Search

Friction is a complex phenomenon. This can be seen, for example, in laboratory experiments where stick-slip motion of various kind (i.e., slow and fast instabilities) can be produced when adapting the normal stress applied to the system. Similarly, natural earthquakes also producecomplex stick-slip behaviour. A first challenge in the description of friction comes from the potentially high number of degrees of freedom (dofs) involved in the description of the dynamics of the sliding surfaces. Nonetheless, it was shown that friction can be described with a reduced number of dofs or variables of the dynamics. These may include the shear stress, the relative sliding slip rate, and one or more variables that describe the state of the contact of the sliding surfaces. We investigate the possibility to extract directly from the data the governing equations of friction starting from a simplified synthetic example. We further study the laboratory data with the Hankel Alternative View Of Koopman (HAVOK) theory, a method rooted in dynamical system theory that leverages data driven techniques and produces a Reduced Order Model (ROM) to reconstruct a shadow of the attractor of a system from observational data. We finally compare the results obtained for the laboratory experiments with Cascadia slow earthquakes.

Pore fluids are ubiquitous throughout the lithosphere and are commonly cited as a major factor producing slow slip and complex modes of tectonic faulting. Here, we investigate the role of pore pressure on slow slip and the frictional stability transition and find that the mode of fault slip is largely unaffected by pore pressure once we account for effective stress. Ambient temperature experiments are done on synthetic fault gouge composed of quartz powder with a median grain size of 10μm with an average permeability of 8E-17m2 – 6E-18m2 from shear strains 0 - 26. We conduct constant velocity experiments at 20MPa σn’, with Pp/σn’ratios of λ from 0.05 to 0.28. Under these conditions, dilatancy strengthening is minimal and we find that slip rate dependent changes in the critical rate of frictional weakening are sufficient to explain slow slip.

Serpentinites are poly-mineralic rocks distributed almost ubiquitously in active tectonic regions worldwide. They are composed of rheologically weak (lizardite and crysotile) and strong (e.g., magnetite and pyroxene) phases. In particular, lizardite typically shows low friction coefficients and is supposed to localise deformation along weak shear zones characterized by aseismic behaviour. Major faults hosting serpentinite lithologies are characterised by seismic activity, tremors, and other slip modes. We advance the hypothesis that low strain domains, which are enriched in rheologically strong phases, can act as potential site of nucleation of unstable slip as the result of the velocity-dependent rheology of magnetite-rich serpentinites. Through an experimental and microstructural approach, we explore the different mechanisms whose interplay controls the complex behaviour of these lithologies.For this study we collected natural samples of lizardite-magnetite rich serpentinites within the low strain domains of the Elba Island ophiolites (Italy). Rocks were characterised, powdered, and deformed in a set of shear experiments at four different normal stresses (25, 50, 75 and 100 MPa) in the biaxial apparatus BRAVA. The experiments consist of an initial phase of sliding at 10 μm/s, a slide-hold-slide test, and two series of velocity stepping (sliding velocity from 0.1 to 300 μm/s). Fundamental parameters to quantify the frictional properties of serpentinites are individuated in the (a-b) value, the critical slip distance Dc, and the critical stiffness kc, which is derived by their combination.The material shows friction values of ~0.4 with velocity weakening behaviour and negative frictional healing. The module of the negative (a-b) parameter increases neatly with decreasing sliding velocity while Dc decreases, causing kc to rise. At low velocities (< 3 μm/s) sliding is unstable and the fault undergoes stick-slip behaviour. This is explained by the increase of the critical stiffness to values higher than the loading system stiffness. This systematic change of mechanical properties and fault slip behaviours with sliding velocity is interpreted to be the result of the time-dependent arrangement of grains in a heterogeneous experimental fault architecture.Back-scattered SEM images of the principal slip zones of recovered samples support this hypothesis. Elasto-frictional behaviour is controlled by the build-up of a partial (granular) load-bearing framework of strong magnetite grains, while visco-frictional rheology is controlled by the (phyllosilicatic) anastomosing and foliated lizardite matrix. At low sliding velocities, the granular phase interacts creating force chains thus promoting frictional instabilities. At higher velocities, dilation promotes the activity of throughgoing weaker phyllosilicate planes thus favouring stable slip.Our experiments shed light on the role of fault rock heterogeneity in nucleating dynamic slip in nature as well as in controlling the slip mode during earthquakes or slow-slip events in serpentinite terrains.

