The transition from stable to slow to fast earthquake slip: the influence of surface morphology, fault normal stiffness and lithology

Over the last decades, new types of earthquakes have been discovered. The most well-known group of ordinary earthquakes might be the most dangerous as they emit the largest amount of seismic radiation and cause ground-shaking, but repeating slow earthquakes can also damage buildings and infrastructure. Ordinary earthquakes occur when movement on a fault is unstable and a run-away process accelerates the movement to seismogenic velocities. During slow earthquakes, there are also clearly defined phases of faster slip along the fault, but the maximum slip velocity reached during these phases is lower. Then, there are aseismic faults, where slip accumulates constantly by stable creep at a rate close to the far-field stressing rate. The mechanisms that control the nature of sliding behavior of faults are multiple and studied in more or less detail. In this thesis, I explore how three factors influence fault stability: fault surface roughness and roughness anisotropy, fault-normal stiffness and stiffness contrasts across a fault, and the lithological controls on the extraordinary shallow slow slip events in the Hikurangi subduction zone margin (New-Zealand).
Here, I present results using direct shear experiments, while varying one of the studied variables. To study the influence of fault surface morphology, I use two materials; a velocity-weakening and therefore potentially unstable pure quartz powder, and Rochester shale powder, which is velocity-strengthening and therefore likely to show stable sliding. Fault surface morphology evolves with displacement and its influence on frictional behavior is therefore studied by varying the amount of displacement on the samples. To test the influence of host-rock stiffness, the testing device is fitted with springs of variable stiffness in both the shear-parallel and fault-normal directions. Testing occurs on the intrinsically unstable quartz powder and I analyze both the frictional properties as well as the slip instabilities that occur. For the study about the Hikurangi margin, I use samples of the sediments on the incoming plate and use realistically low deformation rates, to study the frictional behavior and the occurrence of spontaneous slow slip events during the experiments.
The results show rough, isotropic faults can host slip instabilities, because these show the required velocity-weakening frictional behavior. Striated, smooth surfaces are velocity-strengthening and promote stable sliding. The formed fault surfaces obey the typical self-affine fractal scaling, that make these results directly applicable to natural faults. Reducing the fault-normal stiffness causes the fault to become less velocity-weakening and would therefore promote stable sliding. However, slip instabilities occur when the fault-normal stiffness is reduced, which I explain by a different mechanism that requires a stiffness asymmetry. The asymmetry is the result of reducing the fault-normal stiffness on one side of the fault. The plate-rate shear experiments on Hikurangi sediments show spontaneous slow slip events occur in the calcite-rich lithologies, whereas the weakest lithologies are velocity-strengthening.
Altogether, the results presented in this thesis suggest unstable sliding will occur on rough, isotropic fault patches. The slow slip events in the Hikurangi margin can only occur when the slow slip event-hosting lithologies are introduced into the deformation zone. This could be explained by a geometrically complex deformation zone due to subducting seamounts. Stiffness contrasts, due to lithological contrast across a fault or due to asymmetric damage, may cause slip instabilities that are not explained by the traditional critical stiffness theory. I show the three studied variables are closely linked and fault surface roughness, fault stiffness and stiffness contrast, as well as fault zone lithology may affect each other.