Volume 124, Issue 7 p. 7067-7080
Research Article
Free Access

Upper Crustal Structure and Magmatism in Southwest Washington: Vp, Vs, and Vp/Vs Results From the iMUSH Active-Source Seismic Experiment

E. Kiser

Corresponding Author

E. Kiser

Department of Geosciences, University of Arizona, Tucson, Arizona, USA

Correspondence to: E. Kiser,

[email protected]

Search for more papers by this author
A. Levander

A. Levander

Department of Earth, Environmental and Planetary Sciences, Rice University, Houston, Texas, USA

Search for more papers by this author
C. Zelt

C. Zelt

Department of Earth, Environmental and Planetary Sciences, Rice University, Houston, Texas, USA

Search for more papers by this author
B. Schmandt

B. Schmandt

Department of Earth and Planetary Sciences, University of New Mexico, Albuquerque, New Mexico, USA

Search for more papers by this author
S. Hansen

S. Hansen

Department of Earth and Planetary Sciences, Macquarie University, Sydney, New South Wales, Australia

Search for more papers by this author
First published: 08 July 2019
Citations: 5

Abstract

Structural details of the crust play an important role in controlling the distribution of volcanic activity in arc systems. In southwest Washington, several different regional structures associated with accretion and magmatism have been invoked to explain the broad distribution of Cascade volcanism in this region. In order to image these regional structures in the upper crust, Pg and Sg travel times from the imaging Magma Under St. Helens (iMUSH) active-source seismic experiment are inverted for Vp, Vs, and Vp/Vs models in the region surrounding Mount St. Helens. Several features of these models provide new insights into the regional structure of the upper crust. A large section of the Southern Washington Cascades Conductor is imaged as a low Vp/Vs anomaly that is inferred to represent a broad sedimentary/metasedimentary sequence that composes the upper crust in this region. The accreted terrane Siletzia is imaged west of Mount St. Helens as north/south trending high Vp and Vp/Vs bodies. The Vp/Vs model shows relatively high Vp/Vs regions near Mount St. Helens and the Indian Heaven Volcanic Field, which could be related to the presence of magmatic fluids. Separating these two volcanic regions below 6-km depth is a northeast trending series of high Vp and Vs bodies. These bodies have the same orientation as several volcanic/magmatic features at the surface, including Mount St. Helens and Mount Rainier, and it is argued that these high-velocity features are a regional-scale group of intrusive bodies associated with a crustal weak zone that focuses magma ascent.

Key Points

  • North/south trending high Vp and Vp/Vs anomalies west of Mount St. Helens are inferred to represent the accreted terrane Siletzia
  • A low Vp/Vs anomaly spatially correlates with the Southern Washington Cascades Conductor indicating a sedimentary/metasedimentary origin for this feature
  • Northeast trending high-velocity anomalies aligned with volcanic features are inferred to be Tertiary igneous rocks intruded along a crustal weak zone

1 Introduction

The Cascade arc is a series of Quaternary volcanoes that run parallel to the west coast of the United States and Canada from northern California to southern British Colombia (Figure 1a). The most active volcano in this arc during the past 4,000 years is Mount St. Helens (Mullineaux & Crandell, 1980) with recent activity from 1980 to 1986 and 2004 to 2008 that was initiated on 18 May 1980 with a Plinian eruption that was the most destructive in U.S. history (Lipman & Mullineaux, 1982). Several prominent regional features may account for both the high activity and unique position of Mount St. Helens, which sits 50 km west of the main arc. This volcanic edifice is near the eastern boundary of an accreted terrane commonly referred to as Siletzia. This terrane is a basaltic large igneous province that formed 56-49 Ma and accreted onto the North American Plate around 51-49 Ma (Wells et al., 2014). Seismic profiles indicate that the thickness of this terrane varies significantly along its length parallel to the Cascade arc, from 25 to 35 km in central Oregon to 6 km near Vancouver Island (Brocher et al., 2001; Parsons et al., 1999; Ramachandran et al., 2004; Ramachandran et al., 2005; Trehu et al., 1994). It has been argued that seismicity and volcanism focus near the eastern boundary of this terrane, though the location of this boundary is still debated and may vary with depth (Parsons et al., 1999; Schmandt & Humphreys, 2011; Wells et al., 1998). Several lines of evidence also indicate that this terrane is a continuous body throughout much of Oregon, whereas in southern Washington the terrane seems to be broken into smaller bodies interspersed with sedimentary material (Egbert & Booker, 1993; Wells, 1990).

