Imaging Magma Under St. Helens

In late July and early August 2014, a ~70 person field crew conducted the iMUSH active source seismic experiment (Figure 1, left). The experiment consisted of two 8 shot refraction profiles centered on Mount St Helens, one oriented SE-NW, and the second SW-NE, and areal arrays close to and distant from Mount St Helens. With offsets to 150 km, we expect to be able to see as deep as the Moho, and with closer shots (~15 km) near the volcano, we expect to also image details in the shallow crust. We used 800-1000 Reftek 125A (Texan) recorders with 4.5 Hz geophones along each profile and devoted the other 1600-1800 Texans to areal arrays deployed along a series of radial lines leading away from the volcano (see Figure 1 for details). Within the immediate vicinity of Mount St Helens, i.e., within ~7.5km radius, about 300 Texans and an additional 920 Nodal Seismic units (managed by Brandon Schmandt, University of New Mexico) were deployed on foot on the trail system around Mount St. Helens (Figure 1, right).  Lastly, more than a dozen 21-element 1 km by 1 km arrays were positioned around the volcano to record earthquakes.

Figure 1: Locations of shots (stars) and receivers (dots). Left is a view of the entire deployment, where purple denotes nodal receivers that were deployed during the entire experiment, blue is the first deployment of shots receivers, and red is the second deployment. Right is a detail of the deployment closer to the summit of Mount St. Helens.

The focus of the experiment was to image with transmitted, forward-scattered waves, since volcanic systems are notoriously poorly-resolved in backscattered (i.e., traditional reflection) experiments due to the complicated targets and heterogeneous near surface. The experiment provided two traditional refraction profiles with offsets to 150 km to image from the upper crust to Moho depths (see Parsons et al., 1998).  Two rings of shots at 15 km and 30 km radius from the summit provided fairly dense ray coverage at shallow depths beneath and around Mount St Helens.  Eight additional shots at greater offsets (50-80 km radius from the summit) provided sampling to the base of the crust.  The iMUSH arrays recorded ~ 8 x 10⁴ traces from the 23 shots. Two of the larger shots registered as M_L=2.3 events by PNSN permanent stations.  We also recorded numerous regional and local earthquakes, including a number occurring directly under Mount St. Helens, as well as some regional quarry blasts.

The two profiles will be interpreted using the hybrid tomography/layered inversion developed by C.A. Zelt for the highly heterogeneous crustal structure the southern Caribbean margin (Clark et al. 2008; Magnani et al. 2009). The areal array data will be analyzed using the 3D methods developed by Zelt and Barton 1998, which we employed on areal array data from the Caribbean (Arogunmati, 2006; unpublished Rice MS thesis). To extract the most information from the data, we will employ waveform tomography on the two profiles and from 2D slices through the areal arrays (Pratt et al. 1998), having experience with this method used on other types of transmission data (Gao et al. 2006; Gao et al. 2007). The travel-time models will be used as starting models for the waveform tomography. Having completed this we will attempt 3D adjoint tomography on data from the inner two rings using the SPECFEM_SESAME forward and adjoint codes available through the Computational Infrastructure for Geodynamics (e.g. Tape et al. 2007). Other researchers working on volcanic systems have had success applying traveltime tomography to volcanic systems (e.g., Zollo et al. 1998; Paulatto et al. 2010), and waveform tomography to crustal scale data (Operto et al. 2004). No 2D or 3D waveform or adjoint tomography has been previously attempted on data from a volcanic system to our knowledge.