Episodic slow slip process in a non-planar fault model constrained by non-volcanic tremor locations along Cascadia subduction zone

Abstract:

Slow slip events (SSEs) are observed in the circum-Pacific subduction zones and exhibit a wide diversity of source parameters (including equivalent moment, duration and recurrence interval). Gao et al [2012] compiled source parameters of SSEs around the world and revealed their empirical scaling relations distinct from those of regular earthquakes. However, the nature of this diversity is unclear. Previous 3-D numerical simulations in a simplified plate model have reproduced the along-strike segmentation of episodic SSEs in Cascadia margin and their source scaling relations in the framework of rate- and state- friction law [Liu, 2014]. But the planar fault model is inefficient to investigate the effect of the fault geometry on the source characteristics of SSEs in a specific subduction zone.

In this study, we adopt a non-planar Cascadia subduction fault geometry constrained by relocated seismicity [McCrory et al. 2012] in the rate-state friction model. We have reproduced episodic SSEs beneath Vancouver Island and Washington arising every ~1.5 year with a maximum cumulative slip of ~2.5 cm. We find three phases characteristic in cumulative moment rate and slip rate in each episode. These three phases, defined as nucleation, fast-spreading and healing, lasting for ~160, ~60 and ~140 days, respectively. Both the nucleation and healing phases are beneath GPS detection threshold but the nucleation phase is spatially correlated with the small tremor activities arising deeper than episodic tremor and slip (ETS).

To further investigate the diversity of SSEs in Cascadia, we introduce the recent 6-year tremor locations in Cascadia (http://pnsn.org/tremor) to constrain the near-lithostatic pore pressure distribution at the SSE depths in an 800km-long Cascadia fault model. Additionally, we set a step change of effective normal stress in SSE region from 1.5 MPa in Vancouver Island and Washington to 2.0 MPa in Oregon. The results show that the modeled SSEs exhibit distinct slip ‘patches’ along the strike (Fig. 1). This distribution of average cumulative slip is correlated with the frictional heterogeneity introduced by tremor locations. We also calculate the rupture area, equivalent moment and duration of each event and compare their scaling relations with the results of GPS inversions[Schmidt and Gao, 2010].