Details are in the caption following the image
(a) Tectonic overview of the Cascade arc. The red triangles are active volcanoes. The thick red line is the Cascadia trench. The yellow dots are the locations of Corvalis, OR; Portland, OR; and Seattle, WA. The background colors show topography/bathymetry in the region. The black box is the study region shown in (b). GR, Goat Rocks; MA, Mount Adams; MH, Mount Hood; MJ, Mount Jefferson; MR, Mount Rainier, MSH, Mount St. Helens. (b) Regional features associated with Mount St. Helens. The thick black dashed line is the approximate eastern boundary of Siletzia estimated from aeromagnetic data (Wells et al., 1998). The red line is the outer boundary of the SWCC (Egbert & Booker, 1993). The thin black dashed lines are majors rivers in the region. The white diamonds are Quaternary volcanic vents in the region (Hildreth, 2007), and the circles are earthquake locations for earthquakes M≥2 from the Pacific Northwest Seismic Network (PNSN) earthquake catalogue between 1 January 2000 and 1 February 2019 (pnsn.org). The colors of the dots indicate earthquake depth. Other symbols are the same as (a).

Another important regional geophysical feature is the Southern Washington Cascades Conductor (SWCC), a zone of high electrical conductivity primarily extending between Mount St. Helens, Mount Rainier, and Mount Adams (Stanley et al., 1987). This feature was originally attributed to deep marine and/or accretionary prism sedimentary rocks (Stanley et al., 1987; Stanley et al., 1996), possibly emplaced during the collision of Siletzia with North America. An alternative interpretation was given by Hill et al. (2009), which imaged the SWCC during activity at Mount St. Helens between 2004 and 2008. This study showed that the SWCC shallows near the edifices of Mount St. Helens and Mount Adams and concluded that the SWCC is a broad region of partial melt connecting volcanic edifices in the region. Recent 3-D conductivity models from the magnetotelluric (MT) component of the imaging Magma Under St. Helens (iMUSH) project have revealed that high conductivities associated with the SWCC surround a shallow high resistivity body east of Mount St. Helens that has been interpreted as a large batholith (Bedrosian et al., 2018). Though several MT studies have found evidence for the SWCC, previous regional-scale seismic tomographic models found no velocity pattern that matched the unique shape of the SWCC (Moran et al., 1999). A better understanding of the seismic properties of this feature can help answer fundamental questions related to the effect of crustal structure on volcanism and the spatial extents and interactions of magmatic systems in arc settings.

A final striking feature of this part of Cascadia is the broad distribution of Quaternary basalt vents that extend west to east from Portland, OR, in the forearc to the Simcoe Mountains Volcanic Field in the backarc (Figure 1b). A particularly dense cluster of vents that trend north/south, known as the Indian Heaven Volcanic Field, is located in a 300-km2 region southeast of Mount St. Helens. Volcanic output at this series of vents over the past 760 kyr is similar to that at Mount St. Helens (60-80 km3), though they have been dormant for the past 9 kyr (Hildreth, 2007).

The iMUSH experiment was carried out between 2014 and 2016 to image the magmatic system beneath Mount St. Helens from the volcanic edifice down to the subducting Juan de Fuca slab. This experiment included active-source seismic, passive-source seismic, MT, and petrologic data collection efforts (Bedrosian et al., 2018; Hansen et al., 2016; Hansen & Schmandt, 2015; Kiser et al., 2016, 2018; Mann et al., 2019; Wanke et al., 2019). Here we present the first regional-scale 3-D Vp (P-wave velocity), Vs (S-wave velocity), and Vp/Vs (the ratio of P and S wave velocities) models produced from seismic wave travel time data from the active-source seismic component of this project. These seismic velocity models are combined with results from previous geologic and geophysical studies to understand the origin of regional features and their relationships to magmatism in the region.

