S32B-01 10:20h
Internal Structure of a Strike-Slip Dilational Fault Jog: Overlander Fault, Mt Isa Inlier, Australia
The Overlander Fault is one of a set of NE-SW subvertical dextral strike-slip faults which, together with a NW-SE conjugate sinistral set, disrupt the Mt Isa Proterozoic orogen (1590-1500 Ma) in NW Queensland, Australia. These late- to post-orogenic faults thus define a regional stress field with $\sigma$$_{1}$ oriented approximately E-W and $\sigma$$_{3}$ oriented approximately N-S. The Overlander Fault trends $\sim$$060\deg$ across the metamorphic assemblage except where it refracts to 070-$074\deg$ across an outcropping granitic pluton, the margins of which it offsets dextrally by $\sim$1.5 km. The stepover width of this dilational fault jog approaches 1 km, comparable to dilational stepovers within active strike-slip faults (e.g. the San Andreas fault at Parkfield). In the surrounding amphibolite facies metamorphic assemblage the fault trace is comparatively inconspicuous and unmineralized but where it crosses the granite it is defined by upstanding ridges of silicified microbreccia and associated quartz veining. The stepover region provides opportunities for studying incremental and finite dilatation associated with slip transfer across the jog, and associated influx of hydrothermal fluids. Shearing across the stepover region is accommodated by a mesh structure with principal components that include: (1) a series of silicified microbreccia-cataclasite `walls' $<$10 m or so thick with associated quartz veins $<$1 m or so thick trending $070\deg$ and defining a `main zone' about 100$\pm$20 m wide; (2) parallel subsidiary strike-slip cataclastic shear zones occurring $<$200 m laterally from the main zone; (3) a set of subvertical $<$1-2 m thick extension veins oriented 090-$100\deg$, some with evidence of marginal shearing (both sinistral and dextral); (4) a conspicuous sinistral extensional-shear curving eastwards for $\sim$250 m from the main fault core on a trend of 100-$115\deg$; and (5) a set of unmineralized faults with sinistral separations trending 120-$130\deg$. Slickenfibers and striations along the main fault-parallel components indicate predominantly strike-slip motion on subvertical planes but there is some local evidence for dip-slip. Mutual cross-cutting relationships between all the principal components indicate penecontemporaneous development. Orientations of the various components are broadly compatible with the inferred regional stress field but there is some evidence for fluctuating stress trajectories. Vein textures record histories of incremental growth and are generally consistent with hydrothermal deposition under low effective stresses, probably in the epizonal environment ($<$1-2 km depth). Recorded dextral separations along the major shear fracture components are commonly of the order of 1-10 cm, consistent with small-to-moderate seismic slip increments. It is common for shearing increments along the main strike-slip faults in the stepover to be accompanied by significant dilatation. Rigid-body analysis of the deflected fault trace across the granite suggests that total dilatation in the stepover should be of the order of 300 m but the major veins account for c. 10-20$%$ of this. Nonetheless, it is clear that progressive slip transfer across the jog involved substantial dilatation and massive fluid influx. The complex array of sub-structures within the stepover mesh invites comparison with structural complexity revealed by high resolution aftershock studies of dilational jog structures on seismically active strike-slip faults (e.g. Parkfield, Coyote Lake, Landers earthquakes) which are recognized sites of rupture perturbation or arrest.
S32B-02 10:35h
Space Geodetic Constraints on the Structure and Properties of Compliant Damage Zones Around Major Crustal Faults
Geologic and seismologic studies of large crustal faults indicate that the fault interface that accommodates most of seismic slip is often surrounded by heavily damaged material characterized by high crack density and reduced seismic velocities. Recently such damage zones were imaged by space geodetic observations. I present results of Interferometric Synthetic Aperture Radar (InSAR) observations of deformation across kilometer-wide compliant fault zones in response to nearby earthquakes. In particular, a number of faults in the Eastern California Shear Zone, including the Calico, Rodman, Pinto Mountain, and Lenwood faults, were strained by both the 1992 Landers and the 1999 Hector Mine earthquakes. Analysis of deformation on these faults indicates that the fault zone displacements depend on the magnitude, but are independent of the sign of the co-seismic stress changes, implying a linearly elastic deformation. Other examples include faults adjacent to the North Anatolian fault (Turkey) that were strained by the 1999 Izmit earthquake. Analytic and numerical (finite element) modeling of the observed deformation suggests that the compliant fault zones have width of 1-2 km, depth extent of several km (or greater), and reductions in the effective shear modulus of about a factor of two. Stacked interferometric data from the Eastern California Shear Zone spannig a time period of more than 10 years reveal time-dependent (post- or inter-seismic) deformation on some of the inferred compliant fault zones. In particular, the fault zone associated with the Pinto Mountain fault was subsiding over several years following the Landers eartquake, with the total amplitude of subsidence comparable to the amplitude of the co-seismically-induced uplift. This behavior may be indicative of the poro-elastic deformation of the fluid-saturated fault zone.
