Spatial and Temporal Comparisons of Tremor and Slow Slip in Cascadia

Abstract:

Tremor is often thought to be a proxy for slip during ETS events and has been shown to have a relatively abrupt updip boundary, implying an abrupt updip limit of slip. However, as shown by Houston (AGU abstract, 2012) and Hall and Houston (AGU abstract, 2014), slip inferred from GPS extended updip of the seismically-detected tremor in the 2010 M6.8 and 2012 M6.7 ETS events. If slip extending updip of tremor is a persistent phenomena, tremor cannot be directly used as a proxy for slip. Following the methods used on the 2010 and 2012 ETS event, we found that the August 2009 ETS around Portland, OR also showed slip updip of tremor. Principal Component Analysis was implemented to automatically select the direction and magnitude of the maximum displacement vector. Our Green’s functions use the Okada formulation of buried rectangular faults in a halfspace for a grid of 8x8 km subfaults based on the McCrory slab model. We then performed static inversions with 2nd order Tikhonov regularization to find slip on the fault surface. We also compared two different inversions for 2009, one where slip was allowed on a broad regional grid and a tremor-restricted inversion (TRI) where slip was restricted to subfaults in which tremor occurred. We found the 2009 ETS event released the equivalent of a M6.8 in slip. We also found that, as in the previous ETSs, the TRI forced up to 10 cm of slip to the updip edge of the grid, which is exceeds the amount of plate convergence expected in the inter-ETS periods and is therefore physically unsustainable over several ETS events. The excess slip along the updip edge in the TRI models also suggests that the geodetic data prefer slip with a larger footprint than the spatial pattern of tremor, and supports our conclusion that tremor does not represent all of the slip during an ETS event. We see consistent and clear spatial relationship between tremor and slip with some slip occurring updip of tremor. Our static inversions show where slip is occurring we also want to know when slip is occurring relative to tremor. We will further explore this relationship by using the Extended Network Inversion Filter (ENIF) (Segall and Matthews, 1997; McGuire and Segall, 2005). By using the ENIF we can estimate the slip and slip rate on subfaults, and find in particular, when the updip slow slip tends to occur relative to the tremor front.