2 Data

The active-source seismic component of the iMUSH project included two deployments of ~2,500 Texan Reftek 125A data loggers with one-component 4.5-Hz geophones and one deployment of ~900 one-component nodal instruments with 10-Hz geophones (Hansen & Schmandt, 2015). The Texans recorded for 8 to 48 hr, while the nodal instruments recorded for 2 weeks continuously. Twenty-three 450- to 900-kg borehole shots were set off during the experiment, 15 during the first Texan deployment, and 8 during the second Texan deployment (Figure 2). Shots were recorded across the arrays with maximum shot/receiver offsets around 190 km. Data were bandpass filtered between 1 and 32 Hz, and direct P (Pg) and S (Sg) wave travel times were manually picked from all shots (Figures 3 and S1 in the supporting information). Additional seismic phases are observed in the data (e.g., PmP; Figure S1) but are not included in the current study. During the picking process, data are divided into shot gathers based on the azimuth of the receivers from the source. The azimuth ranges of these shot gathers overlap, and therefore, multiple picks are made on each trace. The mean of these picks at each trace is the pick time used in the inversion, and the standard deviation is used to define pick uncertainty. When the estimated uncertainty is below a defined minimum value, the uncertainty is set to the minimum value. This minimum value is set to 0.055 s for P wave picks (one quarter of the dominant period of the data) and 0.222 s for S wave picks (the dominant period of the data). The mean uncertainties of all P and S wave picks are 0.058 and 0.224 s, and the standard deviations of the uncertainties are 0.013 and 0.011 s. In total, there are around 61,700 P wave picks and 46,700 S wave picks. Approximately one quarter of these picks were made on data from the nodal seismometers.

Details are in the caption following the image
The imaging Magma Under St. Helens (iMUSH) active-source seismic experiment. (a) Northeast/southwest deployment of ~3,400 instruments (white dots). These instruments are a combination of Reftek 125A Texans with 4.5-Hz geophones and nodal instruments with 10-Hz geophones. The 15 blue stars are 450- to 900-kg shot locations. All other symbols are the same as Figure 1. (b) Northwest/southeast deployment of ~3,400 instruments (white dots) with eight 450- to 900-kg shot locations (blue stars).
Details are in the caption following the image
Shot gather. (a) Deployment 1 of the imaging Magma Under St. Helens (iMUSH) active-source seismic experiment. The white dots are all seismograph locations, the blue dots are seismograph locations used for the shot gather, the pink star is the shot location, and the red triangles are major volcanic edifices in the region. The background colors show topography. (b) Shot gather from the stations shown in (a). The reducing velocity is 7 km/s. (c) The same as (b) except with P and S wave picks plotted as red and green vertical bars to account for pick uncertainty. Also plotted are the P and S wave arrival times predicted by the final Vp and Vs models (blue and pink dots).

3 Methods

3.1 Vp, Vs, and Vp/Vs Inversion

Pg and Sg travel times are inverted for Vp, Vs, and Vp/Vs using the First Arrival Seismic Tomography (FAST) program, an iterative, regularized, ray-based inversion (Zelt & Barton, 1998). The regularized inversion seeks to minimize an objective function that is the weighted combination of the L2 norm of data misfit and model smoothness. A least squares conjugate gradient method is used to determine model updates. Iterations continue until the normalized chi-squared value is close to one (Figure S3). For all inversions, a 1-km grid spacing is used to calculate travel times (Vidale, 1990) and to define the cells used to update the model parameters.

The starting model for the Vp inversion is the 1-D model used in Waite and Moran (2009); Figure S2). Using this starting model, a normalized chi-squared value of 1.08 is achieved after 30 iterations of the inversion. For each of these iterations, five trial values are used to determine a weight for model smoothness that allows the normalized chi-squared value to decrease. A minimum model smoothness weight is applied to avoid model perturbations at the scale of the grid spacing (Figure S3). Similar results are achieved using different 1-D starting models (Figures S2, S4, and S5).

Sg travel times are inverted using a slightly different procedure. The starting model of this inversion is the final Vp model divided by 1.75, which also produces a uniform starting Vp/Vs model. A modified version of the FAST program is then used to invert for Vs using Sg travel times with an additional spatial smoothness constraint applied to Vp/Vs values (Schmandt & Humphreys, 2010). For this inversion, a normalized chi-squared value of 1.13 is achieved after 20 iterations of the inversion. As with Vp, similar Vs results are achieved using different 1-D starting models (Figures S2, S6, and S7).

3.2 Resolution and Uncertainty

Checkerboard tests are used to estimate the regions of the models that are well resolved (Figures S8 and S9). A semblance value is used to determine the similarity between recovered velocity models and the input checkerboard perturbations (Zelt, 1998). Results from these checkerboard tests indicate that for the Vp model, lateral dimensions of 5, 10, and 15 km can be resolved to maximum depths of 5, 10, and 14 km, respectively (Figures S11-S13). Similar relationships exist for the Vs model (Figures S20-S22). These tests also show that anomalies with depth dimensions of 6 km or less can be resolved throughout much of the center of the model (Figures S16-S18 and S25-S27). For Figures 4-12, regions of the models are shown where checkerboard anomalies of any size can be resolved (Figures S10 and S19). All other regions are masked out. In addition to the checkerboard tests, synthetic tests that include high-amplitude anomalies that mimic the features of the Vp and Vs models are also included in Figures S28S39. These tests show that the data can resolve the basic shapes of the dominant features in our models, though the P wave data set is better for amplitude recovery.