After many large earthquakes aftershocks activity can reach depths greater than the base of the seismogenic zone that is defined by background seismic activity. This observation is generally explained by strain rate induced embrittlement associated with the increase of post-mainshock strain rate, which favors a transition from ductile to brittle behavior. However, the underlying physical processes are not well understood. Here we integrate geological and geophysical data for the 2016–2017 Central Italy seismic sequence with laboratory experiments to provide a geological and physical interpretation for the post-mainshock transient deepening of the base of the seismogenic zone.The base of the seismogenic zone in the central-northern Apennines is set typically at 9-10 kilometers and corresponds to the top of the phyllitic basement. Structural studies on exhumed basement rocks highlight a heterogeneous basement fabric consisting of competent, 10 to 200 m wide, quartz-rich lenses surrounded by an interconnected and frictionally weak phyllosilicate-rich matrix. The matrix controls the overall rheology of the basement due to its interconnectivity, and promotes aseismic deformation because its rate-strengthening behavior.Following each major (Mw > 5.5) event of the 2016–2017 sequence, a dramatic and abrupt increase in seismic rate is observed below 10 km, hence within the basement. Here we document the presence of seismicity clusters made of more than 4 earthquakes and characterized by small magnitude (Mw < 2.5), small dimensions (< 500 meters), small temporal duration (< 14 days) and a swarm-like behavior. Furthermore, these clusters are often represented by multiple or repeating earthquakes with a cross correlation coefficient higher than 0.7 for all the three components. These observations suggest that the increase of shear stressing rate within the basement is responsible for deepening of seismicity. To further explore this idea, we performed laboratory experiments on rocks from exhumed outcrops of basement rocks. We found that shear stressing rate promotes accelerated creep on the phyllosilicate-rich matrix and dynamic instabilities on the quartz-rich gouge belonging to the lenses. Our integrated analysis suggests that the mainshocks of the 2016-2017 seismic sequence promoted an increase of shear stressing rate within the basement allowing the phyllosilicate-rich matrix to creep faster hence favoring the loading and the repeated failures of locked seismogenic patches represented by the quartz-rich lenses.

Natural earthquakes occur in clusters or sequences that arise from complex triggering mechanisms, but direct measurement of the mechanisms responsible for complex temporal sequences and delayed triggering is rarely possible. A central question involved whether delayed triggering is due to slow slip and stress transfer or local weakening/fatigue processes such as stress corrosion. We investigate the origins of this complexity and its relationship to fault heterogeneity using a biaxial loading apparatus with an experimental fault that has two dominant seismic asperities. The fault is composed of a 5 mm layer of quartz powder, a velocity weakening material common to natural faults, sandwiched between 760 mm long polymer blocks that deform similar to the way 10 meters of rock would behave. Due to the higher local normal stress and the free surface boundary condition on the sample ends, the sample behaves like two asperities, one at each end, that can fail independently. As the quartz powder was continuously sheared, the friction properties changed, and we observed a transition from steady sliding to periodic repeating earthquakes that transitioned into aperiodic and complex sequences of fast and slow events. There is also reason to believe that friction properties evolved differently on the higher normal stress asperities and made them more unstable than the center part of the laboratory sample. Sequential ruptures on the two different asperities were linked via migrating slow slip which resembles creep fronts observed in numerical simulations and on tectonic faults. The propagation velocity of the creep fronts ranged from 0.1 to 10 m/s, which is broadly consistent with the velocity of slow slip fronts inferred from migrating tectonic tremor sources. Utilizing both local stress measurements and numerical simulations, we observe that the speed and strength of creep fronts are highly sensitive to fault stress levels left behind by previous earthquakes and may serve as on-fault stress meters.

In the last decades, it has been observed that faults can slip both by slow aseismic creep and seismic events (i.e., earthquakes). Between these two slip modes, a wide variety of fault slip behavior can be observed, including low-frequency earthquakes, slow slip events and tremors. This wide variety of slip modes can radiate seismic energy at different frequencies whose content may be linked to the physical mechanisms at play. In the laboratory, it is possible to reproduce the entire spectrum of fault slip modes by modulating the loading stiffness of the apparatus depending on the critical fault rheologic stiffness (i.e. k/kc). This technique allows us to study, under controlled laboratory conditions, the acoustic signature of different fault slip modes to infer the physical mechanisms at their origin. To shed light on the nucleation mechanisms and seek for reliable precursors to failure of different slip modes, we performed friction experiments on powders that differ for granulometry and grain shape (i.e., glass beads with a grainsize < 150 µm; and quartz powders MinUSil with an average grain size of 10.5 µm), to simulate fault gouge. The experiments were conducted in a double direct shear configuration, instrumented with an array of piezoelectric sensors to record continuously Acoustic Emissions (AEs) at high recording rate (~10MHz). The experiments are performed at a constant displacement rate of 10 µm/s and using a spring to reduce the apparatus stiffness k, to match the critical fault rheological stiffness (kc). Following this procedure we obtain slow slip events (i.e., k = kc) and fast events (i.e. k

