| This dissertation seeks to further understand the LFE source process, the role LFEs play in generating slow slip, and the utility of using LFEs to examine plate interface structure. The work involves the creation and investigation of a 2-year-long catalog of low-frequency earthquakes beneath the Olympic Peninsula, Washington.;In the first chapter, we calculate the seismic moments for 34,264 low-frequency earthquakes (LFEs) beneath the Olympic Peninsula, WA. LFE moments range from 1.4x1010-- 1.9x1012 N-m (M W=0.7-2.1). While regular earthquakes follow a power-law moment-frequency distribution with a b-value near 1 (the number of events increases by a factor of 10 for each unit increase in MW), we find that while for large LFEs the b-value is ∼6, for small LFEs it is <1. The magnitude-frequency distribution for all LFEs is best fit by an exponential distribution with a mean seismic moment (characteristic moment) of 2.0x1011 N-m. The moment-frequency distributions for each of the 43 LFE families, or spots on the plate interface where LFEs repeat, can also be fit by exponential distributions. An exponential moment-frequency distribution implies a scale-limited source process. We consider two end-member models where LFE moment is limited by (1) the amount of slip or (2) slip area. We favor the area-limited model. Based on the observed exponential distribution of LFE moment and geodetically observed total slip we estimate that the total area that slips within an LFE family has a diameter of 300 m. Assuming an area-limited model, we estimate the slips, sub-patch diameters, stress drops, and slip rates for LFEs during ETS events. We allow for LFEs to rupture smaller sub-patches within the LFE family patch. Models with 1-10 sub-patches produce slips of 0.1-1 mm, sub-patch diameters of 80-275 m, and stress drops of 30-1000 kPa. While one sub-patch is often assumed, we believe 3-10 sub-patches are more likely.;In the second chapter, using high-resolution relative low-frequency earthquake (LFE) locations, we calculate the patch areas (Ap) of LFE families. During Episodic Tremor and Slip (ETS) events, we define AT as the area that slips during LFEs and ST as the total amount of summed LFE slip. Using observed and calculated values for AP, AT and ST we evaluate two end- member models for LFE slip within an LFE family patch (models 2 and 3 from chapter 1). In the ductile matrix model (model 3), LFEs produce 100% of the observed ETS slip (SETS) in distinct sub-patches (i.e., AT AP). In the connected patch model (model 2), AT =AP, but ST<< SETS. LFEs cluster into 45 LFE families. Spatial gaps (∼10-20 km) between LFE family clusters and smaller gaps within LFE family clusters serve as evidence that LFE slip is heterogeneous on multiple spatial scales. We find that LFE slip only accounts for ∼0.2% of the slip within the slow slip zone. There are downdip trends in the characteristic (mean) moment and in the number of LFEs during both ETS events (only) and the entire ETS cycle (Mc,ETS and NT,ETS and Mc,all and NT,all respectively). During ETS, Mc decreases with downdip distance but NT does not change. Over the entire ETS cycle, Mc decreases with downdip distance, but NT increases. These observations indicate that downdip LFE slip occurs through a larger number (800-1200) of small LFEs, while updip LFE slip occurs primarily during ETS events through a smaller number (200-600) of larger LFEs. This could indicate that the plate interface is stronger and has a higher stress threshold updip.;In the third chapter, we use high-precision, relative low-frequency earthquake (LFE) locations for LFEs beneath the Olympic Peninsula, WA to constrain the depth, geometry, and thickness of the plate interface. LFE depths correspond most closely with the McCrory et al. (2012) plate model, but vary from that smooth model along strike. The latter observation indicates that the actual plate interface is notably rougher and more complex than smooth plate models. Our LFEs lie directly above low-velocity zone (LVZ) and approximately 5 km above intraslab earthquakes. This supports the proposal of Bostock (2013), that the LVZ comprises the upper oceanic crust and that fluids are responsible for the velocity contrast across the LVZ and likely play a large role in generating slow slip and LFEs. Within each of our LFE families, LFEs group into tight clusters around the family centroid. The width of these clusters in the depth direction, which is an indicator of the thickness of slow slip deformation on the plate interface, is 130 to 340 meters. |