Details are in the caption following the image
Vp model depth slices. The depth of each panel is indicated in the top left corner. The red triangles are volcanic edifices. The contour interval is 0.2 km/s. Labels P1-P11 correspond to features discussed in the text.
Details are in the caption following the image
Vs model depth slices. The depth of each panel is indicated in the top left corner. The red triangles are volcanic edifices. The contour interval is 0.1 km/s. Labels S1-S9 correspond to features discussed in the text.
Details are in the caption following the image
Vp/Vs model depth slices. The depth of each panel is indicated in the top left corner. The red triangles are volcanic edifices. The contour interval is 0.025. Labels R1-R5 correspond to features discussed in the text.
Details are in the caption following the image
Vp model at 1-km depth. The red triangles are Quaternary volcanic edifices. The dashed lines are major rivers in the study area. CB, Chehalis Basin; CP, Cinnamon Peak; MA, Morton Anticline; MM, Marble Mountain; PB, Portland Basin; QV, Quaternary Volcanics; SL, Spirit Lake Pluton; SM, Spud Mountain Pluton; SS, Silver Star Pluton; and WRB, Willamette River Basin. The thick black lines show mapped exposures of Tertiary intrusive rocks within the resolved area.
Details are in the caption following the image
Vp/Vs model of the Indian Heaven Volcanic Field. (a) Depth slice at 5-km depth. The red triangles are major volcanic edifices, and the white diamonds are Quaternary volcanic vents (Hildreth, 2007). MA, MSH, and IHVF indicate Mount Adams, Mount St. Helens, and Indian Heaven Volcanic Field, respectively. Lines B-B′ and C-C′ show the orientations of the cross sections shown in (b) and (c), respectively. The white dot is the origin of these cross sections. (b) Cross section B-B′. The white diamonds are vents within 2 km of B-B′. (c) Cross section C-C′. The white diamonds are vents within 2 km of C-C′. The contour interval for (a)-(c) is 0.025, though the 1.75 contour has been removed.
Details are in the caption following the image
Vp/Vs model of the Southern Washington Cascades Conductor (SWCC). (a) Map view of the model at 7-km depth. The dashed line is the outline of the SWCC as estimated by a previous magnetotelluric study (Egbert & Booker, 1993). The thick black line outlines the low Vp/Vs bodies inferred to represent the SWCC at this depth. The lines B-B′ and C-C′ show the orientations of cross sections shown in (b) and (c). The black dots are the origins of these cross sections. (b) East/west cross-section through the Vp/Vs model (B-B′). The thick black line outlines the low Vp/Vs anomalies inferred to be associated with the SWCC. The arrow marking the SWCC is the western boundary of this feature based upon electrical conductivity. C line indicates the intersection point of the line from (c). (c) North/south cross-section through the Vp/Vs model (C-C′). The thick black line outlines the low Vp/Vs anomalies inferred to be associated with the SWCC. The arrow marking the SWCC is the southern boundary of this feature based upon electrical conductivity. B line indicates the intersection point of the line from (b).
Details are in the caption following the image
Vp/Vs and Vp models at 7-km depth. (a) Vp/Vs model plotted with the estimated location of the eastern boundary of Siletzia (Wells et al., 1998). The thick white contours are high Vp/Vs anomalies west of the Siletzia boundary. (b) The same as (a) for the Vp model. The thick white contours are from (a).
Details are in the caption following the image
Seismicity (white dots) from the Pacific Northwest Seismic Network (PNSN) earthquake catalogue between 1 January 2000 and 1 February 2019 (pnsn.org) plotted with the Vp model. (a) Comparison at 1-km depth. All earthquakes within 3 km of this depth are included. (b) Comparison at 9-km depth. All earthquakes within 3 km of this depth are included.
Details are in the caption following the image
Regional-scale high Vp structures. (a) Vp model at 9-km depth. The white diamonds are the locations of Quaternary volcanic vents (Hildreth, 2007). The thick white line outlines the north/south trending high Vp region that is inferred to represent Siletzia. The thick black line outlines the northeast/southwest trending high Vp region inferred to represent Tertiary magmatic intrusions. The thin horizontal lines show the locations of slices shown in b-i. The black dots are the origins of these slices. (b–i) East-west cross sections of the percent perturbation of the Vp model from an average 1-D model. The regions outlined in thick white lines show deep north/south trending structures associated with Siletzia west of Mount St. Helens. The regions outlined in black show deep northeast/southwest trending structures associated with Tertiary magmatic intrusions that extend to the surface. The white diamonds are the locations of Quaternary volcanic vents within 3 km of the lines. The Chehalis Basin is labeled with CB. The eastern boundary of Siletzia (Wells et al., 1998) is labeled with ESB. Mount St. Helens is labeled with MSH and a red triangle in cross section F.

