Study of Tectonic Tremor in Depth: Triggering Stress Observation and Model of the Triggering Mechanism
- Author(s): Wang, Tien-Huei
- Advisor(s): Oglesby, David D
- et al.
Non-volcanic tremor (NVT) has been discovered in recent years due to advances in seismic instruments and increased density of seismic networks. The NVT is a special kind of seismic signal indicative of the physical conditions and the failure mechanism on the source on the fault where NVT occurs. The detection methods used and the sensitivity of them relies on the density, distance and instrumentation of the station network available. How accurately the tremor is identified in different regions varies greatly among different studies. Therefore, there has not been study that rigorously documents tectonic tremors in different regions under limited methods and data. Meanwhile, many incidences of NVTs are observed during or after small but significant strain change induced by teleseismic, regional or local earthquake. The understanding of the triggering mechanisms critical for tremor remains unclear. In addition, characteristics of the triggering of NVT in different regions are rarely compared because of the short time frame after the discovery of the triggered NVTs. We first explore tectonic tremor based on observations to learn about its triggering, frequency of occurrence, location and spectral characteristics. Then, we numerically model the triggering of instability on the estimated tremor-source, under assumptions fine-tuned according to previous studies (Thomas et al., 2009; Miyazawa et al., 2005; Hill, 2008; Ito, 2009; Rubinstein et al., 2007; Peng and Chao, 2008). The onset of the slip reveals that how and when the external loading triggers tremor. It also holds the information to the background stress conditions under which tremor source starts with.
We observe and detect tremor in two regions: Anza and Cholame, along San Jacinto Fault (SJF) and San Andreas Fault (SAF) respectively. These two sections of the faults, relative to general fault zone on which general earthquakes occur, are considered transition zones where slip of slow rates occurs. Slip events including NVT occur on these sections have slower slip rates than that of the general earthquakes (Rubin, 2008; Ide, 2008). In Azna region, we use envelope and waveform cross-correlation to detect tremor. We investigate the stress required to trigger tremor and tremor spectrum using continuous broadband seismograms from 11 stations located near Anza, California. We examine 44 Mw≥7.4 teleseismic events between 2001 and 2011, in addition to one regional earthquake of smaller-magnitude, the 2009 Mw 6.5 Gulf of California earthquake, because it induced extremely high strain at Anza. The result suggests that not only the amplitude of the induced strain, but also the period of the incoming surface wave, may control triggering of tremor near Anza. In addition, we find that the transient-shear stress (17-35 kPa) required to trigger tremor along the SJF at Anza is distinctly higher than what has been reported for the well-studied SAF (Gulihem et al. 2010).
We also use a newly deployed mini-array to detected tremor along the SAF. The sWe use a beam-backprojection method (Ghosh et al., 2009, 2012) that enhanced imaging power to detect tremor more accurately than most conventional methods. This method finds the coherent tremor signal by stacking waveform data in slowness domain with a delay-and-sum approach. We detect tremor along the Cholame section of the SAF using the beam-backprojection method along with visual inspection to confirm the accuracy. We found at least 2 times more of the tremor rates in comparison with the rate reported by previous studies (Nadeau and Guilhem, 2009). With the enhanced imaging power of beam-backprojection, we observed the triggering potential of 11 earthquakes which induced significant shear stress (0.07-4.5 kPa) in this region. The 11 earthquakes consist of 7 teleseismic (Mw >=7.0) and 4 regional (ML ≥ 4.0) earthquakes. We observe the tremor rate before, during and after the 11 earthquakes. We find, in 7 out of 11 earthquakes, the tremor rate after the P-wave arrival of the major events is at least 2 times greater than the rate before. The results may also suggest that, for external loading of peak shear stress more than 0.25 kPa, triggering of tremor occurs 4 times out of 5. However, for peak shear stress less than 0.25 kPa, no apparent trend can be indicated.
We model slip initiation using the analytical solution of rate-and-state friction. We verify the correctness of this method by comparing the results with that from the dynamic model, implemented using the Multi-Dimensional Spectral Boundary Integral Code (MDSBI) written by Eric M. Dunham from Sanford University. We find that the analytical result is consistent with that of the dynamic model. We set up a patch model with which the source stress and frictional conditions best resemble the current estimates of the tremor source. The frictional regime of this patch is rate-weakening. The initial normal and shear stress, and friction parameters are suggested by previous observations of tectonic tremors both in this and other studies (Brown et al., 2005; Shelly et al., 2006; Miyazawa, 2008; Ben-Zion, 2012). Our dynamic loading first consists of simple harmonic stress change with fixed periods, simplifying the transient stress history to resemble teleseismic earthquakes. We tested the period and amplitude of such periodic loading. We find that the period of the transient shear stress is less important relative to the amplitude. The triggering depends mainly on the ratio between amplitude of the shear stress loading and the background normal stress. We define a range of ratio indicative of the occurrence of the triggering. We later test the triggering of the instability using the shear stress history from 44 large teleseismic earthquakes (data equivalent to those used in Chapter 1). With the constraints of these observations, we find that the background normal stress should be in the range of ~400-700 kPa. The background normal stress suggested agrees with the common hypothesis that the tremor source is under low normal stress. In addition, our results provide a first estimation of the background normal stress with numerical method. We also demonstrate how our model find constrains on the background physical stress or frictional conditions,, with several true incidences that transient shear stress triggers or not-triggers tremor..