Slow Earthquakes: Tremor, Low-frequency Earthquakes and Slow Slip Events

Abstract

The focus of this thesis is slow earthquakes, that is earthquake-like events that release energy over a period of hours to months, rather than the seconds to minutes characteristic of a typical earthquake. Slow slip events were discovered in many subduction zones during the last two decades thanks to recordings of the displacement of Earth's surface by Global Navigation Satellite Systems (GNSS) networks. As ordinary earthquakes, slow slip events are caused by slip on a fault (for instance, the plate boundary between a tectonic plate subducting under another tectonic plate). However, they take a much longer time (several days to several years) to happen relative to ordinary earthquakes, they have a relatively short recurrence time (months to years), compared to the recurrence time of regular earthquakes (up to several hundreds of years), and the seismic waves they generate are much weaker than the seismic waves generated by ordinary earthquakes and may not be detectable. A slow slip event is inferred to happen when there is a reversal of the direction of motion at GNSS stations, compared to the inter-seismic motion of the surface displacement. In many places, tectonic tremor is also observed in relation to slow slip. Tremor is a long (several seconds to many minutes), low amplitude seismic signal, with emergent onsets, and an absence of clear impulsive phases. Tectonic tremor has been explained as a swarm of small, low-frequency earthquakes (LFEs), that is small magnitude earthquakes (M ~ 1) with frequency content (1-10 Hz) lower than for ordinary earthquakes (up to 20 Hz). Low-frequency earthquakes are usually grouped into families of events, with all the earthquakes of a given family originating from the same small patch on the plate interface and recurring more or less episodically in a bursty manner. Due to the lack of clear impulsive phases in the tremor signal, it is difficult to determine the depth of the tremor source and the distance of the source to the plate interface with great precision. The thickness of the tremor region is also not well constrained. The tremor may be located on a narrow fault as the low-frequency earthquakes appear to be or distributed over a few kilometers wide low shear-wave velocity layer in the upper oceanic crust, which is thought to be a region with high pore-fluid pressure. In the second chapter of this thesis, I compute lag times of peaks in the cross-correlation of the horizontal and vertical components of tremor seismograms, recorded by small-aperture arrays in the Olympic Peninsula, Washington, and interpret them to to be S minus P times. I estimate tremor depths from these S minus P times using epicenters from a previous study using a multibeam back-projection method. The tremor is located close to the plate boundary in a region no more than 2-3 kilometers thick and is very close to the depths of low-frequency earthquakes. The tremor is distributed over a wider depth range than the low-frequency earthquakes. However, due to the uncertainty on the depth, it is difficult to conclude whether the source of the tremor is located at the top of the subducting oceanic crust, in the lower continental crust just above the plate boundary, or in a narrow zone at the plate boundary. In the third chapter of this thesis, I extend the LFE catalog obtained by Plourde et al. (2015) during an episode of high tremor activity in April 2008, to the 8-year-long period 2004-2011. All of the tremor in the Boyarko et al. (2015) catalog south of 42 degrees North has associated LFE activity, but I have identified several other, mostly smaller, clusters of LFEs, and extend their catalog forward and backward by a total of about 3 years. As in northern Cascadia, the down-dip LFE families have recurrence intervals several times smaller than the up-dip families. For the April 2008 Episodic Tremor and Slip event, the best recorded LFE families exhibit a strong tidal Coulomb stress sensitivity starting 1.5 days after the rupture front passes by each LFE family. This behavior is very similar to what has been observed in northern Cascadia, even though the predicted Coulomb stress is about half the magnitude in the south. The southernmost LFE family, which has been interpreted to be on the subduction plate boundary, near the up-dip limit of tremor, has a very short recurrence time. Also, these LFEs tend to occur during times when predicted tidal Coulomb stress is discouraging slip on the plate boundary. Both observations suggest this LFE family may be on a different fault, perhaps a crustal fault. In many places, tectonic tremor is observed in relation to slow slip and can be used as a proxy to study slow slip events of moderate magnitude where surface deformation is hidden in GNSS noise. However, in places where no clear relationship between tremor and slow slip occurrence is observed, these methods cannot be applied, and we need other methods to be able to better detect and quantify slow slip. In the fourth chapter of this thesis, I use the Maximal Overlap Discrete Wavelet Transform (MODWT) to analyze GNSS time series and seismic recordings of slow slip events in Cascadia. I use detrended GNSS data, apply the MODWT transform and stack the wavelet details from several neighboring GNSS stations. As an independent check on the timing of slow slip events, I also compute the cumulative number of tremors in the vicinity of the GNSS stations, detrend this signal, and apply the MODWT transform. I then assume that there is a transient, interpreted as a slow slip event, whenever there is a positive peak followed by a negative peak in the wavelet signal. I verify that there is a good agreement between slow slip events detected with only GNSS data, and slow slip events detected with only tremor data. The wavelet-based detection method detects well events of magnitude higher than 6 as determined by independent event catalogs (Michel et al., 2019).