Automatic detection of secondary slip fronts in Cascadia (preliminary results)

Abstract:

Slow slip events (SSEs) in subduction zones environments propagate along the plate interface at velocities on the order of 10 km/day and are largely confined to the region known as the transition zone, located down-dip of the locked subduction thrust. As these SSEs propagate, small on-fault asperities capable of generating seismic radiation fail in earthquake-like events known as low-frequency earthquakes. Recently, low-frequency earthquakes have been used to image smaller scale secondary slip fronts (SSFs) that occur within the actively slipping portion of the fault after the main front associated with the SSE has passed. SSFs appear to occur over several different length and timescales and propagate both along dip and along strike.

To date, most studies that have documented SSFs have relied on subjective methods such as visual selection to identify them. While such approaches have met with considerable success, it is likely that many small-scale fronts have nevertheless managed to escape detection. Obara et al.[2012] successfully identified secondary slip fronts that occurred on multiple length and time scales by applying principal component analysis to a tremor catalog in Japan. We adapt their implementation to automatically detect SSFs during the three SSEs that occurred in Cascadia in 2003, 2004 and 2005. We show preliminary results of this automatic detection.

The work is still in progress, but we plan on calculating the stress drops associated with the detected SSFs relative to the main front, which should provide useful insight on the mechanical behavior of the fault zone. Ultimately, automatic detection of SSFs should provide an opportunity to discriminate different mechanical treatments and associated parameter choices by identifying the models that are successful not only capturing the primary features of SSEs – such as propagation speeds or stress drops – but also those of SSFs.