A central goal of this work is to understand the extent to which fault friction vs. pore fluids and other factors is the cause of slow earthquakes and the spectrum of fault slip behaviors. Slow earthquakes and quasi-dynamic modes of fault slip such as tremor and LFEs have now been observed in essentially every tectonic setting, which suggests that the underlying mechanism(s) are generic rather than specific to a particular fault setting, rock type, or tectonic regime. Here, I discuss lab data that illuminate the mechanics of slow slip. I focus on frictional stick-slip failure events, the lab equivalent of earthquakes, that reproduce slow slip and the full range of slip behaviors observed on tectonic faults. These studies document repetitive slip events and the complete lab seismic cycle for the full spectrum of slip behaviors from aseismic creep to slow slip and aperiodic elastodynamic failure. Working with data for repetitive slip events is critical for understanding the underlying mechanics. In the lab, we also document the rate of frictional weakening with slip kc = σn (b-a)/Dc ––the so-called critical stiffness–– where σn is fault normal stress, (b-a) is the friction rate parameter and Dc is the critical slip distance. We measure kc for the same conditions of the slow slip events by altering the machine loading stiffness. These works assess directly frictional instability theory, which predicts the slip stability transition when the elastic stiffness of the fault zone k equals the frictional weakening rate kc. The lab work confirms friction theory in relation to the transition from stable to unstable slip but it also reveals additional complexity showing that kc varies with slip rate. Several works now document the velocity dependence of kc(V) and its role in dictating slow slip in the lab. These works show complex behavior near k/kc ≈ 1 including slow slip, aperiodic failure and chaotic motion. I discuss these results in the context of basic questions that remain regarding how slow ruptures can propagate quasi-dynamically, at speeds far below the Rayleigh wave speed, and how tectonic faults can host both slow slip and dynamic earthquake rupture.

Seismic waves contain information about the earthquake (EQ) source and many forms of noise deriving from the seismometer, anthropogenic effects, background noise associated with ocean waves, and microseismic noise. Separating the noise from the EQ signal is a critical first step in EQ physics and seismic waveform analysis. However, this is difficult because optimal parameters for filtering noise typically vary with time and may strongly alter the shape of the waveform. A few recent works have employed Deep Learning (DL) model for seismic denoising, among which we have taken as a benchmark Deep Denoiser and SEDENOSS. These models turn the noisy trace into a 2D signal (spectrograms) within the model to denoise the traces, making the process pretty heavy. We propose a novel DL-powered seismic denoising algorithm based on Diffusion Models (DMs), keeping the signal in 1D. DMs are the latest trend in Machine Learning (ML), having revolutionized the application fields of audio and image processing for denoising (DiffWave), synthesis (Stable Diffusion), and sequence modeling (STARS). The training of DMs proceeds by polluting a signal with noise until the signal has completely vanished into noise, then reversing the process by iterative denoising, conditioned on the latent signal representation. This makes DMs the ideal tool for seismic traces cleaning, as the model naturally learns from seismic sequences by denoising, which aligns the ML training procedure and the final task objective. In a preliminary evaluation, we used the Stanford Earthquake Dataset (STEAD); our proposed Diffusion-based Seismic Denoiser (DiffSD) outperforms the state-of-the-art DL methods on the Signal Noise Ratio (SNR), Scale-Invariant Source to Distortion Ratio (SI-SDR), and Source to Distortion Ratio (SDR) metrics. DiffSD also yields qualitatively pleasing EQ traces out of visual inspection in time and frequency. Finally, DiffSD proceeds from regenerating clean EQ signals from noise, which opens the way to data-driven EQ sequence generations, potentially instrumental to further study and dataset augmentations.

Large earthquakes take place on mature faults with hundreds of meters to kilometres of cumulative slip. At shallow depths, the fault zone is generally composed of non-cohesive rock wear products, often referred to as gouge. Seismic and aseismic slip occur in this fault gouge and fracture/brecciation of the wall rock and damage zone can add to the fault gouge as part of the wear process. Gouge thickness generally increases linearly with the cumulative fault shear displacement and laboratory work shows that gouge tends to stabilize fault frictional stability. Previous works show that frictional stability of simulated fault gouge varies as a function of shear displacement. The stability evolution is interpreted as a consequence of the degree of shear localisation within the simulated fault gouge: the more the deformation is localized, the more the fault slip is unstable. This implies that for bare rock surfaces, unstable behaviour is expected as the deformations are forced to be localized at the interface between the two sheared surfaces.On natural faults at large shear displacement (or for faults having a high gouge production rate), a competition must take place between 1) the localization of the deformation at rock-on-rock surfaces, 2) the delocalization of deformation due to gouge production and wall rock brecciation, 3) fault zone lithification and frictional healing and 4) shear localization within the gouge and wear material. The competition and interaction between these phenomena are modulated by cumulative fault slip, temperature and fluid chemistry. In turn, this competition may influence the frictional stability of faults with increasing shear displacement, and thus, their potential seismic activity.To characterise the influence of shear displacement on fault stability, constant velocity and velocity step experiments were performed to large displacement. Two initially intact rocks were chosen as starting material: a high porosity Fontainebleau sandstone and a low porosity quartzite. These samples represent very different resistances to abrasion (i.e., wear production with slip) for the same initial mineral composition (< 95% quartz), which allows us to investigate wear and wear rate on fault stability. Additionally, simulated quartz gouge was tested for comparison. Mechanical data are analysed within the rate-and-state framework, and post-mortem microscopic analyses of the sample were performed. For initially bare surface experiments a threshold shear displacement is required to transition from stable to unstable sliding. Stick-slip events (laboratory earthquakes) evolve systematically as a function of fault zone shear displacement. The inversion of the rate-and-state parameters shows that shear displacement has a dominant influence on both (a-b) and Dc. For all the faults tested, (a-b) decreases with increasing shear displacement. For high wear rates and simulated gouge, Dc decreases with increasing shear displacement. However, for low wear rate faults, Dc is constant within the tested shear displacement. These results demonstrate that, under the tested boundary conditions, fault stability varies systematically with fault maturity and in particular that shear displacement and strain localization are the dominant parameters controlling fault slip stability.

