3D Cascadia slow slip event model constrained by tremor locations and gravity

Abstract:

Slow slip events (SSEs) and non-volcanic tremors are proposed to be the geodetic and seismic manifestations of the same fault slip process in the transition zone from seismogenic and continuous sliding depths [Peng and Gomberg, 2010]. Numerous observations in world-wide subduction zones have found SSEs and tremors arising quasi-periodically with various source parameters[Gao et al., 2012] that may be characteristic fault slip controlled by subduction structure. Two-decade records of synchronized SSEs and tremors in Cascadia render it an ideal site to investigate the physical mechanism of such slow slip phenomena. We build a comprehensive forward model for SSEs in Cascadia with fault geometry, frictional property, effective normal stress loading and plate convergence rates constrained by various types of observations along the margin. The fault surface is constructed from the slab geometry determined by relocated seismicity, covering from Vancouver Island to Oregon where GPS records have good coverage and slip inversions can be acquired[Schmidt and Gao, 2010]. The automatically detected tremors between August 2009 and March 2015 [Wech and Creager, 2008] are used to constrain the distribution of rate-and-state frictional parameter a-b in the slow slip zone. The along-strike effective normal stress distribution follows the gravity anomaly profile along the SSE depth contours. Five patches of Mw6.0 SSEs spontaneously appear where tremors span a wider along-dip distance. The spatial correlation between SSE and tremor distributions confirms the controlling effect of fault frictional properties. The segmented recurrence intervals of SSEs in Cascadia are related to the loading of the overlying continent, as revealed by gravity anomalies[Brudzinski and Allen, 2007]. In our model, we use gravity anomalies as a proxy for the effective normal stress, hence segments of longer SSE recurrence intervals correspond to where higher gravity anomalies are observed. We also found the source parameter scaling, in particular the moment-duration scaling, varies as different selection criteria are applied. The moment-duration scaling changes from Vave>Vpl to Vave > 0.5Vpl when SSEs are defined by decreasing the slip rate threshold from which suggests the scaling may be affected by current geodetic resolution limit.