S34B-01 INVITED
Episodic tremor and slip - a kinder and gentler variety of earthquake
Episodic Tremor and Slip (AKA ETS) is appearing numerous unsuspected places as seismic instruments and analysis gears up into the 21st century. From its initial and loudest lair in subduction zones, ETS observations have spread to strike-slip faults and fold-thrust belts. ETS differs from old-school earthquakes. ETS can last weeks while the grandest earthquakes run out of steam in minutes. ETS unleashes seismic moment roughly linearly with duration, as opposed to run-of-the- mill earthquakes, whose moment scales with the cube of duration. ETS has a redder spectrum and more emergent beginnings. ETS is more sensitive to stress perturbations, as revealed by susceptibility to triggering by teleseismic waves and amplitude modulation by weak tidal stressing. In the rare cases for which good location is possible, ETS most often unfolds near or well below the lower edge of the seismogenic portion of major faults, perhaps marking the transition between locked and freely slipping rheologies. Mechanically, it appears sensible that fluids play an integral role regulating the ETS process, although how fluids would be able to mediate ETS episodes spanning 100+ km of fault is a head- scratcher. Maybe ETS regions underlie most locked faults. Maybe ETS offers possibilities to delineate locked faults and detail their temporal loading with clues to when they will erupt. This story is still unfolding.
S34B-02
TRANSIENT CREEP EVENTS AND THE TRANSITION FROM ASEISMIC TO SEISMIC FAULTING OBSERVED IN LABORATORY EXPERIMENTS
Dry triaxial compression experiments performed on Carrara marble and Volterra Gypsum at confining pressures ranging from 10 to 100MPa revealed an important dependence of the strain rate on the confining stress. At constant shear stress, transient exponential increases in axial strain, similar to transient creep events observed on the field, were observed for small decreases of the confining pressure (1-10MPa), which could eventually lead to ductile failure nucleation. In such case, rupture propagation was silent and slow (several hundreds of seconds), although accompanied by stress drops of the order of several tens of MPas and millimetric slips. However, as the slip velocity increased during failure propagation, continuous acoustic waveform recordings enabled us to illustrate the transition from aseismic to seismic faulting. These experiments provide a clear experimental case of silent, slow localized failure in rocks as a result of an interplay between intragranular plasticity and microcracking. In the case of ductile failure as in the marble, dislocation and twin accumulation is such that cracks propagation steps are small and/or slow, and thus the radiated energy release rate remains small at early stages of rupture and increases with rupture speed. This last observation clearly highlights the dependence of radiated acoustic (and microseismic?) energy during crack propagation not only on the rupture propagation speed and the slip velocity but most importantly on the rock's lithology and rheology. This could have important implications as carbonates are prevalent within sedimentary basins. At even at shallower depths and prevalent within fault gouges, clay minerals are also expected to behave in a similar way.
S34B-03 INVITED
Interaction of Earthquakes and Aseismic Slip: Insights From 3D Fault Models Governed by Lab-Derived Friction Laws
Recent improvements in availability and quality of seismic and geodetic data have revealed complex interactions of seismic and aseismic slip. This rich information, interpreted through modeling, can help us understand mechanics and physics of faults. Our simulation approach (Lapusta and Liu, 2008) produces spontaneous long-term fault slip and the resulting stress redistribution with full inclusion of inertial effects during simulated earthquakes in the context of a 3D fault model. The approach incorporates laboratory- derived rate and state friction laws, involves slow, tectonic-like loading, resolves all stages of seismic and aseismic slip, and results in realistic rupture speeds, slip velocities, and stress drops. We use the simulations to study two phenomena that arise due to interaction of seismic and aseismic slip: (i) supershear transition in 3D models due to rheological boundaries and (ii) mechanics of small repeating earthquakes. In our simulations, aseismic slip in creeping regions concentrates stress at rheological boundaries and promotes supershear transition during dynamic events (Liu and Lapusta, 2008). Transition of earthquakes from subshear to supershear speeds has important implications for strong ground motion and fault properties. Simulations of supershear transition are typically done in models of single events on linear slip- weakening faults. We simulate long-term seismic and aseismic slip on a strike-slip fault with a rectangular velocity-weakening (VW) region surrounded by velocity-strengthening (VS) regions. Existence of shallow and deeper VS regions is supported by lab friction studies as well as observations of interseismic creep, postseismic slip, limits on the depth extent of seismicity, and clustering of small events. In our simulations, steady slip throughout the interseismic period in the VS areas concentrates stress next to rheological boundaries and promotes faster rupture speeds and local supershear propagation of dynamic events. Under a range of conditions, this local behavior causes the entire rupture to become supershear. Our modeling of small repeating earthquakes reproduces several observational constraints, including their nontrivial scaling of seismic moment with the recurrence time (Chen and Lapusta, 2008). In our 3D model, a small circular patch with velocity-weakening friction is surrounded by a much larger velocity-strengthening region. When the patch size is smaller than the nucleation size implied by the underlying rate and state formulation, all slip on the patch is aseismic. For larger patch sizes, small repeating events occur, with slip rates of the order of 1 m/s and sharp stress drops; however, the patch also experiences significant aseismic slip. By varying the patch radius, we are able to produce repeating earthquakes that (i) reproduce the observed scaling T∝ M01/6 of the repeat time T and seismic moment M0 and (ii) have source dimensions and stress drops typical for earthquakes of comparable sizes and similar to recent inversions for Parkfield repeaters. Remarkably, the scaling T∝ M01/6 is independent of the variation, within a factor of 4, of rate and state parameters a and b. The scaling is also reproduced in a model with a rectangular velocity-weakening patch and with quasi-dynamic approach.