http://sioviz.ucsd.edu/~fialko/res_compliant.html
S32B-03 10:50h
Modelling Crustal Stress in Southern Ontario
Analysis of stress measurement data from the near-surface to crustal depths in southern Ontario show a misalignment between the direction of tectonic loading and the orientation of the major horizontal principal stress. The compressive stress field appears to be oriented subparallel to the major terrane boundaries such as the Grenville Front, the Central Medisedimentary Belt boundary zone and the Composite Arc Belt boundary zone. This suggests that the stress field has been modified by these deep crustal scale fault zones. In order to test this hypothesis, a geomechanical model was constructed using the three-dimensional discontinuum stress analysis code 3DEC. The model consists of a 45 km thick crust of southern Ontario in which the major crustal scale fault zones are represented as discrete faults. Lateral velocity boundary conditions were applied to the sides of the model in the direction of tectonic loading in order to generate the horizontal compressive stress field. Preliminary results show that for low strength (low friction angle and cohesion), fault slip results in the stress field rotating toward the strike of the faults, consistent with the observed direction of misalignment with the tectonic loading direction. Further analysis of the model may be used to provide constraints on the strength of these faults.
S32B-04 11:05h
Estimation of Fault Strength Before the 1995 Kobe Earthquake
We propose a new method to estimate the strength of a fault. The most direct way to measure fault strength is to observe the stress field just before an earthquake. However, it is difficult to monitor in-situ stress continuously, making the observation of pre-shock stress nearly impossible. Instead, we have reconstructed the stress field before the 1995 Kobe earthquake ({\it M}$_{w}$6.9) using in-situ post-shock stress measurements by hydraulic fracturing experiments and a kinematic source model. The in-situ stress measurements were taken about one year after the earthquake at four sites near the fault (Ikeda {\it et al}., 2001; Tsukahara {\it et al}., 2001). The background stress field in this region was considered to be east-west compression from the focal mechanisms of small earthquakes (Katao {\it et al}., 1997). After the earthquake, however, the maximum horizontal stress direction around the fault was estimated to be northwest-southeast, which is perpendicular to the fault, from in-situ stress measurements and the mechanisms of aftershocks. This rotation of the principal stress direction was due to the fault slip of the earthquake. We estimated the pre-shock stress field by subtracting the stress change due to the coseismic slip from the post-shock stress field. For a kinematic source model, we selected Yoshida's model inverted from geodetic and seismic data (Yoshida {\it et al}., 1996). The estimated stress state shows that at the center of the fault, the maximum principal stress direction is east-west, which is consistent with the background stress before the earthquake. Moreover, the horizontal differential stress was estimated to be large: linear gradient of the differential stress was estimated to be 19 MPa/km. On the contrary, at both edges of the fault, the stress field did not change due to the earthquake and its principal direction was perpendicular to the fault. After the estimation of the pre-shock stress field, we were then able to estimate the coefficient of static friction using the shear and normal stress on the fault plane. At each depth, pore pressure is assumed to be hydrostatic. The average coefficient of static friction at the center of the fault was estimated to be 0.60, which is consistent with Byerlee's law in the laboratory (Byerlee, 1978). The strength of the fault is found to be equivalent to the strong crust. On the other hand, the coefficients at the edge of the fault were estimated to be very small (0-0.25), which suggests that the faults in those regions could not sustain the shear stress and an aseismic slip might occur even before the earthquake.
S32B-05 INVITED 11:20h
Variation of large elastodynamic earthquakes on complex fault systems
One of the biggest assumptions, and a source of some of the biggest uncertainties in earthquake hazard estimation is the role of fault segmentation in controlling large earthquake ruptures. Here we apply a new model which produces sequences of elastodynamic earthquake events on complex segmented fault systems, and use these simulations to quantify the variation of large events. We find a number of important systematic effects of segment geometry on the slip variation and the repeat time variation of large events, including an increase in variation at the ends of segments and a decrease in variation for the longest segments. We find both quantitative and qualitative differences between slip variation and time variation, so slip variation and time variation are not simple proxies for eachother. The model both generates self-consistent complex fault geometries, and generates self-consistent elastodynamic events on those geometries. This geometrical self-consistency is important in insuring strain is compatibly accommodated in t he long run over many earthquake cycles. The self-consistency also reduces the number of things which must be specified, by allowing the fault system to self-organize from a simple physics. Because of the numerical efficiency of the model, we can generate long sequences of events, and study the statistics of the populations. The long sequences are critical here in that the stresses left over by previous events form the setting for subsequent events. With this model, we can thus begin to address the fundamental questions of the interaction of geometry and dynamics over many earthquake cycles.