Uncertainties of the model parameters are estimated using the jackknife method (Lees & Crosson, 1989). For both Pg and Sg travel times, 50 subsets of the data are randomly selected for separate inversions. Each subset of the data involves the removal of around 1,200 and 900 travel time picks for Pg and Sg, respectively. For each data subset, three inversions are performed using three different 1-D starting models. The 1-D starting models are depth-averaged models from previous studies in this region (Kiser et al., 2016; Parsons et al., 1999; Waite & Moran, 2009; Figure S2). Uncertainty is calculated for each Vp and Vs model parameter from these 150 inversions. The uncertainties of Vp and Vs model parameters are propagated to estimate Vp/Vs uncertainties. Figures S40, S41, and S42 show uncertainties of Vp, Vs, and Vp/Vs model parameters, respectively. For the Vp, Vs, and Vp/Vs models, most of the well-resolved model parameters have uncertainties below 0.1 km/s, 0.1 km/s, and 0.05, respectively. Outside of well-resolved regions, model uncertainties increase to the standard deviations of the three 1-D starting models with respect to depth. The average Vp and Vs models from the 150 jackknife inversions (Figures S43 and S44) are very similar to the models produced using the entire data set and the 1-D starting model from Waite and Moran (2009; Figures 4 and 5).

4 Results

4.1 Vp, Vs, and Vp/Vs Models

The Vp model shows complicated structure throughout the study region (Figure 4). The highest-amplitude anomaly is a large low-velocity region (50 by 25 km) in the northwest corner of the model that extends from 0- to 6-km depth (P1 in Figure 4). The two low-velocity peaks within this region correspond to shot locations and therefore are likely not robust features. Low velocities extend discontinuously south and east from this prominent feature, with an additional large-amplitude low-velocity region in the southwest corner of the model (P2 in Figure 4), and moderately low Vp anomalies in the northeast corner of the model (P3 in Figure 4) and 40 km north of Mount St. Helens (P4 in Figure 4). Both the northwest and southwest low-velocity features terminate abruptly in depth at 6 km at high-velocity bodies, which extend to at least 11-km depth (P5 in Figure 4).

At shallow depths near Mount St. Helens, we observe two high-velocity features northeast (P6 in Figure 4) and northwest (P7 in Figure 4) of the edifice in the Vp model. Around 20-30 km south/southeast of Mount St. Helens, a third high-velocity anomaly is observed (P8 in Figure 4). These three high-velocity anomalies extend from 0- to 3-km depth and seem to merge beneath the volcano at about 2-km depth. Between 8- and 13-km depth, we observe broad northeast/southwest trending zones of low (P9 in Figure 4) and high (P10 in Figure 4) velocity anomalies roughly parallel to one another, with the transition from low to high velocities approximately below Mount St. Helens. In the eastern part of the model, a north/south trending low velocity zone is observed west of Mount Adams and Goat Rocks between 4- and 9-km depths. In the southeast corner of the model, low velocities are observed from 0- to 11-km depth (P11 in Figure 4).

The Vs model has many similar features to the Vp model (Figure 5), an expected result given the inversion scheme. These features include shallow low-velocity anomalies in the corners of the model (S1-S4 in Figure 5), three shallow high-velocity regions near Mount St. Helens (S5-S7 in Figure 5), and deep low- and high-velocity northeast/southwest trending features that strike close to Mount St. Helens (S8-S9 in Figure 5). The differences in the regional Vp and Vs models can be best observed in the Vp/Vs model (Figure 6). The Vp/Vs model has highest-amplitude features at shallow depths. Starting at 3 km, the model has the basic characteristics of high-amplitude north/south trending anomalies in the western part of the model (R1-R2 in Figure 6) and broad low-amplitude anomalies with isolated high-amplitude features in the eastern part of the model. Directly beneath Mount St. Helens, a region of low Vp/Vs is isolated near the edifice between 0- and 1-km depth. This low is surrounded by large amplitude high Vp/Vs anomalies. Starting at about 3-km depth, a region of relatively high Vp/Vs is observed near Mount St. Helens that extends discontinuously down to 13-km depth (R3 in Figure 6). Another isolated high Vp/Vs feature is present approximately 40 km southeast of Mount St. Helens from 4- to 12-km depth (R4 in Figure 6). East/northeast of Mount St. Helens is a broad region of low Vp/Vs between 3- and 13-km depth (R5 in Figure 6).