S34B-04
Role of Fault Dilatancy in Subduction Zone Aseismic Deformation Transients and Thrust Earthquakes
Numerical simulation in the framework of rate and state friction shows that short-period aseismic deformation transients can emerge spontaneously when interstitial fluids are present and pore pressure p is near- lithostatic around the friction stability transition, for certain friction parameter variations with depth [Liu and Rice, JGR, 2007]. This is precisely the situation for which Segall and Rice [JGR, 1995] suggested that fault stabilization by induced suction from dilatancy during increased shear rates becomes most important. In this study, building on Taylor and Rice [EOS, 1998], Liu and Rice [EOS, 2005] and especially Segall and Rubin [EOS, 2007], we analyze the conditions for short-period aseismic transients and dimensions of coseismic rupture (within the radiation damping approximation) of a fluid infiltrated subduction fault using the rate and state friction model including dilatancy and pore compaction (using membrane diffusion approximation) effects. First, in a simplified situation that the fault is completely locked at one side and is loaded by a constant rate Vpl at the other, extensive simulation cases confirm that the fault response is a function of the non-dimensional parameters E = fo ε / β b σ̅, T = Vpl tp / L (time scale of fluid pressure re-equilibration), W / h★ (length ratio of the velocity-weakening region under near- lithostatic p and the critical nucleation patch size) and a/b. Here, σ̅; is effective normal stress, fo is steady state friction, ε is a dilatancy coefficient representing porosity changes in response to state changes, β is a combination of fluid and pore compressibility, a, b and L are friction parameters. Self-sustained slip rate oscillation remains aseismic at large W / h★, for which earthquakes would occur without dilatancy. The maximum slip rate during transient episodes decreases as E (or T) increases to ~ 1, while the recurrence interval remains relatively constant. We then extend the analysis to a shallow-dipping subduction fault model using recently reported hydrothermal gabbro gouge friction data [He et al., Tectonophys., 2006, 2007]. The along-dip elevated p is constrained by seismological observations and by thermal and petrological models for the northern Cascadia margin (p near-lithostatic around stability transition and lower in the seismogenic zone). Similarly, aseismic transients can exist for a much broader range of W / h★, making it plausible to produce transients with total slips of a few centimeters and recurrence periods of a couple years while using lab values for L of 10s of microns in the low σ̅ zone. Inclusion of dilatancy also reduces the speed and spatial extent of coseismic rupture. For a fixed T = 1 and ε / β = 0.2 MPa, rupture stops ~ 50 km up-dip of the lower stability transition and causes nearly no coseismic slip at the trench. The depth of complete interseismic locking also varies with parameters E and T in the seismogenic zone. This suggests that a subduction fault extending well down-dip of the limit of seismogenesis could be frictionally unstable (a-b<0) but undergo no seismic slip due to effective dilatancy stabilization. This has implications for the relative depths of slow slip events and thrust earthquakes and for the total slip budget in an earthquake cycle.
S34B-05 INVITED
The effects of fault zone architecture on earthquake triggering
We report on laboratory experiments in which stick-slipping shear surfaces are subject to transient stressing to simulate earthquake triggering by seismic waves. Granular layers and bare granite surfaces were sheared in a servo-controlled deformation apparatus in double-direct shear. The seismic waves from an earthquake and tectonic load were simulated by superimposing a loading rate sinusoid on a constant shear loading rate. The dependence of triggered stick-slip failure on fault stress state and architecture was analyzed. Fault architecture was evaluated by varying gouge layer thickness (2 to 6 mm) and studying bare granite surfaces. We compare the shortened recurrence times for faults under transient loading conditions to the consistent recurrence intervals under constant loading rate. Our results imply that triggering depends on oscillation amplitude and frequency, as well as properties of the fault. Larger-amplitude dynamic stresses reduce stick- slip recurrence intervals for granular layers, whereas failure times for granite surfaces are uncorrelated with oscillation amplitude. Granular layers have shorter recurrence rates at higher frequency, whereas the recurrence intervals of granite surfaces are lengthened or unaffected by high-frequency oscillations. Higher frequencies can inhibit failure when fault slip exceeds a critical slip distance prior to peak velocity and encourages failure if a critical slip distance is achieved postpeak velocity. Increasing velocity temporarily strengthens faults, whereas velocity reduction further weakens and promotes failure, as predicted by the rate-and-state friction laws. Our results may explain variations in earthquake triggering thresholds and imply that high-frequency thresholds may not be constant, as has been previously proposed. Our work suggests that triggering thresholds are dependent on amplitude, frequency, critical slip distance, timing in the interseismic cycle, and stressing rate.
