Using Tremor to Predict Strong Ground Motion

Due to its widespread occurrence, frequency content, and location, tectonic tremor presents an exceptional opportunity to test and improve strong ground motion attenuation relations for subduction zones. We characterize the amplitude of thousands of individual tremor events to constrain the distance attenuation relationship of peak ground acceleration (PGA) and peak ground velocity (PGV) of tremor for application to strong ground motion prediction. Ground motion prediction equations (GMPE) relate ground motion amplitude to earthquake magnitude and distance, and are critically important for creating seismic hazard maps. In practice, GMPEs are determined empirically with earthquake data, recorded at many stations. In some areas of high earthquake hazard, such as Cascadia, the data set of recorded earthquakes for ground motion prediction is extremely sparse. Tectonic tremor, however, occurs frequently and abundantly in many subduction zones, including Cascadia. Moreover, the tremor band of 1-10 Hz, corresponds precisely to the frequency band of greatest interest for engineering seismology. Finally, tremor is thought to occur at or near the down-dip limit of rupture in large megathrust earthquakes, which is the area that is likely to control high frequency shaking in large earthquakes. Tremor in Cascadia is strong enough to be recorded at borehole stations of the Plate Boundary Observatory to distances of nearly 200 km, which is sufficient to place important constraints on ground motion decay with distance. We characterize the amplitude of thousands of individual tremor events occurring nearly continuously throughout an ETS event, using tremor locations and timings from independently determined catalogs [Wech and Creager, 2008]. Because tremor bursts do not have a defined absolute magnitude, we normalize all events to a reference magnitude. PGA and PGV for tremor shows a distance decay that is similar to subduction-zone-specific GMPEs developed from both data and simulations [e.g. Atkinson and Boore, 1997]; however, the massive amount of data present in the tremor observations allows us to improve the fit of the GMPEs, to refine distance-amplitude attenuation relationships, and to search for regional variations in behavior.