4.2 Discussion

4.2.1 Surface Geology

Many shallow anomalies in the Vp and Vs models spatially correlate with mapped geologic features in this region (Figure 7). The large low-velocity features in the northwest (Vp: 3.0-4.8 km/s; Vs: 2.0-2.6 km/s) and southwest (Vp: 3.4-4.8 km/s; Vs: 2.4-2.6 km/s) corners of the models agree well with the locations of the Chehalis Basin and the southern Portland Basin/northern Willamette River Basin, respectively. The nearly continuous region of low-velocity features (Vp: 4.4-5.0 km/s; Vs: 2.3-2.8 km/s) extending southeast from Mount St. Helens agrees reasonably well with the distribution of Quaternary volcanic deposits between Mount St. Helens and the main axis of the Cascade arc (Walsh et al., 1987). The low-velocity (Vp: 5.0-5.1 km/s; Vs: 2.8-2.9 km/s) feature 40-50 km north/northwest of Mount St. Helens corresponds to the location of the Morton Anticline, a region of exposed sedimentary rock. Near Mount St. Helens, there are five high-velocity bodies (Vp: 5.6-6.4 km/s; Vs: 3.2-3.6 km/s) that spatially agree with exposures of intrusive Tertiary rocks. These include Cinnamon Peak, Marble Mountain, the Spirit Lake Pluton, the Spud Mountain Pluton, and the Silver Star Pluton. Other high-velocity bodies near the surface are generally close to sparser exposures of intrusive Tertiary rocks.

4.2.2 Mount St. Helens and the Indian Heaven Volcanic Field

Near Mount St. Helens, both the Vp and Vs models have low-velocity anomalies (Vp: 6.0-6.2 km/s; Vs: 3.2-3.6 km/s) between depths of 4 and 14 km. At most depths the Vp model shows that these anomalies are part of a larger-scale low-velocity feature that extends southwest/northeast away from Mount St. Helens. A refined Vp model in this region shows that the velocity anomalies near Mount St. Helens form a vertically continuous body mostly offset from the volcanic edifice to the south and west [Kiser et al., 2018]. This body has a width that is as small as 5 km and exhibits significant shifts in its lateral position over depth ranges of 2 to 3 km. This anomaly has been interpreted as the primary magma reservoir beneath Mount St. Helens, and a smoothed version of this feature is present in the Vp and Vs models of the current study. It would also be expected that this primary reservoir would produce a high Vp/Vs anomaly (Takei, 2002). In the Vp/Vs model, the region near Mount St. Helens has relatively high Vp/Vs values and a few particularly high Vp/Vs anomalies (1.8-1.85) between the depth range of 4 and 14 km, though a continuous body as is observed in the refined Vp model (Kiser et al., 2018) is not present. Given the resolution differences between the Vp/Vs model and the refined Vp model, it is difficult to determine the significance of these model differences.

Another volcanic region that is resolved within the velocity models is the Indian Heaven Volcanic Field 40 km southeast of Mount St. Helens. This region exhibits a high Vp/Vs anomaly that extends from approximately 4- to 12-km depth (Figure 8). A similar feature was observed in a previous 2-D Vp/Vs model using the iMUSH data set (Kiser et al., 2016). Though the Indian Heaven Volcanic Field has produced significant volcanic output over the past 71 ka, the most recent eruption of this series of vents was around 9 ka (Hildreth, 2007). Here it is inferred that the high Vp/Vs anomaly is associated with Indian Heaven Volcanic Field volcanism, though it is not clear if this anomaly is due an active magmatic system or fluids associated with a permeable fault zone that could have acted to focus volcanism in this region.

