Observations and modeling of temporal variability in slow slip events

Abstract

In this thesis, I investigate short-timescale variations in slow slip events in Cascadia. I use these and other observations to assess whether one of the friction laws proposed to govern the slow slip region can adequately reproduce the observed events.

In the first observational component, we use borehole strain data to look for tidal variations in the slow slip moment rate in central Cascadia. We find that slow slip is tidally modulated. On average, the moment rate oscillates 25% above and below the mean at the period of the strongest tide. This modulation implies that slow slip is sensitive to small external stresses. It provides a useful constraint on models of slow slip events.

In the modeling component of this thesis, we examine features of slow slip events simulated with a rate and state friction law that is velocity-weakening at low slip rates but velocity-strengthening at high slip rates. This is one of three friction laws that have been proposed to govern the frictional strength in the slow slip region. These models parameterize the slow slip region as an elongate rectangle. This mimics the geometry of observed events, which often extend farther along strike than along dip. The simulated events propagate approximately steadily "along strike," and slip rate and stress decay gradually behind the propagating front. The recurrence interval of large events is controlled by the requirement that the strain energy released by slip equal the energy dissipated by friction. We identify the sets of model parameters that allow for episodic large events with the stress drops, slip velocities, and propagation rates seen in Cascadia.

Next, we investigate the effect of applying a tidal load to this model. We find that the slip rate varies quasi-sinusoidally, with amplitude proportional to the applied stress. It is possible to choose model parameters that allow the model to reproduce the observed modulation, but if we do so, the model can match only a subset of the observed stress drops.

We also investigate simulated back-propagating fronts. These are small regions of high slip rate that propagate in the opposite direction of the longer-term propagation, back through the region that has already slipped. Such fronts have been inferred from tremor observations. The modeled fronts propagate much more slowly than the observed ones. For the modeled fronts to reach the speeds indicated by the observations, the slip distribution in the slow slip event would have to be highly heterogeneous in space and time. Even if we added significant spatial heterogeneity to the modeled interface, it would be different for our model to reproduce some aspects of the observations.

On the other hand, in the final observational component of this thesis, we show that slow slip does exhibit signicant temporal heterogeneity on timescales shorter than one day. Such variability is frequently observed in tremor but is difficult to observe in slow slip because geodetic observations have limited resolution. We look for short-timescale variations in slow slip that are correlated with variations in tremor amplitude. We find that, on average, the tremor amplitude and the slow slip moment rate are correlated on timescales between 15 minutes and 16 hours. The aseismic moment rate changes by at least a factor of 2 on timescales shorter than 4 hours.