A time-dependent catalog of episodic tremor and slow slip in Cascadia

We present an update on our new self-consistent, continuous time-dependent catalog of ETS events in Cascadia for the period 2006-2019 using the Network Inversion Filter (NIF) and heterogeneous elastic Greens functions computed using the PyLith finite element software (Aagaard et al., 2013, 2019), with density (and by extension, shear modulus) estimated via empirical relationships with VP (Brocher, 2005), using the velocity model of Stephenson (2007). We use this model to automatically detect ETS events based on criteria involving a minimum slip rate for a minimum number of days on a minimum number of subfault patches. We explore the scaling properties of these ETS events and find scaling between time1 (bounded growth) and time3 (unbounded growth), consistent with prior studies (e.g., Gomberg et al., 2016). Some numerical modeling studies predict slow slip events may become smaller and more frequent, and/or the updip limit may become shallower as the megathrust approaches the time of the next great earthquake (e.g., Luo and Liu, 2019; Zhu et al., 2020). We do not see coherent changes in recurrence intervals or updip migration in our catalog, but we explore how much updip migration will need to occur to be detectable with the current GNSS network. We also show that geodetically measured slip rate and seismically detected tremor from the Pacific Northwest Seismic Center catalog (Wech & Creager, 2008; Wech, 2010) track together extremely well in every detected ETS event for the time period covered by the tremor catalog, similar to what has been shown previously for individual ETS events (Bartlow et al., 2011; Wech and Bartlow, 2014). We compare our results for cumulative slip from this time-dependent model to those of Bartlow (2020), which used a static approach, and discuss the role of ETS in the overall slip budget in Cascadia. We show that total slip from the time-dependent approach is consistent with the results of Bartlow (2020), but with considerably less spatial resolution, illustrating the trade-off between spatial and temporal resolution in ETS modeling. We also explore tests of spatial and temporal resolution for the NIF applied to Cascadia and discuss future plans to extend the catalog past 2019 and automate the inversion process.