Understanding the poromechanical response of fractured rock is necessary for applications ranging from hydraulic stimulation of the subsurface to teleseismic impulses from earthquakes that may reactivate faults or otherwise breach reservoir seals. We describe laboratory experiments that seek to decouple the role of fracture interface stiffness and fracture infill (sediment transport) in the hydraulic and elastodyamic properties of fractured rock. Experiments are conducted on multiple samples of Westerly granite with different uniform roughness (silicon carbide grit-roughened or milled) that were loaded under triaxial stresses in a pressure vessel while permeability evolution is measured from the flow-through of deionized water. In some experiments, thin layers of synthetic wear (gouge) material are added to the fracture interface to simulate mature, sheared, fractures whose poromechanical response is dominated by clogging and unclogging of pore throats. Oscillations of pore pressure and normal stress are applied at amplitudes ranging from 0.2 to 1 MPa at 1Hz. The experiments also consider the influence of fracture aperture with effective stress perturbations applied at normal stresses ranging from 5 to 20 MPa (reducing aperture with increasing effective normal stress). Before, during, and after the dynamic stressing, an array of piezoelectric transducers (PZTs) continuously transmits and receives ultrasonic pulses across the fracture to monitor the evolution of fracture stiffness and fluid transport due to the dynamic stressing. These allow evaluation of stress-induced changes in transmitted ultrasonic wave velocity and amplitude to estimate the contact acoustic nonlinearity of the fracture interface concurrent with permeability evolution. We compare the results for samples with and without synthetic wear (gouge) material to understand the role and evolution of fracture stiffness and clogging-unclogging mechanisms in pore throats of porous and fractured media.

Fault zone properties can change significantly during the seismic cycle in response to stress changes, microcracking and wall rock damage. Lab experiments show consistent changes in elastic properties prior to and after lab earthquakes (EQ) and previous works show that machine learning/deep learning (ML/DL) techniques are successful for capturing such changes. Here, we apply DL techniques to assess whether similar changes occur during the seismic cycle of tectonic EQ. The main motivation is to generalize lab-based findings to tectonic faulting, to predict failure and identify precursors. The novelty is that we use EQ traces as probing signals to estimate the fault state.We train DL model to distinguish foreshocks, aftershocks and time to failure of the Mw 6.5 2016 Norcia EQ in central Italy, October 30th 2016. We analyze a 25-second window of 3-component data around the P- and S-wave arrivals for events near the Norcia fault with M>0.5 and ±2 months before/after the Norcia mainshock. Normalized waveforms are used to train a Convolutional Neural Network (CNN). As a first task we divide events into two classes (foreshocks/aftershocks), and then refine the classification as a function of time-to-failure (TTF) for the mainshock. Our DL model perform very well for TTF classification into 2, 4, 8, or 9-classes for the 2 months before/after the mainshock. We explore a range of seismic ray paths near, through, and away from the Norcia mainshock fault zone. Model performance exceeds 90% for most stations. Waveform investigations show that wave amplitude is not the key factor; other waveform properties dictate model performance. Models derived from seismic spectra, rather than time-domain data, are equally good. We challenged the model in several ways to confirm the results. We found reduced performance in training the model with the wrong mainshock time and by omitting data immediately before/after the mainshock. Foreshock/aftershock identification is significantly degraded also by removing high frequencies (filtering seismic data above 25 Hz). We tested data from different years to understand seasonality at individual stations for the time period September to December and removed these effects. Comparing these seasonality effects defined from noise with our EQ results shows that foreshocks/aftershocks for the 2016 Norcia mainshock are well resolved. Training with data containing EQ offers a huge increase in classification performance over noise only, proving that EQ signals are the sole that enable assessing timing as a function of the fault status. To confirm our results and understand which stations are able to detect changes of fault properties we perform a further test cleaning the signals from the seasonality by confounding the DL with a shuffled noise (adversarial training).We conclude that DL is able to recognize variations in the stress state and fracture during the seismic cycle. The model uses EQ-induced changes in seismic attenuation to distinguish foreshocks from aftershocks and time to failure. This is an important step in ongoing efforts to improve EQ prediction and precursor identification through the use of ML and DL.

