Shear Velocity Structure From Ambient Noise and Teleseismic Surface Wave Tomography in the Cascades Around Mount St. Helens

Abstract

Mount St. Helens (MSH) lies in the forearc of the Cascades where conditions should be too cold for volcanism. To better understand thermal conditions and magma pathways beneath MSH, data from a dense broadband array are used to produce high‐resolution tomographic images of the crust and upper mantle. Rayleigh‐wave phase‐velocity maps and three‐dimensional images of shear velocity (Vs), generated from ambient noise and earthquake surface waves, show that west of MSH the middle‐lower crust is anomalously fast (3.95 ± 0.1 km/s), overlying an anomalously slow uppermost mantle (4.0–4.2 km/s). This combination renders the forearc Moho weak to invisible, with crustal velocity variations being a primary cause; fast crust is necessary to explain the absent Moho. Comparison with predicted rock velocities indicates that the fast crust likely consists of gabbros and basalts of the Siletzia terrane, an accreted oceanic plateau. East of MSH where magmatism is abundant, middle‐lower crust Vs is low (3.45–3.6 km/s), consistent with hot and potentially partly molten crust of more intermediate to felsic composition. This crust overlies mantle with more typical wave speeds, producing a strong Moho. The sharp boundary in crust and mantle Vs within a few kilometers of the MSH edifice correlates with a sharp boundary from low heat flow in the forearc to high arc heat flow and demonstrates that the crustal terrane boundary here couples with thermal structure to focus lateral melt transport from the lower crust westward to arc volcanoes.

Amphibious surface-wave phase-velocity measurements of the Cascadia subduction zone

“…Thus, the largest differences are likely in part due to differences in data coverage, which was absent from previous studies using similar methodologies (e.g. Accardo et al 2017;Crosbie 2018). Furthermore, the 20-s-period phase-velocity maps for both the ambient-noise and Helmholtz results have higher average uncertainties relative to their adjacent periods.…”

“…Interstation phase-velocity measurements, expressed as slowness perturbations from a reference average velocity, are weighted by errors estimated from the dispersion-curve inversion. The prior error is calculated as a standard estimate of data uncertainty from the residuals (Crosbie 2018). We use a constant phase velocity calculated from the weighted average from all station pairs at a given period as the starting model; however, we test a range of constant initial velocities (± 0.5 km s −1 ) and observe no significant effects on the final result.…”

Imaging Subduction Beneath Mount St. Helens: Implications for Slab Dehydration and Magma Transport

“…First, data are low-pass filtered at 0.2 Hz to better isolate coherent phases such as slab conversions. The back-projections use a reference 1-D S velocity model derived from an inversion of surface waves from ambient noise and earthquakes (supporting information Figure S1; Crosbie, 2018), estimating crustal Vp/Vs and density in the upper 40 km from Brocher (2005) and mantle Vp/Vs of 1.75. To better visualize the structures responsible for those phases, we plot them with depth in a reference 1-D velocity model (e.g., Figures 2a and 2b), assuming that dominant conversions are the Ppxs phase (free-surface-reflected P converting to S at the scatterer; Figure 2b) due to its high-depth resolution (Rondenay, 2009) and low noise (Pearce et al, 2012).…”

“…To better visualize the structures responsible for those phases, we plot them with depth in a reference 1-D velocity model (e.g., Figures 2a and 2b), assuming that dominant conversions are the Ppxs phase (free-surface-reflected P converting to S at the scatterer; Figure 2b) due to its high-depth resolution (Rondenay, 2009) and low noise (Pearce et al, 2012). The back-projections use a reference 1-D S velocity model derived from an inversion of surface waves from ambient noise and earthquakes (supporting information Figure S1; Crosbie, 2018), estimating crustal Vp/Vs and density in the upper 40 km from Brocher (2005) and mantle Vp/Vs of 1.75. These analyses result in 61 earthquakes after eliminating earthquakes producing ringy or inconsistent RFs.…”

“…The root-mean-square (RMS) variation in depth between the three sections is ≤1 km (Figure 4e) between 49-and 72-km depth. We estimate an additional 2-km uncertainty due to the assumed velocity model, taken here as the RMS variation in travel time through the 3-D model of Crosbie (2018). This gives an aggregate uncertainty in slab depth of 2 km.…”