Crustal Deformation and Fault Mechanics

Migration of Tremor and Slip in Cascadia

The Cascadia subduction zone, where the Juan de Fuca tectonic plate converges with the North American tectonic plate, hosts magnitude 8-9+ megathrust earthquakes with an inter-event time estimated to be 300-500 years. The last such earthquake occurred on January 26, 1700 and had an estimated magnitude of 9. In between these earthquakes the subduction zone is not quiescent, but rather hosts repeating slow slip events and tectonic tremor, together referred to as Episodic Tremor and Slip (ETS). ETS occurs on the deep extent of the plate interface, deeper than the main earthquake region, and repeats with intervals in the range of 1-2 years depending on location. ETS is composed of slow slip on the plate interface, at the rate of about 2 mm/day, and small low-frequency seismic signals called tremor. The combined moment of the recorded tremor from seismic stations is much smaller than the total moment of slip recorded by GPS stations, implying that much of the plate interface is slipping aseismically. This slow aseismic slip is recorded on the surface as ~1cm displacements on a network of GPS stations.

Studying ETS contributes to our general understanding of subduction since it has been detected in multiple places around the world. In particular, it is debated whether a large megathrust would rupture into the ETS region, or stop updip. Further understanding of the physics of ETS is needed to answer this question. If the megathrust does rupture into this region, the shaking in cities near the coast such as Seattle and Portland will be greater. In addition, since ETS occurs frequently, it provides an opportunity to monitor for changes in the subduction zone. It is possible that ETS monitoring may give us a recognizable precursor to any large earthquake, but we don’t yet understand ETS well enough to know.

Noel Bartlow has been working on contributing to the understanding of ETS by studying how slip and tremor migrate during an ETS event. Although it is established that ETS is composed aseismic slip and tremor, which is in turn composed of individual small low-frequency earthquakes, the relationship between these two components is not well understood. Many researchers have proposed a model of a freely slipping interface with individual “stuck” patches representing tremor sources. These patches are loaded by the slip until they “pop” giving off seismic waves (see Figure 1).

There are two ways in which such patches could be loaded. Outside of the slipping region, shear stress is increased, loading the entire plate interface. Inside the slipping region, shear stress goes down in slipping areas, but up in any locked patches. These two models are illustrated in Figure 1. The difference hints at physical properties of the tremor source region. It is possible to resolve the difference between these models because ETS migrates, with both tremor and slip moving at about 10 km/day along strike. In the model shown in Figure 1a, tremor and slip migrate together. In the model of Figure 1b, tremor leads slip. Tremor migration is well recorded; however establishing the migration of the slipping region is more difficult. In this project, we perform the first inversion of its kind for migration of the slipping region in Cascadia, using GPS data from the Plate Boundary Observatory network. We use a Kalman filter based method, the Network Inversion Filter, originally developed by Paul Segall and Mark Matthews (see Segall and Matthews, JGR, 1997). The results are shown in Figure 2 and Animation 1. We find that tremor and slip migrate together, as in Figure 1a, and tremor does not appear to lead slip. The fit of this model to the data is shown in Figure 3. We also inverted synthetic data from a propagating elliptical slip distribution, to verify that peak slip rate is resolved behind the slip front. This is shown in Figure 4. These results was recently published as Bartlow et al. 2011 in Geophysical Research Letters (see publications). We are currently using theoretical models to explore the implications of this result on the size and frictional properties of tremor source regions.

Figure 1. Schematic diagrams of two possible relationships between tremor and slow slip.

Animation 1. An Animation of slip-rate and tremor. Each frame represents 1 day. Tremor locations are shown as black dots, with colors representing the average slip-rate for the day shown. Day labels refer to day of year in 2009.

Figure 2. Slip rate on the plate interface, averaged over two day intervals. Tremor epicenters are plotted in black for the same two-day intervals. The plate interface mesh is shown between 10 and 60 km depth. Note that not all of the modeled days are shown.

Figure 3. Left: Observed and predicted cumulative GPS displacements during the 2009 ETS event. Data and 1σ error ellipses shown in blue; model fit shown in red. Labeled stations refer to time series fits shown to the right. Right: Time series fits to GPS data for selected stations.
Blue dots represent east displacements with model fit in red continuous line; Pink dots represents north displacements with model fit in green. North component offset by 1 cm for clarity.

Figure 4. Selected snapshots from the slip-rate history of the resolution test. Top, smoothed filter result. Bottom, true input distribution from forward model. Red ellipses mark the current position of the slip front. Peak slip rate is resolved behind the slip front.


	Last modified Please contact the webmaster with suggestions or comments.