T22A-01 10:20h
Microstructures, Chemical Composition, and Viscosities of Fault-generated Friction Melts in the Shimanto Accretionary Complex, Southwest Japan: Implication for Dynamics of Earthquake Faulting in Subduction Zones
The pseudotachylytes (PT) were recently found in the Cretaceous Shimanto accretionary complex of eastern and western Shikoku, southwest Japan, but their microstructures under a backscattered electron image, chemical composition, and effects of frictional melting on co-seismic slip in the accretionary prism remains poorly understood. The PT bearing fault is the 1-2 m thick roof thrust of a duplex structure, which bounds the off-scraped coherent turbidites above from the imbricated melange below without a thermal inversion across the fault. The fault zone consists of foliated cataclasite of sandstone-shale melange in origin and dark veins. The PT commonly occurs as brecciated fragments in dark veins. The PT matrix is transparent under plane-polarized light and is optically homogeneous under cross-polarized light, similar to glass matrix. Under a backscattered electron image, the PT clearly shows the evidences for frictional melting and subsequent rapid cooling: rounded and irregularly shaped grains and vesicles in matrix and fracturing associated with grain margins. These textural features of the PT are very similar to those of experimentally generated PT. The EPMA analysis indicates that chemical composition of the PT matrix corresponds to illite with 5.7-9.9 wt% H2O and that partially melted grains are dominated by orthoclase and quartz. This indicates that the temperatures of the PT melt could reach the breakdown temperatures of orthoclase (1150 C) and quartz (1730 C), greater than the maximum temperature recorded in host rocks (170-200 C). We calculated the viscosity of friction melt, based on the chemical composition of the PT matrix and the volume fraction and aspect ratio of grains in the PT. We considered both Arrhenian and non-Arrhenian models for viscosity calculation. Our result demonstrates that the melt viscosity is much lower than PT in continental plutonic and metamorphic rocks: 10^3 Pa s (Arrhenian model) and 10^2 Pa s (non-Arrhenian model) even at 700 C and 10 Pa s (both models) at 1200 C. The extremely low melt viscosity is caused primarily by the formation of liquids (release of OH-) from hydrous illite, and secondarily by small volume fraction (< 20%) of grains in the PT. Because illite is commonly present in accretionary prisms, generation of a low viscosity melt from illite would lead to fault lubrication and hence control the efficiency of stored strain energy release and earthquake magnitude in subduction zones.
T22A-02 10:35h
Maine Pseudotachylyte Localities and the Role of Host Rock Anisotropy in Fault Zone Development and Frictional Melting
Three brittle strike-slip fault localities in coastal Maine have developed pseudotachylyte fault veins, injection veins and other reservoir structures in a variety of host rocks where the pre-existing layering can serve as a controlling fabric for brittle strike-slip reactivation. Host rocks with a poorly-oriented planar anisotropy at high angles to the shear direction will favor the development of R-shears in initial en echelon arrays as seen in the Two Lights and Richmond Island Fault Zones of Cape Elizabeth that cut gently-dipping phyllitic quartzites. These en echelon R-shears grow to through-going faults with the development of P-shear linkages across the dominantly contractional stepovers in the initial arrays. Pseudotachylyte on these faults is very localized, typically up to 1-2 mm in thickness and is restricted to through-going fault segments, P-shear linkages and some sidewall ripouts. Overall melt production is limited by the complex geometry of the multi-fault array. Host rocks with a favorably-oriented planar anisotropy for reactivation in brittle shear, however, preferentially develop a multitude of longer, non-coplanar layer-parallel fault segments. Pseudotachylyte in the newly-discovered Harbor Island Fault Zone in Muscongus Bay is developed within vertical bedding on regional upright folds with over 50 individual layer-parallel single-slip fault veins, some of which can be traced for over 40 meters along strike. Many faults show clear crosscuts of pre-existing quartz veins that indicate a range of coseismic displacements of 0.23-0.53 meters yielding fault vein widths of a few mm and dilatant reservoirs up to 2 cm thick. Both vertical and rare horizontal lateral injection veins can be found in the adjoining wall rock up to 0.7 cm thick and 80 cm in length. The structure of these faults is simple with minor development of splay faults, sidewall ripouts and strike-slip duplexes. The prominent vertical flow layering within the mylonite gneisses of Gerrish Island serves as host to the complex Fort Foster Brittle Zone where it localizes brittle fault slip and contributes to a maximum area of contact between the sliding surfaces which, in turn, yields fault vein thicknesses of 1-2 mm and locally up to 2 cm. The reactivation of this planar anisotropy in brittle shear produces long overlapping geometries that develop linking structures in both extensional and contractional stepovers may reflect the development of sidewall ripouts due to adhesive wear. The prominent development of closely-spaced individual single-slip fault veins suggests frictional welding as an effective strain hardening mechanism for repeated stick-slip.
