This paper reviews laboratory observations of earthquake initiation and describes new experiments on a 3‐m rock sample where the nucleation process is imaged in detail. Many of the laboratory observations are consistent with previous work that showed a slow and smoothly accelerating earthquake nucleation process that expands to a critical nucleation length scale Lc, before it rapidly accelerates to dynamic fault rupture. The experiments also highlight complexities not currently considered by most theoretical and numerical models. This includes a loading rate dependency where a “kick” above steady state produces smaller and more abrupt initiation. Heterogeneity of fault strength also causes abrupt initiation when creep fronts coalesce on a stuck patch that is somewhat stronger than the surrounding fault. Taken together, these two mechanisms suggest a rate‐dependent “cascade up” model for earthquake initiation. This model simultaneously accounts for foreshocks that are a by‐product of a larger nucleation process and similarities between initial P wave signatures of small and large earthquakes. A diversity of nucleation conditions are expected in the Earth's crust, ranging from slip limited environments with Lc < 1 m, to ignition‐limited environments with Lc > 10 km. In the latter case, Lc fails to fully characterize the initiation process since earthquakes nucleate not because a slipping patch reaches a critical length but because fault slip rate exceeds a critical power density needed to ignite dynamic rupture.

[1] We report on laboratory experiments which investigate interactions between aseismic slip, stress changes, and seismicity on a critically stressed fault during the nucleation of stick-slip instability. We monitor quasi-static and dynamic changes in local shear stress and fault slip with arrays of gages deployed along a simulated strike-slip fault (2 m long and 0.4 m deep) in a saw cut sample of Sierra White granite. With 14 piezoelectric sensors, we simultaneously monitor seismic signals produced during the nucleation phase and subsequent dynamic rupture. We observe localized aseismic fault slip in an approximately meter-sized zone in the center of the fault, while the ends of the fault remain locked. Clusters of high-frequency foreshocks (M w~À 6.5 to À5.0) can occur in this slowly slipping zone 5-50 ms prior to the initiation of dynamic rupture; their occurrence appears to be dependent on the rate at which local shear stress is applied to the fault. The meter-sized nucleation zone is generally consistent with theoretical estimates, but source radii of the foreshocks (2 to 70 mm) are 1 to 2 orders of magnitude smaller than the theoretical minimum length scale over which earthquake nucleation can occur. We propose that frictional stability and the transition between seismic and aseismic slip are modulated by local stressing rate and that fault sections, which would typically slip aseismically, may radiate seismic waves if they are rapidly stressed. Fault behavior of this type may provide physical insight into the mechanics of foreshocks, tremor, repeating earthquake sequences, and a minimum earthquake source dimension.

Recent modeling studies have explored whether earthquakes begin with a large aseismic nucleation process or initiate dynamically from the rapid growth of a smaller instability in a "cascade-up" process. To explore such a case in the laboratory, we study the initiation of dynamic rupture (stick slip) of a smooth saw-cut fault in a 76 mm diameter cylindrical granite laboratory sample at 40-120 MPa confining pressure. We use a high dynamic range recording system to directly compare the seismic waves radiated during the stick-slip event to those radiated from tiny (M À6) discrete seismic events, commonly known as acoustic emissions (AEs), that occur in the seconds prior to each large stick slip. The seismic moments, focal mechanisms, locations, and timing of the AEs all contribute to our understanding of their mechanics and provide us with information about the stick-slip nucleation process. In a sequence of 10 stick slips, the first few microseconds of the signals recorded from stick-slip instabilities are nearly indistinguishable from those of premonitory AEs. In this sense, it appears that each stick slip begins as an AE event that rapidly (~20 μs) grows about 2 orders of magnitude in linear dimension and ruptures the entire 150 mm length of the simulated fault. We also measure accelerating fault slip in the final seconds before stick slip. We estimate that this slip is at least 98% aseismic and that it both weakens the fault and produces AEs that will eventually cascade-up to initiate the larger dynamic rupture.

