S14B-01
Is the Chrystalls Beach Accretionary Melange a Fossil Subduction Channel Shear Zone?
In the northeast of the Hikurangi subduction margin, a 1-2 km thick layer of high Vp/Vs, low Qp and distributed microseismicity is present along the subduction megathrust interface (Eberhart-Phillips & Chadwick, 2002). This zone is interpreted as a 'subduction channel' consisting of a fluid-saturated, highly sheared mixture of trench-fill sediments, which have been subducted below (or eroded from) the accretionary prism (von Huene & Scholl, 1991). Seismic style within subduction channels may vary from large megathrust ruptures to aseismic slip associated with microseismic activity. The factors controlling these variations in style are not well understood due to the inaccessible nature of active subduction thrust interfaces. The Chrystalls Beach Complex, SE Otago, New Zealand, is a possible analogue for the seismogenic zone of an active subduction channel. This complex comprises an intensely sheared mixture of chert, terrigenous mud and sand, minor volcanogenic sediments and pillow lavas. It has a 'block-in-matrix' mélange structure, where asymmetric, dismembered beds of sand-rich competent material are enclosed within a relatively incompetent, cleaved pelitic matrix. The rock assemblage has been progressively deformed in a top-to-the-north shear zone, and is pervaded by an anastomosing network of quartz/calcite shear- and extension veins, where individual veins can be traced for tens of metres. The presence of extension veins indicates episodes where the tensile overpressure condition (Pf > σ3) was locally attained. Initially the sediments experienced compaction, volume loss and layer-parallel soft sediment shearing, developing a slaty cleavage and viscous S/C shear structures. The dense vein network developed during subsequent brittle deformation. The mineral assemblage (pumpellyite-chlorite to pumpellyite-actinolite), mica b0 spacing and illite crystallinity indicate deformation in a high pressure - low temperature environment ( ~ 3-6 kbar, ~ 200-300°C). This P-T environment and structural character appear to match that inferred for microseismically active portions of the Hikurangi subduction channel. Slickenfibres coating shear veins in the Chrystalls Beach Complex formed by a 'crack-seal' mechanism, suggesting formation by episodic slip coupled to fluid pressure cycling and solution transfer. These veins may therefore record incremental slip associated with microearthquakes like those seen in highly overpressured regions of active subduction zones and other creeping fault segments. Observations from this fossil shear zone provide constraints for laboratory and theoretical models, highlighting the chaotic nature of natural faults. For example, shear veins in the complex are commonly localised along lithological contacts, while extension veins are concentrated in the more competent units. If the process forming these veins mirrors the triggering mechanism of subduction zone microseismicity, then the structure and composition of the subduction channel shear zone impose a significant control on deformation along the subduction thrust interface. In particular, heterogeneity in fluid pressure and the ratio of competent to incompetent material seem likely to be important sources of strength heterogeneity along the interplate megathrust.
S14B-02
Influence of Rupture Speed on Effective Normal Stress Changes During Slip Between Dissimilar Poroelastic Materials
We consider the poroelastic response of a fault zone consisting of a relatively impermeable fault core surrounded by a more permeable damage zone. In-plane slip compresses one side of the fault and extends the other, generating a gradient in fluid pressure that drives flow across the fault. Any asymmetry across the fault, either in elastic properties and density measured over the scale of the rupture, or in poroelastic properties and permeability measured over the hydraulic diffusion length (as would occur if slip localizes at the boundary between the fault core and damage zone), leads to effective normal stress changes on the fault [Rudnicki and Rice, 2006]. The sign of the effective normal stress change reverses if the rupture propagates in the opposite direction, in a manner similar to the well-known bimaterial effect of normal stress changes during slip between dissimilar elastic solids. The sign of the effective normal stress change cannot always be predicted solely from the contrast in elastic properties across the fault. In numerical models with opposing elastic and poroelastic effects, we observe, as the rupture accelerates, a reversal in the sign of effective normal stress change from that predicted by the local poroelastic mismatch to that predicted by the larger scale elastic mismatch, provided that the wave-speed contrast exceeds about 5--10% (the precise value depends on the poroelastic contrast and Skempton's coefficient). For faults separating more elastically similar materials, there exists a minimum poroelastic contrast above which the poroelastic effect always determines the sign of the effective normal stress change, no matter the rupture speed.