We analyze frictional motion for a laboratory fault as it passes through the stability transition from stable sliding to unstable motion. We study frictional stick-slip events, which are the lab equivalent of earthquakes, via dynamical system tools in order to retrieve information on the underlying dynamics and to assess whether there are dynamical changes associated with the transition from stable to unstable motion. We find that the lab seismic cycles exhibit characteristics of a low-dimensional system with average dimension similar to that of natural slow earthquakes (10, indicating that some regions of the phase space require a high number of degrees of freedom (dofs). Our analysis does not preclude deterministic chaos, but the lab seismic cycle is best explained by a random attractor based on rate- and state-dependent friction whose dynamics is stochastically perturbed. We find that minimal variations of 0.05% of the shear and normal stresses applied to the experimental fault influence the large-scale dynamics and the recurrence time of labquakes. While complicated motion including period doubling is observed near the stability transition, even in the fully unstable regime we do not observe truly periodic behavior. Friction's nonlinear nature amplifies small scale perturbations, reducing the predictability of the otherwise periodic macroscopic dynamics. As applied to tectonic faults, our results imply that even small stress field fluctuations (less or about 150 kPa) can induce coefficient of variations in earthquake repeat time of a few percent. Moreover, these perturbations can drive an otherwise fast-slipping fault, close to the critical stability condition, into a mixed behavior involving slow and fast ruptures.

We exploit nonlinear elastodynamic properties of fractured rock to probe the micro-scale mechanics of fractures and understand the relation between fluid transport and fracture aperture and area, stiffness proxy, under dynamic stressing. Experiments are conducted on rough, tensile-fractured Westerly granite specimen subject to triaxial stresses. Fracture permeability is measured from steady-state fluid flow with deionized water. Pore pressure oscillations are applied at amplitudes ranging from 0.2 to 1~MPa at 1~Hz frequency. During dynamic stressing we transmit acoustic signals through the fracture using an array of piezoelectric transducers (PZTs) to monitor the evolution of fracture interface properties. We examine the influence of fracture aperture and contact area by conducting measurements at effective normal stresses of 10, 12.5, 15, 17.5, and 20~MPa. Additionally, the evolution of contact area with stress is characterized using pressure sensitive film. These experiments are conducted separately with the same fracture and they map contact area at stresses from 9 to 21~MPa. The resulting ‘true’ area of contact measurements made for the entire fracture surface and within the calculated PZT sensor footprints, numerical modeling of Fresnel zone. We compare the elastodynamic response of the the fracture using the stress-induced changes ultrasonic wave velocities for a range of transmitter-receiver pairs to image spatial variations in contact properties, which is informed by fracture contact area measurements. These measurements of the nonlinear elasticity are related to the fluid-flow, permeability, in response to dynamic stressing and similar comparisons are made for the slow-dynamics, recovery, of the fracture interface following the stress perturbations.

Nonlinear elastodynamic response of fractured rocks carries crucial information on fracture features that can be exploited to forecast flow properties, friction constitutive behavior and poromechanical response. Well-controlled laboratory experiments are designed to measure the nonlinear elastodynamic response of Westerly granite in three states: dry intact, dry fractured and saturated fractured. We study the effect of fracturing and saturation in modifying the elastodynamic response of the rock. Each sample is tested at a normal stress level of 15 MPa. We measure the elastodynamic response of an intact L-shaped sample of Westerly Granite subjected to normal stress oscillations of prescribed amplitudes (0.2–1 MPa) and frequencies (0.1, 1, 10 Hz). Ultrasonic waves transmitted across the sample are used to monitor the evolution of wave velocity before, during and after dynamic stressing. The nonlinearity of the elastodynamic response is measured in terms of: (1) the offset in normalized wave velocity; (2) the amplitude of wave velocity fluctuation during the oscillations; and (3) recovery rate of the wave velocity post-oscillation. We observe that the three nonlinearity parameters show a similar trend. Irrespective of the parameter, the nonlinearity measures higher for sample in dry-intact condition than that for dry-fractured and the saturated-fractured sample exhibits smaller nonlinearity than the dry-fractured sample. As expected, the saturated sample exhibits less nonlinearity than the dry intact and fractured samples due to the presence of interstitial fluid and the resulting increased interface stiffness. Conversely, the dry intact rock shows a higher nonlinearity than the dry fractured. We use numerical simulations to show that the presence of fracture significantly alters the strain distribution across the bulk of the sample and only the contacting asperities are highly strained, thus resulting in a decrease in the measured elastodynamic nonlinearity.

Understanding the connection between seismic activity and the earthquake nucleation process is a fundamental goal in earthquake seismology with important implications for earthquake early warning systems and forecasting. We use high-resolution acoustic emission (AE) waveform measurements from laboratory stick-slip experiments that span a spectrum of slow to fast slip rates to probe laboratory foreshocks and nucleation. We measure the waveform similarity of AE templates and use differential travel-times to track their relative locations. Fast laboratory earthquakes are preceded by a late, rapid increase in waveform similarity prior to failure, whereas slow slip events show a modest increase in waveform similarity before failure. Differential travel-time and waveform similarity measurements reveal a spatiotemporal coalescence of foreshocks prior to failure. Our work suggests that laboratory foreshocks evolve systematically prior to stick-slip failure and are a byproduct of a slow nucleation process driven by pre-seismic fault slip.

