[1] A mid-ocean ridge transform fault (RTF) of length L, slip rate V, and moment release rate _ M can be characterized by a seismic coupling coefficient c = A E /A T , whereA global set of 65 RTFs with a combined length of 16,410 km is well described by a linear scaling relation (1) A E /A T , which yields c = 0.15 ± 0.05 for T ref = 600°C. Therefore about 85% of the slip above the 600°C isotherm must be accommodated by subseismic mechanisms, and this slip partitioning does not depend systematically on either V or L. RTF seismicity can be fit by a truncated Gutenberg-Richter distribution with a slope b = 2/3 in which the cumulative number of events N 0 and the upper cutoff moment M C = mD C A C depend on A T . Data for the largest events are consistent with a self-similar slip scaling, D C / A C 1/2 , and a square root areal scaling (2) A C / A T 1/2 . If relations 1 and 2 apply, then moment balance requires that the dimensionless seismic productivity, n 0 / _ N 0 /A T V, should scale as n 0 / A T À1/4 , which we confirm using small events. Hence the frequencies of both small and large earthquakes adjust with A T to maintain constant coupling. RTF scaling relations appear to violate the single-mode hypothesis, which states that a fault patch is either fully seismic or fully aseismic and thus implies A C A E . The heterogeneities in the stress distribution and fault structure responsible for relation 2 may arise from a thermally regulated, dynamic balance between the growth and coalescence of fault segments within a rapidly evolving fault zone.

Abstract. Two-dimensional numerical simulations were conducted using the distinct element method (DEM) to examine the influences of particle size distribution (PSD) and interparticle friction /at, on the nature of deformation in granular fault gouge. Particle fracture was not allowed in this implementation but points in PSD space were examined by constructing assemblages of particles with self-similar size distributions defined by the twodimensional power law exponent D. For these numerical "experiments," D ranged from 0.81 to 2.60, where D = 1.60 represents the two-dimensional equivalent of a characteristic PSD to which cataclastically deforming gouge is thought to evolve. Experiments presented here used /at, values of 0.10 and 0.50 and were conducted using normal stress c• n on the shear zone walls of 70 MPa. Shear strain within these simulated assemblages was accommodated by intermittent displacement along discrete slip surfaces, alternating between broadly distributed deformation along multiple slip planes and highly localized deformation along a single, sharply defined, subhorizontal zone of slip. Slip planes corresponded in orientation and sense of shear to shear structures observed in natural gouge zones, specifically Riedel and Y shears; the oblique Riedel shears showed more extreme orientations than typical, but their geometries were consistent with those predicted for lowstrength Coulomb materials. The character of deformation in the shear zone varied with PSD due to changes in the relative importance of interparticle slip and rolling as deformation mechanisms. A high degree of frictional coupling between large rolling particles in low D (coarse-grained) assemblages resulted in wide zones of slip and broadly distributed deformation. In higher D assemblages (D >= 1.60), small rolling particles selforganized into columns that separated large rolling particles, causing a reduction in frictional resistance within the deforming assemblage. This unusual particle configuration appears to depend on a critical abundance of small particles achieved at D = 1.60 and may enable strain localization in both real and simulated granular assemblages.

No abstract

[1] We investigate the strength and frictional behavior of olivine aggregates at temperatures and effective confining pressures similar to those at the base of the seismogenic zone on a typical ridge transform fault. Triaxial compression tests were conducted on dry olivine powder (grain size 60 mm) at effective confining pressures between 50 and 300 MPa (using Argon as a pore fluid), temperatures between 600°C and 1000°C, and axial displacement rates from 0.06 to 60 mm/s (axial strain rates from 3 Â 10 À6 to 3 Â 10 À3 s À1 ). Yielding shows a negative pressure dependence, consistent with predictions for shear enhanced compaction and with the observation that samples exhibit compaction during the initial stages of the experiments. A combination of mechanical data and microstructural observations demonstrate that deformation was accommodated by frictional processes. Sample strengths were pressure-dependent and nearly independent of temperature. Localized shear zones formed in initially homogeneous aggregates early in the experiments. The frictional response to changes in loading rate is well described by rate and state constitutive laws, with a transition from velocity-weakening to velocity-strengthening at 1000°C. Microstructural observations and physical models indicate that plastic yielding of asperities at high temperatures and low axial strain rates stabilizes frictional sliding. Extrapolation of our experimental data to geologic strain rates indicates that a transition from velocity weakening to velocity strengthening occurs at approximately 600°C, consistent with the focal depths of earthquakes in the oceanic lithosphere.