T22A-03 10:50h
Can Pseudotachylytes be Used to Infer Earthquake Source Parameters? Limitations in the Study of Exhumed Faults
Tectonic pseudotachylytes might be used to constrain earthquake source parameters, such as dynamic shear stress resistance, average dynamic friction and slip-weakening distance. Estimation of dynamic shear stress resistance and dynamic friction from field studies is based on the assumption that the volume of melt produced during coseismic slip is proportional to the frictional work converted to heat on the fault surface. Conditions conducive to a realistic estimate of dynamic shear resistance are: i) the presence of large outcrop exposures that allow for estimation of the volume of pseudotachylyte, ii) the presence of structural markers offset by faults to relate the displacement accommodated by the fault with the volume of melt produced, iii) data that provide an estimate of the initial melt temperature and of host-rock temperature during seismic faulting. An independent indication that steady-state friction in the presence of melts might be achieved during coseismic slip arises from the dependence of the fractal dimension of the fault profile (intersection of the fault surface with the outcrop surface) with displacement. This relation could also indicate the slip-weakening distance (Hirose and Shimamoto, 2003). The above conditions are all satisfied in the case of the Gole Larghe Fault Zone, which consists of hundreds subparallel strike-slip faults that cut tonalites of the Adamello batholith (Italy). The thickness of pseudotachylyte-bearing faults increases with displacement. From displacement/thickness ratios and energy balance calculations we determined the dynamic shear resistance for several pseudotachylyte-bearing faults. In the same faults the fractal dimension of the fault profile increases from 1.0 to 1.16 with displacement. This was also observed in experiments where steady state friction in the presence of melt was achieved (Hirose and Shimamoto, 2003). However, we will show that, since pseudotachylytes often overprint preexisting cataclasites, the estimate of the earthquake source parameters from field studies is limited by the uncertainties to attribute the measured displacement to a single seismic rupture.
T22A-04 11:05h
The Effect of Bandwidth Limitations on the Inference of Slip-Weakening and Energy Parameters from Seismograms
Numerous researchers have obtained estimates of slip-weakening distance Dc and fracture energy for recent earthquakes. Dc is often observed to be a significant fraction of the total slip and tends to correlate with total slip. While these observations may well be true of real earthquakes, we show that low-pass filtering of strong motion seismograms can also produce some of these effects in inverted rupture models. We test the accuracy of Dc estimates by calculating them in low-pass filtered versions of models A and B of Guatteri and Spudich (2000). Models A and B are two different rupture models for a hypothetical M6.5 earthquake, and they have nearly identical rupture time, slip, and stress drop distributions, as well as nearly identical predicted seismograms, but Dc for model B is about twice that for model A. By low-pass filtering slip models A and B at 1.0 Hz, we simulate the blurring effects of band-limited waveform inversions on these slip models. At each point on a fault, Dc' is defined to be the slip at the time of the peak slip speed at that point. Low-pass filtering the slip models causes an upward bias in Dc inferred from stress-slip curves, and it causes an artificial correlation between Dc' and the total slip. Low-pass filtering might also bias fracture energy high and radiated energy low. These biases should be considered when interpreting Dc derived from band-limited slip models of real earthquakes.
