Improvements to Absolute Locations from an Updated Velocity Model at Mount St. Helens, Washington

Over 700,000 earthquakes have been detected at Mount St. Helens, Washington since it reawakened last fall. Over 99% of these earthquakes have occurred within 1.5 km of the surface, under or near the active dome. Currently, picks of well recorded P-waves on 10 to 15 permanent seismic stations are used with a simple 1-D velocity model with station corrections to generate catalog locations for a subset of earthquakes. Formal errors are 100-500 meters; however, because of the large station corrections needed to minimize average residuals, the systematic errors due to unknown and variable velocity structure is likely to be larger than this. Using cross-correlation techniques with hypoDD on event families, we find hundreds of events that cluster into very small volumes, however these clusters have little constraint on the actual location of the cluster in space, since only the relative lag time between similar earthquakes is kept. This highlights the need to determine and apply a realistic shallow velocity model to achieve improved absolute locations that may supplement high-resolution relative locations. To improve the velocity model, we installed a 4-km long transect that extended from the 1986 Dome north to the Pumice Plain. Our transect consisted of 39 vertical component stations at 100 m spacing, which alternated between 1 Hz and 4.5 Hz sensors to enhance the recording of low frequencies across the transect. During our 3-day deployment, we recorded 11 earthquakes above magnitude 2.0, as well as coherent energy from rockfall sources across the entire array. In order to constrain the dip of our unreversed refraction profile, we utilize Rayleigh wave dispersion with the Refraction Microtremor (ReMi) method to get multiple S-wave soundings and track common velocities from different sections of the line. We combine the results from our transect with previous near surface S-wave soundings to establish an updated 2-D velocity model of the edifice that extends past 2 km depth. Shallow p-wave velocities are as low as 350 m/s while our deepest p-wave velocities are nearly 4.0 km/s. This improved velocity model, allows us to use arrivals from the 13 permanent stations within 10 km of the active vent, and up to 25 additional 3-component sensors that were temporarily installed during the summer of 2005, to greatly improve the absolute accuracy in earthquake locations.