East Pacific Rise transform faults are characterized by high slip rates (more than ten centimetres a year), predominantly aseismic slip and maximum earthquake magnitudes of about 6.5. Using recordings from a hydroacoustic array deployed by the National Oceanic and Atmospheric Administration, we show here that East Pacific Rise transform faults also have a low number of aftershocks and high foreshock rates compared to continental strike-slip faults. The high ratio of foreshocks to aftershocks implies that such transform-fault seismicity cannot be explained by seismic triggering models in which there is no fundamental distinction between foreshocks, mainshocks and aftershocks. The foreshock sequences on East Pacific Rise transform faults can be used to predict (retrospectively) earthquakes of magnitude 5.4 or greater, in narrow spatial and temporal windows and with a high probability gain. The predictability of such transform earthquakes is consistent with a model in which slow slip transients trigger earthquakes, enrich their low-frequency radiation and accommodate much of the aseismic plate motion.

We use three-dimensional fi nite element simulations to investigate the temperature structure beneath oceanic transform faults. We show that using a rheology that incorporates brittle weakening of the lithosphere generates a region of enhanced mantle upwelling and elevated temperatures along the transform; the warmest temperatures and thinnest lithosphere are predicted to be near the center of the transform. Previous studies predicted that the mantle beneath oceanic transform faults is anomalously cold relative to adjacent intraplate regions, with the thickest lithosphere located at the center of the transform. These earlier studies used simplifi ed rheologic laws to simulate the behavior of the lithosphere and underlying asthenosphere. We show that the warmer thermal structure predicted by our calculations is directly attributed to the inclusion of a more realistic brittle rheology. This temperature structure is consistent with a wide range of observations from ridge-transform environments, including the depth of seismicity, geochemical anomalies along adjacent ridge segments, and the tendency for long transforms to break into small intratransform spreading centers during changes in plate motion.

There is a global seismic moment deficit on mid-ocean ridge transform faults, and the largest earthquakes on these faults do not rupture the full fault area. We explore the influence of physical fault structure, including step-overs in the fault trace, on the seismic behavior of the Discovery transform fault, 4S on the East Pacific Rise. One year of microseismicity recorded during a 2008 ocean bottom seismograph deployment (24,377 0 M L 4.6 earthquakes) and 24 years of Mw 5.4 earthquakes obtained from the Global Centroid Moment Tensor catalog, are correlated with surface fault structure delineated from high-resolution multibeam bathymetry. Each of the 15 5.4 Mw 6.0 earthquakes that occurred on Discovery between 1 January 1990 and 1 April 2014 was relocated into one of five distinct rupture patches using a teleseismic surface wave cross-correlation technique. Microseismicity was relocated using the HypoDD relocation algorithm. The western fault segment of Discovery (DW) is composed of three zones of varying structure and seismic behavior: a zone with no large events and abundant microseismicity, a fully coupled zone with large earthquakes, and a complex zone with multiple fault strands and abundant seismicity. In general, microseismicity is reduced within the patches defined by the large, repeating earthquakes. While the extent of the large rupture patches on DW correlates with physical features in the bathymetry, step-overs in the primary fault trace are not observed at patch boundaries, suggesting along-strike heterogeneity in fault zone properties controls the size and location of the large events.

No abstract