| Slow earthquakes have been observed in major plate boundaries worldwide, and accommodate a significant part of the plate motion through slow slip in the transition zone of the faults. They occur down-dip of the locked zone, where large damaging fast earthquakes nucleate. The physical processes that control slow quakes, however, remain enigmatic. To understand slow earthquakes, I study non-volcanic tremor, a form of seismic radiation associated with slow quakes. It is challenging to detect and locate tremor due to its lack of clear impulsive arrivals. I develop a new beam-backprojection technique to image slow earthquakes in high resolution by detecting and precisely locating tremor using small aperture seismic arrays. This technique can detect more duration of tremor, gives high resolution in tremor locations compared to a conventional envelope cross-correlation method, and also resolve tremor depth. I apply this technique in Cascadia, and show that the majority of tremor is occurring near the plate interface suggesting that they are possibly a result of shear slip on the subduction fault. Transition zone producing tremor appears to be fairly heterogeneous. Three patches down-dip of the transition zone produce majority of the tremor during small to moderate-sized tremor episodes. The patches repeat 10--15 times in 15 months. On the other hand, several up-dip patches are responsible for most of the tremor activity during large slow quakes. Moreover, I find that tremor behavior changes dramatically over different time scales. Over the time scale of several minutes, tremor propagates rapidly sub-parallel to the slip direction of the subduction zone at a velocity of ∼100 km/hr. This quasi-continuous streaking of tremor produces slip-parallel tremor bands over the time scale of several hours. Tremor bands migrate along-strike resulting in the slow rupture propagation at an average velocity of ∼8 km/day. Along-strike slow rupture propagation velocity during a large slow quake, however, varies by at least a factor of five suggesting a variable fault strength in the transition zone. In addition, I found a wide range of tremor propagation velocity indicating convoluted evolution of slip during slow quakes. Complex interaction between stress transfer, fault structures, and fluid flow may cause the variegated nature of tremor propagation. The observations indicate that strong heterogeneity, possibly rheological, in the transition zone may control tremor generation and rupture propagation during slow quakes. In this study, I develop new tremor detection and location techniques that allow us to image slow earthquakes in great detail, and give new insights into their underlying physics and the tectonic behavior of transition zone. |