S34B-06
Rocks pulverized near San Andreas Fault: insight from high strain rate testing experiments
Pulverized rocks have been found at the surface near the San Andreas Fault, up to 400m away from the fault core. This intense damage is unusual at such distance from the main slip surface. At these distances from the fault core, rocks are usually fragmented along prominent localized fractures. Here, we show that high strain rate loading may inhibit strain localization and induce intense pervasive crack nucleation. We loaded non-pulverized rocks from the San Andreas Fault damage zone using a Split Hopkinson Pressure Bars apparatus. This apparatus is classicaly used for high strain rate testing of materials in engineering mechanics, but scarcely used in Earth Sciences. When strain rate is smaller than 100/s, the rocks are split in two or three fragments along well-localized fractures. Conversely, when submitted to a strain rate larger than 150/s, the rock samples break into numerous small fragments, smaller than the initial grain size. Such strain rates are difficult to attain 100m away from the fault zone with a classical subshear rupture, but are compatible with the shock wave accompanying a supershear rupture. We therefore propose that pulverized rocks may be (1) a non-seismological tool to prove the seismological pertinence of supershear earthquakes (2) a paleomarker of previous supershear rupture useful for the risk assessment of such potentially damaging earthquakes.
S34B-07
Thermal Pressurization Induced by Coseismic Shear Heating Within Thermally Unstable Rocks
The frictional heating generated during an earthquake can cause the thermal pressurization of the fluids trapped within the slip zone, which favours dramatic fault zone weakening and rupture propagation. Many slip weakening mechanisms have been proposed to account for coseismic fault zone weakness: flash heating, thermal pressurization, frictional melting, gel formation, thermal decomposition and moisture drainage. Thermal pressurization of fluids has not been previously reproduced in the laboratory and, as a consequence, the experimental verification of the effective normal stress principle at seismic slip rates (m/s) is lacking. Here we present data from high velocity friction experiments where a new type of thermal pressurization has been reproduced as the fluids are not initially present within the slip zone but are released by decarbonation (dolomite and Mg-rich calcite) and dehydration (gypsum) reactions, both activated by frictional heating during seismic slip. The coseismic shear strength of experimental faults dramatically reduces to almost zero when fluids are trapped and pressurized within the slip zone, in accord with the effective normal stress principle. The microstructures observed in the areas adjacent/within the slip zone can be used in future studies as new diagnostic features to aid in the recognition of seismic faulting within thermally unstable rocks and, possibly, of thermal pressurization slip weakening processes. Earthquake source parameters (e.g. slip weakening distance and fracture energy), calculated from our experimental data and extrapolated at seismogenic depths, match very closely the seismological data from the mainshocks of the 1997 Colfiorito earthquake, nucleated within dolomite/anhydrite rocks in the Northern Apennines of Italy. The gap between mechanical and seismological observations can be bridged when thermal pressurization is considered to act in combination with other slip weakening mechanisms.
S34B-08 INVITED
Frictional Melting of Peridotite and Seismic Slip
The evolution of the frictional strength along a fault at seismic slip rates (about 1 m/s) is one of the main factors controlling earthquake mechanics. In particular, friction-induced rock melting and melt lubrication during seismic slip may be typical at mantle depths, based on field studies, seismological evidence, torsion experiments and theoretical studies. To investigate the (1) dynamic strength of faults and (2) the frictional melting processes in mantle rocks, we performed 20 experiments with the Balmuccia peridotite in a high- velocity rotary shear apparatus. Experiments were conducted on cylindrical samples (21.8 mm in diameter) over a wide range of normal stresses (5.4 to 16.1 MPa), slip rates (0.23 to 1.14 m/s) and displacements (1.5 to 71 m). The dynamic strength of experimental faults evolved with displacement: after a peak (first strengthening) at the initiation of slip, fault strength abruptly decreased (first weakening), then increased (second strengthening) and eventually decreased (second weakening) towards a steady-state value. The microstructural and geochemical (FE-SEM, EPMA and EDS) investigation of the slipping zone from experiments interrupted at different displacements, revealed that second strengthening was associated with the production of a grain-supported melt-poor layer, while second weakening and steady-state with the formation of a continuous melt-rich layer. The temperature of the frictional melt was up to 1780 Celsius. Microstructures formed during the experiments were identical to those found in natural ultramafic pseudotachylytes. By performing experiments for increasing normal stresses and slip rates, steady-state shear stress slightly increased with increasing normal stress (friction coefficient of 0.15) and, for a given normal stress, decreased with increasing slip rate. The dependence of steady-state shear stress with normal stress and slip rate is described by a constitutive equation for melt lubrication. The presence of microstructures similar to those found in natural pseudotachylytes and the determination of a constitutive equation that describes the experimental data, might allow to extrapolate the experimental observations to natural conditions and to the study of rupture dynamics in mantle rocks.