T43C-3008
Slow Slip Events on a 760 mm Long Granite Sample

Thursday, 17 December 2015
Poster Hall (Moscone South)
Gregory Mclaskey, USGS California Water Science Center Menlo Park, Menlo Park, CA, United States and Futoshi Yamashita, National Research Institute for Earth Science and Disaster Prevention, Tsukuba, Japan
Abstract:
We describe slow slip events and dynamic rupture events generated on a newly constructed large-scale biaxial friction apparatus at Cornell University that provide insights into the mechanisms of aseismic and seismic slip. We find that, under nominally similar experimental conditions, the 760 mm long granite sample sometimes slips in dynamic stick-slip events and sometimes relieves accumulated shear stress through slow slip events. To provide insights into this curious behavior and the underlying mechanisms, fault slip and shear stress are each measured at 8 locations along the 760 mm long fault. This allows us to map slow slip fronts and the nucleation and propagation of dynamic fault rupture. The granite sample is also instrumented with an array of piezoelectric sensors that are the laboratory equivalent of a seismic network. When the sample is loaded relatively slowly, at 0.03 MPa/s, slow slip occurs on large sections of the fault and the slow slipping region soon expands to the sample boundary. In this case, stress is released in a slow slip event with peak slip velocities < 2 mm/s. Alternatively, when one end of the sample is loaded rapidly (4 MPa/s), or the sample is allowed to heal in stationary contact for a few minutes, slow slip initiates near the load point and accelerates to slip velocities exceeding 200 mm/s before the slow slipping region expands all the way to the sample boundary. This produces a dynamic slip event (stick-slip). The dynamic slip events radiate seismic waves equivalent to a M = -2.5 earthquake. In contrast, the laboratory-generated slow slip events are predominantly aseismic and produce only bursts of tiny and discrete seismic events (M = -6) reminiscent of swarms of microseismicity. The experiments illustrate how a single fault can slide slowly and aseismically or rapidly and dynamically depending on stress state and loading conditions. We compare the behavior observed on this Cornell apparatus to the behavior of other large-scale biaxial machines at the NIED in Tsukuba, Japan, and at the USGS in Menlo Park, California. While the Cornell machine is smaller than the NIED (1.5 m) and USGS (2.0 m) machines, the stress levels are higher (14 MPa compared to 7 MPa), and the relatively high stiffness (6 GPa/m) of the loading frame likely promotes slow slip.