Recent geodetic observations indicate that a single fault may slip slowly and silently or fast and seismically in different circumstances. We report laboratory experiments that demonstrate how the mode of faulting can alternate between fast “stick‐slip” events (70 to ~500 mm/s sliding rates) and slow silent events (~0.001 to 30 mm/s) as a result of loading conditions rather than friction properties or stiffness of the loading machine. The 760 mm long granite sample is close to the critical nucleation length scale for unstable slip, so small variations in nucleation properties result in measurable differences in slip events. Slow events occur when instability cannot fully nucleate before reaching the sample ends. Dynamic events occur after long healing times or abrupt increases in loading rate which suggests that these factors shrink the spatial and temporal extents of the nucleation zone. Arrays of slip, strain, and ground motion sensors installed on the sample allow us to quantify seismic coupling and study details of premonitory slip and afterslip. We find that seismic coupling decreases when slip rates fall below about 70 mm/s. The slow slip events we observe are primarily aseismic (less than 1% of the seismic coupling of faster events) and produce swarms of very small M −6 to M −8 events. These mechanical and seismic interactions suggest that faults with transitional behavior—where creep, small earthquakes, and tremor are often observed—could become seismically coupled if loaded rapidly, either by a slow slip front or dynamic rupture of an earthquake that nucleated elsewhere.

This paper describes sensor calibration and signal analysis techniques applicable to the method of acoustic emission (AE) and ultrasonic testing. They are particularly useful for obtaining absolute measurements of AE wave amplitude and shape, which can be used to constrain the physics and mechanics of the AE source. We illustrate how to perform calibration tests on a thick plate and how to implement two different mechanical calibration sources: ball impact and glass capillary fracture. In this way, the instrument response function can be estimated from theory, without the need for a reference transducer. We demonstrate the methodology by comparing calibration results for four different piezoelectric acoustic emission sensors: Physical Acoustics (PAC) PAC R15, PAC NANO30, DigitalWave B1025, and the Glaser-type conical sensor. From the results of these tests, sensor aperture effects are quantified and the accuracy of calibration source models is verified. Finally, this paper describes how the effects of the sensor can be modeled using an autoregressive-moving average (ARMA) model, and how this technique can be used to effectively remove sensor-induced distortion so that a displacement time history can be retrieved from recorded signals.

Faults strengthen or heal with time in stationary contact, and this healing may be an essential ingredient for the generation of earthquakes. In the laboratory, healing is thought to be the result of thermally activated mechanisms that weld together micrometre-sized asperity contacts on the fault surface, but the relationship between laboratory measures of fault healing and the seismically observable properties of earthquakes is at present not well defined. Here we report on laboratory experiments and seismological observations that show how the spectral properties of earthquakes vary as a function of fault healing time. In the laboratory, we find that increased healing causes a disproportionately large amount of high-frequency seismic radiation to be produced during fault rupture. We observe a similar connection between earthquake spectra and recurrence time for repeating earthquake sequences on natural faults. Healing rates depend on pressure, temperature and mineralogy, so the connection between seismicity and healing may help to explain recent observations of large megathrust earthquakes which indicate that energetic, high-frequency seismic radiation originates from locations that are distinct from the geodetically inferred locations of large-amplitude fault slip.

Ball impact has long been used as a repeatable source of stress waves in solids. The amplitude and frequency content of the waves are a function of the force-time history, or force pulse, that the ball imposes on the massive body. In this study, Glaser-type conical piezoelectric sensors are used to measure vibrations induced by a ball colliding with a massive plate. These measurements are compared with theoretical estimates derived from a marriage of Hertz theory and elastic wave propagation. The match between experiment and theory is so close that it not only facilitates the absolute calibration the sensors but it also allows the limits of Hertz theory to be probed. Glass, ruby and hardened steel balls 0.4 to 2.5 mm in diameter were dropped onto steel, glass, aluminum, and polymethylmethacrylate plates at a wide range of approach velocities, delivering frequencies up to 1.5 MHz into these materials. Effects of surface properties and yielding of the plate material were analyzed via the resulting stress waves and simultaneous measurements of the ball's coefficient of restitution. The sensors are sensitive to surface normal displacements down to about +/-1 pm in the frequency range of 20 kHz to over 1 MHz.

Earthquakes are dynamic rupture events that initiate, propagate, and terminate on faults within the Earth's crust. Understanding rupture termination is essential for accurately estimating the maximum magnitude earthquake a region might experience. We study termination on sequences of M − 2.5 earthquakes that rupture a 3‐m granite laboratory sample. At this large scale, nucleation, propagation, and termination are either completely or partially confined within the sample–unique observations for experiments on rock. We compare measured termination locations to estimates from a fracture mechanics‐based model to quantify the fracture energy of the laboratory earthquakes, which compare well with estimates from small natural quakes. Our results provide a mathematical framework that links micrometer‐scale friction parameters to meter‐scale earthquake mechanics, shows that a 3‐m slab of granite can behave similar to a 200‐mm sheet of glassy polymer, and demonstrates how small events can prime a fault for larger, damaging ones.