Modeling Episodic Slow Slip Events on a 3D Non-planar Cascadia Subduction Fault

English Abstract:

Previous numerical simulations with a planar plate model can reproduce slow slip recurrences (e.g. [Liu and Rice, 2009]), but do not address the effect of fault geometry in the diverse slow slip phenomena all around the world. Here we build up a numerical model in the framework of rate- and state-dependent friction for a three-dimensional Cascadia subduction fault with geometry constrained by seismicity relocation [McCrory et al. 2012], to investigate the source processes of episodic slow slip events (SSEs). Our modeling results show that episodic slow slip events appear every ~1.5 year beneath Port Angeles. Each episode lasts ~40 days and propagates ~150 km bilaterally along the strike at the speed of 2~7 km/day. Slow slip is distributed between 35-45 km with a maximum cumulative slip of ~3.5 cm in each episode. All the above features are comparable to the geodetically inferred SSE source characteristics in northern Cascadia [Schmidt and Gao, 2010]. We also find small and gradual slip signals (velocity lower than 0.5 mm/day) extending several months before and after each active SSE episode, similar to the nucleation and healing phases of seismic rupture events.

Our preliminary result shows an approximate along-strike anti-correlation between the cumulative slow slip and the average dipping angle of the SSE zone. However, larger slip amplitude could also be the result of a wider SSE zone. Our next step is to pinpoint which of the two conditions (dipping angle and SSE zone width) is responsible for the accumulated slow slip variation. Additionally, we will use Cascadia tremor locations to constrain the low effective normal stress distributions, and consequently to elaborate the SSEs patterns. We will rely on the quantitative comparison with GPS inversion results to constrain model parameters.