S32B-06 INVITED 11:35h
3D non-Planar Finite Difference Dynamic Rupture: Application to the Landers Earthquake
Many aspects of seismic complexity have been explained in the last thirty years thanks to the development of numerical approaches allowing seismologists to simulate the dynamic rupture of earthquakes. Heterogeneities in both the initial stress field and the surrounding medium are extremely important elements. The constitutive law describing the physics of the breakdown process which relates the fault friction to fault kinematics is also determinant. However, given the increasing amount of high quality seismological data, more sophisticated approaches are needed to explain observations so that others important physical factors, such as the real fault geometry, could be integrated into simulations. Bearing in mind these new high quality observations along with the current computational power, a great interest has arisen in the last five years to develop 3D numerical codes to simulate earthquakes with real fault geometries. Recently, Cruz-Atienza and Virieux (2004) have introduced a 2D finite difference (FD) approach for modeling the dynamic rupture of non-planar faults. In this work we analyze the 3D extension of such an approach. On that account, the new 3D code may consider arbitrary heterogeneous media, composite friction laws and non-planar fault geometries. The numerical criteria for rupture boundary conditions to model rupture processes accurately were determined experimentally finding consistency with those determined for the 2D case: the source is discretized by a set of numerical cells. Given a spatial grid step for wave propagation, the number of grid nodes contained in each cell should be adapted accordingly. The smaller the spatial step the greater the number of nodes. We have performed dynamic rupture simulations for different curved 3D faults and compared results with those given by a BIE method (Aochi et al., 2000). Consistency between solutions yielded by different numerical approaches is essential since it is the only way to have confidence in these kinds of complex simulations for which no theoretical solutions are available. This benchmarking exercise has also allowed us to better understand and quantify the effect of fault curvature on near-source seismograms and fault solutions. Finally, we applied our numerical approach to model the 1992 Landers earthquake (Mw=7.3). Several simulations were carried out including a heterogeneous initial stress field, layered elastic medium and the non-planar fault trace geometry. Complexity in near-field seismograms enhances the importance of both a heterogeneous surrounding medium and non-planar fault geometry due to their intimate interaction during rupture process. Aochi, H., E. Fukuyama and M. Matsu'ura, 2000, Pure. Appl. Geophys., 157, 2003-2027. Cruz-Atienza, V.M. and J. Virieux, 2004, Geophys. J. Int., 158, 939-954.
S32B-07 11:50h
3D Hybrid Numerical Modeling of Seismic Motion in Sedimentary Valleys due to a Dynamic Source Model
We have developed a 3D hybrid modeling technique based on the combination of the finite-element (FE) and finite-difference (FD) methods to simulate dynamic rupture propagation on a fault, seismic wave radiation and propagation in a heterogeneous viscoelastic medium with realistic model of the attenuation. In an arbitrarily shaped part of the whole computational region, the 2nd-order displacement FE method is used to simulate rupture propagation on a possibly non-planar fault. The FE part of the region may also have a non-planar free surface. The rest of the computational region is solved by the 4th-order velocity-stress staggered-grid FD scheme. The schemes may share the same time step while the major part of the model for the seismic wave propagation away from the radiating fault is efficiently covered by the grid with twice larger spatial grid spacing. The FE and FD schemes communicate at each time level in the transition zone in which both spatial grids overlap. The TSN (traction-at-split-node) method (Andrews 1999) is used for modeling the rupture propagation. The numerical tests show very good accuracy of the hybrid modeling. We apply the method to model earthquake ground motions in the Grenoble Valley, France, due to two hypothetical earthquakes on the Belledonne fault. We analyze the effect of different positions of the earthquake hypocenter on the ground motion in the sediment valley. We also compare the motion due to dynamic rupture on the fault with that produced by the equivalent double-couple point source for both considered earthquakes.
S32B-08 12:05h
Spectral Element simulation of rupture dynamics on curvilinear faults
Numerical simulation of fault rupturing process requires today the resolution of several time and space scales, to capture the nucleation, the rupture front propagation, and the short wave radiation associated with heterogeneous fault systems of complexgeometries. Two classes of methods are usually used in seismology: finite differences and boundary integral equations. Classical mixed formulation of finite differences suffers from smoothing and smearing of the rupture front due to the inherent interpolation of staggered schemes. Although if extensions to curved faults have recently been proposed (Cruz-Atienza and Virieux, 2004), using Saenger's stencils, up to now applications of FD methods have been mostly restricted to planar faults. On the other hand, boundary integral equations (Andrews, 1976; Fukuyama and Madariaga, 2000) have been shown to accurately model 3D curvilinear fault segments but are is restricted to homogeneous or layered elastic media. A important issue, still be correctly resolved is the physics of the rupture propagation when reaching the surface. In this framework, Spectral Element method, combining both the geometrical flexibility of finite elements and convergence rate of high-order spectral methods is an attractive tool for numerical simulation of earthquake dynamic rupturing on realistic fault segments in complex geological media. We present numerical simulations of 2D inplane dynamic faulting using the SE method. The results are discussed paying a special attention to the sub- to super-shear transition for both planar and non planar faults, to the influence of different frictional laws on the rupture propagation and to the influence of layered geolgical media both on the dynamics of the rupture process and the short wave radiation. On going work on two main extensions will be discussed : interactions as the faulting process reach the surface and 3D geometries of faults.