A growing body of observations worldwide has documented fault slip transients that radiate little or no seismic energy. The mechanisms that govern these slow slip events (SSEs) and their wide range of depths, slip rates, durations, stress drops and recurrence intervals remain poorly known. Here we show that slow slip can be explained by a transition from rate-weakening frictional sliding at low slip rates towards rate-neutral or rate-strengthening behaviour at higher slip rates, as has been observed experimentally. We use numerical simulations to illustrate that this rate-dependent transition quantitatively explains the experimental data for natural fault rocks representative of materials in the source regions of SSEs. With a standard constant-parameter rate-and-state friction law, SSEs arise only near the threshold for slip instability. The inclusion of velocity-dependent friction parameters substantially broadens the range of conditions for slow slip occurrence, and produces a wide range of event characteristics, which include stress drop, duration and recurrence, as observed in nature. Upscaled numerical simulations that incorporate parameters consistent with laboratory measurements can reproduce geodetic observations of repeating SSEs on tectonic faults. We conclude that slip-rate-dependent friction explains the ubiquitous occurrence of SSEs in a broad range of geological environments. A transition from rate-weakening to rate-strengthening frictional behaviour with increasing slip rate could explain the observed diversity of slow slip events on faults, according to numerical simulations.

Earthquakes are a complex natural phenomenon. They typically are the result of frictional instabilities along preexisting weakness zones called faults. The strain slowly builds up in the fragile Earth crust because of the presence of an external loading counterbalanced by friction forces at the faults’ interface. When the load cannot be balanced by the friction any further, the fault slips releasing the accumulated strain. Friction is a nonlinear phenomenon, and as such frictionally controlled systems may be subject to chaotic behavior. Seismic cycle analogs can be reproduced with rock friction experiments in the laboratory with a double direct shear apparatus. We show that laboratory earthquakes follow a low-dimensional random attractor. We explain the observations with a model of stochastic differential equations based on the rate- and state-friction framework. We show that small perturbations (less than 1‰) on the shear and normal stress can induce laboratory earthquakes aperiodic behavior with coefficient of variations of the order of some percent. The nonlinear nature of friction amplifies small scale perturbations, making mid-long term predictions of the system possible only statistically even for stick-slip events in a well controlled environment like the laboratory.

I summarize a broad suite of laboratory data sets showing that stick-slip failure events –lab earthquakes– are commonly preceded by both measurable changes in fault zone properties and acoustic emission (AE) events that foretell catastrophic failure. These works show that both types of data can be used to predict labquakes with machine learning (ML) methods and deep learning (DL) approaches. The first works used continuous measurements of AE to predict the timing of labquakes and the fault zone shear stress. Subsequent studies showed that catalogs of AE events could also predict labquakes and that ML approaches could also predict stress drop, peak fault slip velocity and the duration of failure. Recently, DL has been used to predict and autoregressively forecast labquakes and fault zone shear stress. Consistent with previous works, we see that seismic b-value begins to decrease as faults unlock and start to creep. This provides a sensible connection between the ML-based predictions, fault zone elastic properties, and the physics of failure. In the lab, AE events represent a form of foreshock and, not surprisingly, the rate of foreshock activity correlates with fault slip rate and its acceleration toward failure. Our work shows precursory changes in wave speed prior to labquakes, consistent with many well known past studies, but the early studies did not provide a method to predict impending failure. ML and DL predicts with fidelity the time of impending failure and other aspects of it. This suggests the possibility of physics-based models for prediction. We are working to connect ML prediction of labquakes with the evolution of fault zone elastic properties, frictional contact mechanics and constitutive laws. A central goal is to learn from lab earthquake prediction to improve forecasts of earthquake precursors and tectonic faulting.

Pore fluids are ubiquitous throughout the lithosphere and are a major part of the stress distribution along faults. Elevated pore fluid pressure reduces the effective normal stress, allowing slip and potentially changing the mode of faulting. Mature brittle faults are characterized by deca- to hecto-meter damage zones composed of gouge and complex shear localization fabrics that can host zones of low, anisotropic permeability. Such zones can include undrained pore fluid conditions that may result in a spectrum of slip behaviors including slow slip events. Despite the obvious importance of pore fluids for fault mechanics, their role in dictating fault stability is poorly constrained. Early results for rate-strengthening accretionary wedge materials suggest pore fluid has a stabilizing effect, as the friction parameter (a - b) increases in response to increased pore fluid pressure (Pf ). Here, we describe early stages of a laboratory investigation of the role of fluid pressure on friction rate and memory effects. We present experimental results from rate-weakening synthetic gouge samples at a range of pore fluid pressure conditions. Experiments use a servo-controlled biaxial load frame enclosing a pressure vessel to apply a true triaxial stress-state with pore fluid pressures. Samples are assembled in a double-direct shear configuration with two uniform 3-millimeter-thick gouge layers. Sample forcing blocks include shear wave piezoelectric transducers for ultrasonic monitoring of shear wave amplitude and velocity. We conducted stable sliding experiments at both drained and undrained conditions to explore role of pore fluids on the RSF parameters. Undrained stick-slip experiments were also done at a range of pore fluid pressures to investigate the role of fluid pressure on the nature of fault slip. We explore differences between the drained and undrained conditions with particular attention on changes from rate-weakening to rate-strengthing friction behavior due to localized overpressure. Additionally, we evaluate interplay between the modes of fault slip, due to poroelastic processes which will result in changes in the transmitted shear wave amplitude and velocity. In subduction environments pore fluid pressure can approach lithostatic pressures leading to localized overpressure. Therefore, it is important to understand the contributions of fluids and effective stress state on frictional stability and the mode of fault slip, whether it be aseismic creep, slow slip, or earthquake rupture.