T22A-05 11:20h
Earthquake Sequences on Rate and State Faults With Strong Dynamic Weakening
The laboratory-derived rate and state friction laws provide a unique tool for simulating earthquake sequences in their entirety, from accelerating slip in slowly expanding nucleation zones to rapid dynamic propagation of earthquake rupture to post-seismic slip and interseismic creep to fault restrengthening between seismic events. We (Lapusta et al., JGR, 2000) developed an efficient methodology that allows simulating all these stages of the earthquake process within a single algorithm in a model based on the Dieterich-Ruina rate and state laws. However, the laws have been derived from experiments with slip velocities small compared to those in seismic range. Hence these logarithmic laws, although reflecting important physics at slow slip velocities, fail to capture potential much stronger variations of fault strength during dynamic slip. Such variations, in particular much stronger weakening, have been receiving more and more experimental and theoretical support. Hence we study the behavior of 2-D rate and state faults with the constitutive relation modified at high seismic slips and/or slip velocities. One option, motivated by theoretical and experimental results on flash heating and, potentially, thermal pressurization of partially drained fault gauge, is to modify the steady-state frictional strength by the factor 1/(1+V/Vw), where Vw ~ 0.1-1 m/s is the parameter regulating at what slip velocities V the additional weakening steps in. Earthquake sequences simulated with such a law, that also includes static (slow-velocity) friction coefficient of about 0.6 as determined in the lab and effective normal stresses comparable to the overburden minus hydrostatic pore pressure, produce earthquakes that nucleate at high static shear stresses and propagate (and arrest) at low dynamic shear stresses, resulting in huge static stress drops. To avoid large static stress drops, it is NOT necessary to assume uniformly low static friction coefficient and/or uniformly low effective normal stress. It is enough to incorporate isolated weak regions that would nucleate earthquake ruptures under low overall shear stresses. Some of these ruptures then continue into the statically strong fault regions due to dynamic weakening. Such model results in fault operation at low shear stresses and low heat production, with reasonable static stress drops and earthquake ruptures propagating as self-healing pulses and hence satisfies a number of basic observational constraints. Similar results have been obtained (Rice, AGU, 1996) when the rate and state friction was modified by adding slip-dependent pore pressure evolution appropriate to undrained adiabatic shear heating of pore fluids. The simulations were done using quasi-dynamic approximation. We will report on our current efforts to incorporate such pore pressure evolution into a fully dynamic simulation of earthquake sequences as well as to combine it with the flash heating weakening mechanism discussed above. Flash heating and pore pressurization are likely to coexist during rapid frictional sliding, as the former occurs at large slip velocities and small slips while the latter occurs at large slip velocities and large slips. Flash heating decreases frictional resistance very early in the sliding process and hence it can influence the effectiveness of the pore pressurization mechanism.
T22A-06 INVITED 11:35h
Dynamic fault weakening caused by thermal pressurization in an earthquake model governed by rate- and state-dependent friction
We model the traction evolution during the dynamic propagation of an earthquake rupture governed by rate- and state-dependent friction with thermal pressurization of pore fluids. The adopted numerical procedure allows us to perform 3-D simulations in which the heat generated during sliding raises the pore pressure and reduces fault friction. The goals are to investigate how dynamic traction varies with slip or time and to understand the physical mechanisms controlling dynamic weakening during slip episodes. These features have important implications on the estimate of fracture energy as well as on the size of the characteristic slip-weakening distance. We have performed different numerical experiments varying the thickness of the slip zone as well as the hydraulic diffusivity value. The variations of diffusivity are associated to changes in permeability comprised between $10^{-20}$ and $10^{-16}$ $m^2$. Porosity can be constant or it can evolve with time according to a specified analytical law. Our results show that the thickness of the slip zone and the hydraulic diffusivity value modify the shape of the traction versus slip curves. Numerical simulations performed with different constitutive formulations reveal that the evolution law strongly affects the traction dependence on slip or time. For particular configurations (for instance, when the effective normal stress changes are not accounted in the evolution of the state variable or when porosity evolves with time), the traction evolution shows a gradual and continuum weakening with increasing slip; for these behaviors the definition of $D_c$ might become rather meaningless. Our results confirm that the breakdown stress drop is inversely proportional to the fault thickness and to the hydraulic diffusivity. A similar relation has been found for the characteristic slip-weakening distance $D_c$. The increase of $D_c$ caused by thermal pressurization is relevant: in a set of simulations we have found that values larger than $0.6 m$ are measured for a hydrated fault zone, while the resulting value for a dry fault is equal to $0.04 m$. Thermal pressurization yields large peak slip velocity values, exceeding 1 $m/s$. We observe that, if diffusivity is comparable or slightly larger than laboratory values, the breakdown stress drop (i.e., the difference between the minimum and the yield stress values) is very large. This suggests that earthquake stress drop might be nearly complete. The estimated fracture energy values are consistent with those inferred seismically.