4.2.3 The SWCC

The broad low Vp/Vs region (1.65 to 1.7) east/northeast of Mount St. Helens between 3- and 13-km depth correlates well with the location and shape of the SWCC, though this set of anomalies is more discontinuous than what has been implied for this feature based upon electrical conductivities (Figure 9). Early studies of the SWCC attributed it to a sedimentary sequence that was emplaced during the accretion of Siletzia. If this is a quartz-rich unit, it should have low Vp, Vs, and Vp/Vs (Brocher, 2005; Ji et al., 2002; Palacky, 1988; Wang et al., 2005). These general characteristics describe the seismic properties within the SWCC with the exception of a series of high Vp and Vs northeast/southwest trending anomalies that cut across the SWCC and connect to the Spirit Lake Pluton at shallow depths. A reasonable interpretation for the high Vp and Vs values in the center of the SWCC is the presence of Oligocene to Miocene intrusive igneous rocks (Brocher, 2005; Christensen, 1996), an interpretation in agreement with a recent electrical conductivity model, which shows that these high Vp and Vs regions also have high resistivity (Bedrosian et al., 2018). Given that the Spirit Lake Pluton is younger than the accretion of Siletzia (Evarts et al., 1987; Wells et al., 2014), it is likely that the high Vp and Vs igneous bodies intruded into the sedimentary sequence represented by the broader low Vp/Vs region.

4.2.4 Siletzia

A series of north/south trending high-velocity bodies are observed 30 to 40 km west of Mount St. Helens (Figure 10). The range of Vp and Vs values associated with these bodies are around 6.5 to 7.0 km/s and 3.7 to 4.1 km/s, respectively. Though resolved over a smaller volume, these high-velocity bodies generally also have high Vp/Vs values (1.8-1.9; Figure 10) and the low-velocity regions (Vp: 5.8-6.2; Vs: 3.3-3.5) separating them generally have low Vp/Vs values (1.5-1.7). The eastern boundary of these north/south trending features agrees well with the location of the eastern boundary of Siletzia estimated from aeromagnetic data (Wells et al., 1998), indicating that these velocity anomalies are associated with this accreted terrane. The location of this terrane within the upper crust suggests that it does not play an important role in focusing magma beneath Mount St. Helens, though this may not be the case at greater depths in the crust. The Vp, Vs, and Vp/Vs values associated with Siletzia are compatible with gabbro or serpentinized peridotite in the crust or mantle sections of this terrane [Brocher, 2005; Brocher et al., 2001; Christensen, 1996; Christensen & Mooney, 1995; Horen et al., 1996; Kiser et al., 2016; Ramachandran et al., 2005; Ramachandran et al., 2004; Trehu et al., 1994]. The low Vp and Vp/Vs regions between these blocks are interpreted as sedimentary rocks (Brocher, 2005; Castagna et al., 1985). Note that a distinct gap exists between the northern and southern high Vp and Vp/Vs features, which is also a region of reduced magnetic anomaly amplitudes (Wells et al., 1998). This suggests that Siletzia is a discontinuous body in southwestern Washington, an interpretation that is in agreement with conclusions from modeled electrical conductivities and differential paleorotation rates (Egbert & Booker, 1993; Wells, 1990). Another interesting aspect of this terrane is that the high Vp anomalies sit directly below the shallow low-velocity anomalies associated with the Chehalis Basin and the junction of the Portland and Willamette River Basins (Figure 12). Similar relationships between high-velocity bodies associated with Siletzia and Cascadia forearc basins have been observed beneath Puget Sound (Lees & Crosson, 1990). This relationship between accreted terranes and forearc basins has also been reported in other settings. For example, a high-velocity body has been observed beneath the Great Valley in California. This high-velocity body is thought to be an oceanic terrane that accreted onto North America during Farallon Plate subduction (Godfrey et al., 1997). In all of these cases, the dense accreted terranes would be expected to form low-lying forearc basins between the subduction zone and the active arc (Dickinson & Seely, 1979).

4.2.5 Seismicity

Seismicity within the study region is focused along a northwest/southeast trending zone known as the St. Helens Seismic Zone (SHSZ). The SHSZ runs directly beneath Mount St. Helens and includes earthquakes with mostly right lateral strike-slip focal mechanisms that accommodate forearc rotation in this region (Weaver & Smith, 1983). At shallow depths, seismicity associated with the SHSZ focuses within a relative low velocity region between the high-velocity bodies north of Mount St. Helens associated with the Spirit Lake and Spud Mountain plutons (Figure 11a). At greater depths, the SHSZ broadens in both the northwest and southeast. Much of this seismicity is focused within a similarly broad low Vp region north of Mount St. Helens (Figure 11b). This region may represent a western boundary of the metasedimentary rocks of the SWCC along which strain and seismicity is focused. In addition to the SHSZ, seismicity extends further to the west at greater depths. Though less abundant in general, it is notable that below 7-km depth, gaps exist in the distribution of the western seismicity that spatially correlate with high Vp and Vs bodies associated with Siletzia (Figures 11b and S45). This may indicate that these are rigid bodies within which strain is limited.

