Seismicity Along the Endeavour Segment of the Juan de Fuca Ridge: Automated Event Locations for an Ocean-Bottom Seismometer Network

From 2003-2006, the W.M. Keck Foundation supported the operation of a network of eight ocean-bottom seismometers (OBSs) that were deployed with a remotely operated vehicle along the central portion of the Endeavour Segment of the Juan de Fuca mid-ocean ridge as part of a multidisciplinary prototype NEPTUNE experiment. Data from 2003-2004 were initially analyzed during a research apprenticeship class at the University of Washington's Friday Harbor Laboratories. Eight student analysts located ~13,000 earthquakes along the Endeavour Segment. Analysis of data from 2004-2005 has to date been limited to locating ~6,000 earthquakes associated with a swarm in February-March 2005 near the northern end of the Endeavour Segment. The remaining data includes several significant swarms and it is anticipated that tens of thousands of earthquakes still need to be located. In order to efficiently obtain a complete catalog of high-quality locations for the 3-year experiment, we are developing an automatic method for earthquake location. We first apply a 5-Hz high-pass filter and identify triggers when the ratio of the root-mean square (RMS) amplitudes in short- and long- term windows exceeds a specified threshold. We search for events that are characterized by triggers within a short time interval on the majority of stations and use the signal spectra to eliminate events that are the result of 20-Hz Fin and Blue whale vocalizations. An autoregressive technique is applied to a short time window centered on the trigger time to pick P-wave times on each station's vertical channel. We locate the earthquake with these picks and either attempt to repick or eliminate arrivals with unacceptable residuals. Preliminary S-wave picks are then made on the horizontal channels by applying a 5-12 Hz bandpass filter, identifying the peak RMS amplitude for a short running window, and making a pick at the time the RMS amplitude rises above 50% of this value. The picks are refined using the autoregressive technique. We then locate earthquakes using the P- and S-wave picks and either repick or discard S-wave picks with unacceptable residuals before obtaining the final location. We have tested the method using data segments from 2003-2004 and find the catalog completeness and quality of locations is comparable to that obtained with the student analysts.