T22A-07 11:50h
Transition from Frictional Weakening to Shear Heating Induced Thermal Pressurization
Earthquake nucleation requires loss of frictional strength $\tau = \mu (\sigma - p)$ with slip or slip rate. For rate and state dependent $\mu$ at fixed $(\sigma - p)$ instabilities can occur when $d \tau_{ss} / d {\rm log} v = (\sigma - p)(a - b)$ is negative. Shear heating increases $p$ and, if dilatancy and pore pressure diffusion are limited, will cause $\tau$ to decrease. We examine how frictional weakening, shear heating, dilatancy and pore-pressure diffusion determine stability in a simplified fault model with a narrow core (thickness $h$) bordered by an impermeable wall, and an outer damage zone. We develop a fluid diffusion model accurate when along strike variations are much greater than $h$, and times are long compared to the pore pressure diffusion time through the fault core. Dilatancy and thermal diffusion are included in an approximate fashion. If the drained behavior is stable ($ a>b$), and wall zone permeability exceeds a critical value (estimated at $\sim 10^{-21}m^2$) then fault slip is linearly stable at all wavelengths. The critical permeability is less than that measured in active fault cores, even at effective stresses acting at 10 km depth. We conclude that shear heating can not generally nucleate slip instability and that frictional weakening is required. However, shear heating may nucleate instability on velocity strengthening faults following strong stress perturbations. On frictionally weakening faults shear heating becomes dominant at slip speeds of order %$v \sim L_p/t_p $\sim$ 1 mm/s, and displacements of $(\sim 0.001-0.01 m)$. Thus, dynamic rupture may be insensitive to frictional variations and dominated by shear heating effects. Time to failure calculations based on rate-state friction alone, however, should be approximately valid
T22A-08 12:05h
Numerical simulation of rupture propagation with thermal pressurization based on measured hydraulic properties :Importance of deformation zone width
Thermal expansion and pressurization of pore water may effectively reduce frictional resistance during slip at high strain rate for faults with impermeable clayey material in shallow depth ($<$5km). Width of deformation zone is an important parameter in this process because the distribution of excess pore pressure is roughly equal to the width of deformation zone for relatively impermeable faults and pore pressure at the center rises more rapidly for thinner cases. In this study, dynamic rupture propagation simulation with thermal pressurization based on measured hydraulic properties of active faults demonstrates the importance of width of deformation zone. In the case that hydraulic properties of Hanaore fault in Southwest Japan (relatively impermeable) is used, when 100m radius asperity is assumed in which initial shear stress is set as static frictional level (0.6) and dynamic frictional coefficient is assumed to be 0.4, rupture velocity is proved sensitive to the width of deformation zone when it is less than 20mm under condition of 3km depth. When deformation width is reduced to 5mm, rupture velocity in direction of mode 2 exceeds S-wave velocity. This study also shows the importance of hydraulic properties by comparing results of a relatively impermeable fault and a permeable fault at different depths.