4.2.6 Crustal-Scale Weak Zone

The Vp and Vs models show two dominant trends of high-velocity features below 6-km depth (Figure 12). As mentioned above, west of Mount St. Helens, these features take on a north/south trend and are interpreted to be associated with the accreted terrane Siletzia. A second southwest/northeast trending high Vp (6.6-7.2 km/s) and Vs (3.7-4.1 km/s) region extends across the entire model at these depths and passes slightly southeast of Mount St. Helens. One of the high-velocity bodies associated with this southwest/northeast trending feature is located in the middle of the SWCC, but the feature as a whole clearly extends beyond this region. Two interesting spatial correlations may indicate the importance of this structure. First, the western edge of this feature aligns with the general trend between Mount St. Helens and Mount Rainier, both of which have distinct westward locations from the main Cascadia arc. Second, Quaternary volcanic vents cluster near the boundaries of this feature, but few vents are located in the center of the anomaly (Figure 12). The general northeast/southwest orientation of volcanic features in this region, as well as the northeast/southwest orientations of faults, has been pointed out in previous studies (Evarts et al., 1987) and has been hypothesized to represent a crustal-scale weak zone that focuses magma ascent. The high Vp and Vs anomalies that follow the same trend as these surface features may represent the intrusive components within this region of focused magma ascent. Cross sections through the Vp model clearly show that in several locations these deeper northeast trending features connect to shallow high Vp anomalies that spatially correlate with the locations of exposed Tertiary intrusive bodies (Figures 7 and 12). These exposed Tertiary intrusive bodies also follow a northeast/southwest trend that extends beyond the area of the velocity models (Evarts et al., 1987; Grant, 1969; Hammond, 1979). Combined these observations point towards a large scale focusing of magma along this northeast/southwest trending feature during the Tertiary (Evarts et al., 1987). Furthermore, reduced Quaternary volcanism above the inferred deep Tertiary intrusive bodies indicates that these igneous bodies from previous magmatic systems have inhibited magma ascent in the upper crust during this time and focused volcanism near their edges (Bedrosian et al., 2018). Given clockwise rotation of the forearc in this region (Wells, 1990), it is possible that this crustal-scale weak zone formed as part of a past magmatic system associated with the main volcanic arc further to the east.

5 Conclusions

The inversion of Pg and Sg travel times from the iMUSH active-source seismic experiment provides unique insights into the complexity of Vp, Vs, and Vp/Vs structure beneath and surrounding Mount St. Helens. On the regional scale, Mount St. Helens sits at the edge of a northeast/southwest trending high-velocity feature that extends across both the Vp and Vs models. Other volcanic vents are focused near the edges of this feature, and the high-velocity bodies have the same approximate trend as that between Mount St. Helens and Mount Rainier. We argue that this feature is a series of Tertiary intrusive bodies that may have aligned with a weak zone in the crust/lithosphere. Volcanic activity seems to currently be focused along the edges of this feature, which may explain the unique positions of both Mount St. Helens and Mount Rainier with respect to the rest of the Cascade Arc. One of the regions where volcanic activity has been focused is the Indian Heaven Volcanic Field. The Vp/Vs model shows a prominent high Vp/Vs body beneath this region. There are several explanations for this anomaly, though one possibility is that this feature represents shallow magmatic fluids. Two other regional features that are well resolved in the velocity models are the SWCC and the accreted terrane Siletzia. The Vp/Vs model exhibits low Vp/Vs anomalies that spatially correlate with the unique shape of the SWCC. The simplest explanation for these seismic characteristics is the presence of a sedimentary/metasedimentary sequence that composes this section of the upper crust. West of Mount St. Helens, north/south trending high Vp, Vs, and Vp/Vs bodies are associated with Siletzia. Large gaps in these features show that this accreted terrane is segmented in southwest Washington.

Acknowledgments

The authors thank two anonymous reviewers and Tom Brocher for their helpful comments. The authors thank the iMUSH group for thoughtful discussions throughout this project. Data collection for this paper was funded by National Science Foundation grants EAR-1144455, EAR-1445937, and EAR-1545750. The iMUSH active-source seismic data and field report are archived at the IRIS DMC (http://ds.iris.edu/pic-ph5/metadata/imush/form.php).