The focus of this study is to elucidate the relation between elastodynamic and hydraulic properties of fractured rock subjected to local stress perturbations in relation to fracture aperture distribution. The goal of our integrated numerical and experimental investigations is to understand the mechanisms responsible for changes in fault zone permeability and elasticity in response to dynamic stressing in the subsurface (anthropogenic or seismic in origin). High-resolution (micron-scale) optical profilometry measurements combined with pressure sensitive films have been used to characterize fracture properties such as ‘true’ contact area, aperture distribution and morphology, as well as asperity deformation under applied loads in our experiments. These measurements allow a direct correlation between fracture properties and our lab measurements of fracture elastic nonlinearity and permeability. Using micron-resolution profilometry of centimeter-scale samples, we calculate the elastic deformation of fracture asperities to varying applied stresses (static and dynamic) using Hertzian contact mechanics. Then, permeability is calculated for each applied stress (deformed asperities) using the parallel plate approximation, in which the Reynolds equation is solved using the finite difference method. This study is uniquely constrained, wherein we investigate the effect of measured deformation of real asperities on creating flow pathways through a fracture. Future work will include implementing contact acoustic nonlinearity (CAN) to model the change in transmission of acoustic waves across the fracture interface during stress perturbation.

It is now recognized that acoustic emissions (AEs) generated during stick-slip frictional sliding (i.e., lab earthquakes) can be considered as microearthquakes. Over the past decades, many laboratory AE studies have addressed issues related to the physics of earthquakes such as fault nucleation and growth in brittle rocks, frequency magnitude statistics of earthquakes, laboratory earthquake precursors and, more recently, laboratory earthquake prediction based on machine learning techniques. Here, we conduct double direct shear experiments on samples of Westerly granite under applied normal loads of 5-15 MPa and with shearing rates of 1-100 μm/s. We use template matching and other cross correlation techniques to study the evolution of AE similarity during the laboratory seismic cycle. The aim of this study is to connect changes in AE similarity to fault stress-loading and kinematics. AE similarity is derived from the correlation matrices of AE catalogs and is found to vary primarily with fault slip velocity. AE similarity is, on average, constant at slow speed (fault slip velocity

Tectonic fault zones are subject to normal stress variations with a wide range of spatiotemporal scales, resulting in stress field alteration. These perturbations can spread over a wide range of frequencies and amplitudes from the high frequency passage of seismic waves generated by earthquakes, to the low frequency of solid earth tides and underground fluid injection cycles. As a result of these normal stress perturbations, critically stressed faults can be reactivated. The resulting slip mode is then controlled by fault friction and elastic properties of the surrounding rock. Existing works show that complex behaviors may arise from the interplay between friction changes with slip and slip rate and stress perturbations.To shed light on the mechanics of fault dynamic triggering we performed experiments in a Biaxial Apparatus in a Double Direct Shear configuration under critically stable stiffness conditions (K/Kc~1). We used powdered quartz gouge (Min-U-Sil 40) as starting material, and conducted experiments at reference normal stress of σn = 10-13.5 MPa. After shearing the material and reaching a steady state sliding, normal stress oscillations were applied with various amplitudes, varying from A = 0.5-2 MPa, and periods, T = 0.5-50 s. In addition, we used the laboratory derived friction parameters as input for forward modeling using Rate-and-State friction laws in order to assess if these laws can explain our data. Our results show that creeping faults, under critical stiffness conditions, are sensitive to normal stress perturbations showing a variety of slip behaviors depending on amplitude and frequency of the oscillations:Oscillation frequency has a major effect on fault stability. Low and high frequencies cause a Coulomb-like response of the shear stress, that is accompanied by a complex frictional response with slow events and period doubling. At the critical frequency predicted by the Rate-and-State friction, we observe dynamic weakening resulting in regular stick-slip events. Oscillation amplitude also plays a role with the main effect depending on the magnitude of the perturbation. Using a modified Rate-and-State equation (Linker and Dieterich, 1992), we are able to accurately model the laboratory data. Our results show that normal stress perturbation on a laboratory creeping fault, at critical stiffness condition, can reproduce the entire spectrum of fault slip behavior depending on the oscillation properties.