S14B-03
Asymmetric pulse-like rupture at bimaterial interface with slip-weakening friction model
Under some circumstances, pulse-like rupture propagation at a bimaterial interface becomes strongly asymmetric, and can be characterized as unilateral in the sense that slip diminishes and eventually dies out in one direction while growing unstably in the other. Pulse-like ruptures capable of this mode of evolution can sometimes be induced for inplane (2D) models with strongly velocity-weakened friction (Ampuero and Ben-Zion, 2007), but have not been seen for 2D slip-dependent friction models (e.g., Harris and Day, 1997). However, we have found that 3D effects (leading to pulse-like rupture) can induce the strongly asymmetric rupture mode even with purely slip-weakening friction. In the slip-weakening case, rupture of faults much longer than their down-dip width initially develops in a crack-like, bilateral mode, and subsequently (due to stopping phases from the top and bottom edges) evolves into two separate slip pulses traveling in opposite directions (e.g., Day, 1982). Under a restricted range of initial conditions, when the fault is at a bimaterial interface (we have so far investigated wavespeed contrasts of ~20%), the slip pulse in the preferred direction propagates indefinitely, while the one in the non-preferred direction dies out. This mode only occurs when the rupture initiates from a localized stress concentration and then propagates into a lower-stress background for which the critical dimension for unstable rupture is tuned closely to the fault width. When initial conditions permit this mechanism to originate, the subsequent propagation distance in the non-preferred direction depends on the value of the quotient (1 + phis)/(1-phid), where phis and phid are, respectively, the static and dynamic friction coefficients (with the die-out distance reducing for high values of this quotient and increasing or transitioning to bilateral rupture for low values). For a surface-rupturing fault, similar relations govern the transition of the rupture mode, provided one interprets the width of the fault as the half-width of an equivalent embedded fault. Both free surface effects and initial normal stress also have some effect on the die-out distance of the non-preferred pulse. If there is no tensile limit imposed on the fault stresses, the preferred-direction slip velocity grows indefinitely with propagation distance, but when fault opening (mode I displacement) is permitted in order to enforce a tensile limit, pulse slip- velocity approaches a steady state value. Whether this 3D mechanism is important in real earthquakes may depend upon a number of phenomena that we have yet to explore, including its sensitivity to natural heterogeneities in initial and frictional stresses, the extent to which it may be amplified by velocity-dependent friction, and the effect of stress limits imposed by off-fault material damage.
S14B-04
Normal Stress and Gaps in Interacting Strike-Slip Faults
One of the pressing issues in seismic hazard analysis is the probability of earthquake rupture jumping between fault segments, resulting in a large cascading event. Researchers (e.g., Wesnousky, 2006) have conducted observational surveys to determine the maximum width of stepover that rupture may jump in strike-slip systems, and other researchers (e.g., Harris et al., 1991; Kase and Kuge, 1998; Harris and Day, 1999) have used numerical methods to study the dynamics of such systems. Both methods imply that ruptures have difficulty propagating across stepovers with widths greater than 3-4 km, with possible differences between compressional and extensional stepovers. The present study extends the numerical modeling work above to address some remaining questions, such as 1) What is the role of the time-dependent normal stress in extensional and compressional stepovers, and 2) How does the presence of an along-strike fault overlap or gap affect the ability of rupture to propagate through stepovers? I find that decreases in the time-dependent normal stress field radiated by the primary fault facilitate jumping rupture in both extensional and compressional stepovers. Removing normal stress dependence from the friction law reduces the maximum jump distance in both extensional and compressional cases. In addition, an along-strike gap makes it much more difficult for rupture to propagate through an extensional stepover than a compressional stepover. The results may have implications for seismic hazard in near multi-segment strike-slip systems.
S14B-05
Segmentation and Maturity of Long-Term Faults Control Earthquake Slip-Length Scaling
Slip and length measurements on earthquakes suggest large stress drop variability. To seek for the possible origin(s) of this apparent variability, we analyze an extended set of surface slip-length measurements that we have compiled for more than 270 large continental earthquakes worldwide (M > 6; the data set has been extended from that provided by Manighetti et al., EPSL, 2007). Using satellite imagery, we simultaneously analyze the geometry and large-scale segmentation of the long-term faults on which the earthquakes have occurred, and identify (when possible) the major fault segments broken by the earthquakes. We find that most long-term faults are composed of a limited number of major segments, commonly 4 more or less 1, separated by discontinuities (inter-segment zones) that commonly appear in map view as significant steps, bends or zones of cross-cutting faulting. We also find that the earthquakes break a variable number of those major segments, with displacement- length ratios which depend on that number and actually decrease as that number increases. We suggest that the number of broken segments depends on the strength of the inter-segment zones, which itself depends on the structural maturity of the long-term faults. We propose new Dmax-L parameterizations based on that idea of multiple segment-ruptures. In such parameterizations, each broken segment roughly scales as a crack, while the total multi-segment rupture does not. Stress drop on individual segments is roughly constant, only varying between 3.5 to 9 MPa. The slight variation that is still observed depends on fault structural maturity; more mature faults have lower stress drops than immature ones. The new Dmax-L functions that we propose reduce uncertainties with respect to available relationships. They thus provide a more solid basis to estimate seismic hazard by integrating fault properties revealed by geological studies.
