iMUSH – Overview

Description: Local earthquake tomography in the area around Mount St. Helens, WA, part of the broader imaging Magma Under St. Helens (iMUSH) project. This interdisciplinary project aims to illuminate the magmatic and geologic structure underneath Mount St. Helens from the subducting plate to the surface. This page is an overview of my work, with more detailed posts about different parts to come (e.g. 3D model viewer, Travel-time database).

Techniques: seismic data collection, data quality analysis, data archival, seismic database creation and management, seismic wave travel-time picking, travel-time tomographic inversion, 3-D model plotting and evaluation, interdisciplinary collaboration

Tools: MATLAB, Bash, C++, Unix, IRIS-WS java library, Antelope, netCDF, Generic Mapping Tools (GMT)

Data Collection: A wide range of field work was performed as part of the iMUSH project, which I summarized in this article.

  • Site locations and routes in Google Earth
  • Mount St. Helens, WA
  • Looking for seismic sites
  • Looking for seismic sites
  • Looking for seismic sites
  • ~20 volunteers deployed 70 seismometers summer 2014
  • Equipment stored in a hangar in Kelso, WA
  • Seismometer being placed
  • Solar panel, instrument box
  • Pipes to hold up solar panel
  • Sometimes it rains...
  • Servicing instruments, checking logs. This one lost power during much of the winter
  • Getting to a site is hard sometimes
  • Oops, water shouldn't be in there
  • Digging a french drain to protect site from water
  • Some sites were tampered with by animals or humans
  • Several large washouts made this site hard to remove
  • This site was mostly destroyed by a forest fire
  • Equipment lost to forest fire
Much of the data for my work came from a 70-element broadband seismometer array that was placed in a ~50 km radius around Mount St. Helens from 2014 to 2016. I organized teams to visit each site 2-3 times a year, make sure the site was working, and swap data disks.

Data cleaning and archival: I used tools provided by the PASSCAL instrument center to confirm quality of the seismic data, then uploaded the data to the IRIS data management center using a shell script written by Steve Malone, UW.

Seismic database, travel-time picking: Next I created and stored the data in an Antelope database, then ran automatic processes to detect and associate seismic wave arrivals (travel-times) with each other. Along with three undergraduate students whom I supervised, we reviewed and adjusted ~20,000 travel-time picks. I also assembled an additional ~100,000 travel-times from four separate sources to use in a tomographic inversion

Travel-time tomography: With travel-times in hand, I performed an inversion to obtain 3-D P- and S-wave seismic velocity models of the Mount St. Helens region. This was done with a C++ program, struct3DP, written by Robert Crosson (UW), which is an iterative, non-linear, conjugate gradient, least-squares inversion code with a finite-difference 3-D eikonal equation solver. The inversion is regularized using a 3-D Laplacian smoothing operator. The optimal tradeoff between model damping and data fit was determined using tradeoff-curve analysis.

3-D model evaluation: I ran a variety of inversions using different datasets and inversion parameters, and came up with several tools to compare models against each other, including an automated script to compile and write out different parameters (e.g. RMS misfit, model size, tradeoff parameter, etc.), and codes to plot cross- and depth-sections through the models (first in matlab, then using Generic Mapping Tools)

Depth sections through the P-wave velocity model. Colors show % variation from a 1-dimensional average (blue is faster, red is slower than average).

Interpretation: Changes in seismic velocity can be related to a number of factors, including rock type, temperature, fractures, and/or the presence of fluids or melt. We interpret low velocities beneath Mount St. Helens to be related to a narrow, roughly cylindrical region between 6-15 km depth that is holding 15-20 cubic kilometers of partial magmatic melt.

Collaboration: Besides the local earthquake tomography which I performed, a variety of other methods were employed by others as part of the iMUSH project, including ambient noise tomography, receiver function analysis, active source tomography, reflection imaging, earthquake analysis, and magnetotelluric and petrologic studies. This included workers at Cornell University, Rice University, University of Arizona, University of New Mexico, ETH Zurich, and the USGS. Some of the early results of this work are summarized in this article which I helped to write. Collaboration between each group is ongoing and has already occurred with group discussions at national meetings (e.g. AGU, IRIS workshops), and two iMUSH workshops held at the USGS Cascades Volcano Observatory in Vancouver, WA, and the University of Washington.