Rocks under stress deform by creep mechanisms that include formation and slip on small-scale internal cracks. Intragranular cracks and slip along grain contacts release energy as elastic waves termed acoustic emissions (AE). AEs are thought to contain predictive information that can be used for fault failure forecasting. Here we present a method using unsupervised classification and an attention network to forecast labquakes using AE waveform features. Our data were generated in a laboratory setting using a biaxial shearing device with granular fault gouge intended to mimic the conditions of tectonic faults. Here we analyzed the temporal evolution of AEs generated throughout several hundred laboratory earthquake cycles. We used a Conscience Self-Organizing Map (CSOM) to perform topologically ordered vector quantization based on waveform properties. The resulting map was used to interactively cluster AEs. We examined the clusters over time to identify those with predictive ability. Finally, we used a variety of LSTM and attention-based networks to test the predictive power of the AE clusters. By tracking cumulative waveform features over the seismic cycle, the network is able to forecast the time-to-failure (TTF) of lab earthquakes. Our results show that analyzing the data to isolate predictive signals and using a more sophisticated network architecture are key to robustly forecasting labquakes. In the future, this method could be applied on tectonic faults monitor earthquakes and augment current early warning systems.

Frictional heterogeneities within fault zones depend strongly on lithology as well as shear fabric and strain localization structures. Here we investigate the impact of fault rock composition and shear fabric on friction constitutive properties for mature High Zagros (Iran) fault rocks. We present field observations along with results from friction experiments on intact and powdered fault rock from carbonate, quartzofeldspathic and schist. We measured frictional strength and rate/state dependence over normal stresses from 25 to 100 MPa and shear slip velocities from 3 to 300 μm/s. The sliding friction coefficients of intact fault rocks are lower than their powdered equivalents. The foliated quartzofeldspathic and lensoidal carbonate rocks from the Main Zagros Reverse Fault (MZRF) exhibit velocity strengthening friction and stable sliding. In contrast, the cataclasite carbonate rocks from the younger, Main Recent Fault (MRF) exhibit potentially unstable, velocity-weakening frictional behavior, consistent with regional seismicity. Our results suggest that solution seams, cleavage processes, and development of stylolite and calcite veins in the carbonate fault rocks of the MZRF continually rejuvenate and rework shear structures, which leads to dominantly aseismic shear. In contrast, younger fault rocks of the MRF form highly localized shear zones and shear fabric that result in velocity-weakening frictional behavior and seismic slip. Our field observations and laboratory data provide insights for using friction data and fault zone structure to build predictive models of the mode of fault slip in zones of tectonic collision.

Seismic stress drop is one of the most important earthquake source parameters, playing a key role in elastic energy release and in scaling relations for seismic moment and fault length. While stress drop does not scale systematically with earthquake size, it varies greatly within seismic catalogs and there is much broad interest in understanding how such variations relate to fault zone properties. Here, we address connections between stress drop and fault zone properties via laboratory experiments that investigate the role of normal stress and layer thickness during frictional sliding. We sheared granular layers at normal stresses from 4 to 22 MPa and document both elastic and inelastic processes that couple with layer dilation to determine the granular fragility and stress drop during stick-slip failure. Stick-slip stress drop scales directly with fault normal stress and inversely with layer thickness. Thicker layers exhibit, greater dilation during shear loading, however shear driven dilatant volume strain is independent of layer thickness. We posit that force chains form rapidly after a dynamic slip event and that bulk inelastic creep occurs via formation and destruction of force chains, interparticle slip, and rolling. Stick-slip recurrence time and stress drop vary with fault normal stress and stiffness, which increases with shear strain, consistent with a model in which stiffness increases as porosity decreases and fault zone density increases. We propose a micromechanical model that accounts for force chains and spectator regions where granular processes are dominated by inelastic slip. We document the role of elastic and inelastic processes during three stages of the stick-slip cycle and show how these change systematically as a function of normal stress and layer thickness. Our work shows that dilation during shear loading contributes to frictional strength via volume strain and that apparent friction scales inversely with the granular thinning ratio.

We explore relationships among bulk modulus, crack density, and permeability through repetitive loading of Marcellus shale. Cumulative cyclic stressing (22–26 MPa with confinement of 24 MPa) is applied at a frequency of 0.05 Hz over 100,000 cycles. Changes in acoustic velocities are used to follow changes in dynamic bulk modulus, Poisson ratio, and crack density and to correlate these with bedding-parallel measurements of methane permeability. The shale is represented as an orthotropic elastic medium containing a dominant, noninteracting fracture set separated by thin laminae. An effective continuum model links permeability evolution to the evolution of the bulk modulus and crack density. Bulk modulus is linearly related to crack density by a scaling parameter representing rock fabric and fracture geometry. The Poisson ratio and bulk modulus of the intact, uncracked shale are approximated from our data. We propose a method for tracking permeability evolution of finely laminated shales using acoustic waves.

Slow slip events (SSEs) have been identified at subduction zones globally as an important link in the continuum between elastodynamic ruptures and stable creep. The northern Hikurangi margin is hom...