S14B-06
Slip-Length Scaling in Large Earthquakes: The Role of Deep Penetrating Slip Below the Seismogenic Layer
Coseismic slip is observed to increase with earthquake rupture length for lengths far beyond the lengthscale set by the seismogenic layer. The observation, when interpreted within the realm of static dislocation theory and the imposed limit that slip be confined to the seismogenic layer, implies that earthquake stress drop increases as a function of rupture length for large earthquakes and, hence, that large earthquakes differ from small. Here a three dimensional elastodynamic model is applied to show that the observed increase in coseismic slip with rupture length may be satisfied while maintaining constant stress drop across the entire spectrum of earthquake sizes when slip is allowed to penetrate below the seismogenic layer into an underlying zone characterized by velocity-strengthening behavior. Is this deep coseismic slip happening during large earthquakes? We point to a number of additional associated features of the model behavior which are potentially observable in the Earth. These include the predictions that a substantial fraction, of order a third of total coseismic moment, is due to slip below the seismogenic layer, and that slip below the seismogenic layer should be characterized by long risetimes and a dearth of high frequency motion.
S14B-07
Exploring the Relationship Between Early Rupture History and Final Earthquake Size
A question of considerable contention in the seismological community is whether earthquakes follow the cascade model or the preslip model of earthquake rupture. In the preslip model, earthquake ruptures are generated by an initial slip distribution in a nucleation zone, which loads the fault in proportion to the magnitude of this early slip. In this model, earthquake size is a deterministic result of the early rupture process, and can be determined through observations while the rupture is ongoing. In the cascade model, the rupture is initiated on a particular asperity on the fault, and that failure applies increased stress to adjacent portions of the fault. Each adjacent patch either propagates the rupture forward or stops its propagation, depending on whether local conditions on each individual patch are favorable to rupture or not. The local geometry of the fault and the strength of the host rock, as well as the local distribution of stresses, all affect how favorable a given fault patch is to rupture. Earthquake size in this model is inherently non-deterministic, as no portion of the rupture front feels the effects of barriers (unfavorable patches) until it has propagated into them and stopped. As a result, this model does not allow for the final size of the earthquake to be known until the rupture has completely stopped propagating. We use a Support Operators (Ely et al., 2007) model to simulate dynamic rupture of a rectangular planar fault with stochastic heterogeneous initial shear stress. By varying the initial shear stress near the point of nucleation we can control the intensity of the early rupture in terms of stress drop, moment release and other parameters. We investigate the effect of different characteristics of early rupture on the final distribution of slip on the fault. Results suggest that under realistic initial stress distributions, earthquake rupture exhibits elements of both cascade and preslip behavior, and that it is possible to alter significantly the final size of the earthquake with comparatively small changes in the character of the early rupture.
S14B-08
Correlations and Non-predictability in the Time Evolution of Earthquake Ruptures
The characterization of the time evolution of ruptures is one of the important aspects of the earthquake process. What makes a rupture, that starts small, to become a big one or end very quickly resulting in a small earthquake is central to understanding the physics of the time evolution of ruptures. Establishing whether there are any correlations in time, between the initiation of the rupture and its ultimate size, is a step in the right direction. Here, we analyze three source-time function data sets. The first is produced by the generation of repeated rupture events on a 2D heterogeneous, in-plane, dynamical model, while the second is produced by an-age dependent critical branching model. The third is the source-time function data base of Ruff [1]. We formulate the problem in terms of two questions. 1) Are there any correlations between the moment release at the beginning of the rupture and the total moment release during the entire rupture? 2) Can we predict the final size of an earthquake, once it has started and without any a posteriori information, by just knowing the moment release up to a certain time τ? Using the three data bases, the answer to the first question is yes and no to the second. The longer τ is, the stronger the correlations are between what goes on at the initiation and the final size. But, for τ fixed, and not a major fraction of the rupture time, there is no predictability of the rupture size. In particular, if a rupture starts with a very large moment release during time τ, it becomes a large earthquake. On the other hand, large earthquakes might start with very small moment release during τ; the non-predictability is due to the heterogeneities. The randomness in the critical branching model mimics the effect of the heterogeneities in the crust and in the 2D model. \begin{thebibliography}{99} \bibitem{ruff} Ruff, L. J., http://www.geo.lsa.umich.edu/SeismoObs/STF.html \end{thebibliography}