistent, intraslab earthquakes (ML<3.2) at >60  km depth below the Georgia Strait in southern British Columbia that is unique in Cascadia and meets the criteria for identification as an earthquake nest. A total of 129 relocated hypocenters define two northwest‐dipping structures in the subducting Juan de Fuca mantle within an ∼30×10×10  km3 volume. Focal mechanisms for 15 events represent a mix of strike‐slip and reverse faulting, and a stress regime of down‐dip tension and plate‐normal compression, consistent with a previous regional study. Converted seismic phases inferred to originate at the boundaries of subducted oceanic crust are observed at several receivers and are consistent with a local slab depth of ∼45  km, shallower than some JdF plate models. The geographical isolation of the nest within the confines of an extrapolated propagator wake suggests that its location is controlled by this pre‐existing and presumably hydrated structure.

No abstract

We propose new methods for assessing temporal changes in seismic velocity using the S-wave coda for repeating earthquakes and cross-correlation functions of ambient noise. For a pair of seismic waveforms representing a common source–receiver path, the relative change in path-averaged velocity over the corresponding time interval is directly proportional to the factor by which one waveform needs to be stretched or compressed with respect to the other to achieve maximum coherence. For an arbitrary number of waveforms, initial pair-wise stretch factors determined through standard approaches can be improved through solution of an overdetermined system and further refined through an iterative approach exploiting the singular value decomposition to minimize rank of the stretched waveform section. We apply this combined approach to both repeating earthquakes and ambient noise correlations for Haida Gwaii in western Canada, the site of a Mw 7.8 thrust earthquake in 2012. Optimal stretch factors for repeating earthquake families indicate that path-averaged S velocities dropped by up to 0.16% after the earthquake. Ambient noise correlations indicate that velocities dropped by between 0.26% and 0.39%, which we interpret to be more pronounced in the uppermost levels of the crust. We explore these results in terms of changes in crustal porosity and hydrogeologic conditions by considering the observation that hot spring activity on Haida Gwaii ceased following the 2012 mainshock and recovered over the next several years.