Earthquake Slip Distribution Research Articles

Global compilation of interferometric synthetic aperture radar earthquake source models: 2. Effects of 3-D Earth structure

[1] We carry out long-period surface wave centroid moment tensor (CMT) inversions using various global tomographic models and two different forward modeling techniques for 32 large earthquakes previously studied using interferometric synthetic aperture radar (InSAR) data. Since InSAR methods provide an alternative and independent way of locating and characterizing shallow continental earthquakes, comparisons of our source parameters with those from InSAR are a novel way to assess limitations in the InSAR models as well as the effects of inaccurate wave propagation formulations and/or 3-D Earth structure on earthquake source determinations. We show that comparing InSAR results with our seismic solutions is valuable to identify inaccuracies in the earthquake slip distribution retrieved using InSAR. Moreover, we find that using more accurate formulations, together with the best fitting Earth models, substantially reduces biases and differences between moment magnitude and fault strike determined using InSAR and seismic data. As expected for long-period surface wave source inversions for shallow earthquakes, the fault dip and rake angles are difficult to constrain, but we show that when using the best fitting Earth models, differences to InSAR estimates are reduced. Moreover, spurious deviations from a pure double-couple earthquake mechanism are on average smaller for the best fitting Earth models and the more accurate formulation of wave propagation. There are large differences between InSAR epicentral locations and those obtained in this study and, on average, no clear improvements to the Global CMT locations are achieved. This suggests that higher-resolution Earth models are necessary to further refine long-period CMT epicentral locations.

Fault Geometry and Slip Distribution of the 2010 Yushu Earthquakes Inferred from InSAR Measurement

We construct a coseismic deformation interferogram for the April 2010 Yushu earthquakes using ALOS/PALSAR data from the ascending track (path 487). We then infer the trace of the Yushu fault using the coherence image, and we build five fault models for the Yushu fault. To determine the fault geometry parameters that give the best fit to the coseismic interferogram, we apply an elastic dislocation algorithm. Our preferred fault model consists of two faults dipping to the northeast. One strikes ∼N60°W with a dip of 82°; the other strikes ∼N67°W with a dip of 86°. Lastly, we infer the coseismic slip distributions of the Yushu earthquakes by the inversion of the displacement in the line of sight (LOS). The results show that three high-slip concentrations are located at a depth of 5∼8 km, with a peak slip of 1.32 m at (96.93 E, 33.03 N). The Yushu fault is a left-lateral strike-slip faulting with small north-side-up, dip-slip components.

Bayesian networks for tsunami early warning

The various uncertainties in the earthquake-triggered tsunami threat assessment are difficult to quantify and/or integrate into the tsunami early warning process. Uncertainties in the (seismic) input parameters and the lack of knowledge about the earthquake slip distribution contribute most to the total uncertainty in real-time evaluated tsunami assessment. We present a method how to integrate and quantify these uncertainties in the warning process by evaluating a tsunami warning level probability distribution with a Bayesian network (BN) approach. As soon as an earthquake is detected, the seismic source parameter estimates are evaluated and a probabilistic overview on different tsunami warning levels is provided, feasible to support a decision maker at a warning center with important additional data. A BN system has been developed exemplarily for the region Sumatra. In this paper, we describe the method of BN generation by ancestral sampling, we critically analyse the assumptions made and weight the pro and cons of the BN approach. A case study demonstrates the workflow of the BN system in real-time and reveals the promising power of a BN analysis in the framework of early warning.

Long-term persistence of subduction earthquake segment boundaries: Evidence from Mejillones Peninsula, northern Chile

[1] The Mejillones Peninsula in northern Chile has been recognized as the surface expression of a segment boundary for large subduction zone earthquakes. The sharp contact between the rupture planes of two instrumentally recorded earthquakes, the Mw = 8.0 Antofagasta (1995) and the Mw = 7.7 Tocopilla (2007) events, is located beneath the central part of Mejillones Peninsula. We present new chronostratigraphic and structural data that allow reconstructing the evolution of the Peninsula at the surface and correlation of the latter with seismic cycle deformation on the plate interface. Uplift commenced after 3.4 Myr, as recorded in the western highland. The central graben area on the Peninsula started uplifting above sea level as an anticlinal hinge zone prior to 400 kyr ago, most probably 790 kyr ago. The resulting E-W trending hinge exactly overlies the limit between the rupture planes of the Antofagasta and Tocopilla earthquakes. By correlating the uplift data with the slip distribution of the above earthquakes, we demonstrate that deformation and uplift is focused during the postseismic and interseismic periods of the megathrust seismic cycle with coseismic deformation opposed to the long-term motion. Additionally, the slip deficit beneath the Peninsula accumulating between events is probably largely recovered by creep. Hence we suggest that Mejillones Peninsula owes its existence to the lateral variation of the propensity for unstable slip at the interface. Since the latter is a material property, the long-term spatial stability of the Peninsula as a barrier to rupture propagation since at least the middle Pleistocene is a necessary consequence.

Slip distribution of the 2003 Tokachi‐oki Mw 8.1 earthquake from joint inversion of tsunami waveforms and geodetic data

We study the 2003 Mw 8.1 Tokachi‐oki earthquake, a great interplate event that occurred along the southwestern Kuril Trench and generated a significant tsunami. To determine the earthquake slip distribution, we perform the first joint inversion of tsunami waveforms measured by tide gauges and of coseismic displacement measured both by GPS stations and three ocean bottom pressure gauges (PG) for this event. The resolution of the different data sets on the slip distribution is assessed by means of several checkerboard tests. Results show that tsunami data constrain the slip distribution offshore, whereas GPS data constrain the slip distribution in the onshore zone. The three PG data only coarsely constrain the offshore slip, indicating that denser networks should be installed close to subduction zones. Combining the three data sets significantly improves the inversion results. Joint inversion of the 2003 Tokachi‐oki earthquake data leads to maximum slip values (∼6 m) confined at depths greater than ∼25 km, between 30 and 80 km northwest of the hypocenter, with a patch of slip (3 m) in the deepest part of the source (∼50 km depth). Slip values are very low (≤1 m) updip from the hypocenter. Furthermore, the rupture does not extend on the plate interface off Akkeshi. As a significant back slip amount (∼4 m) has accumulated there since the last 1952 earthquake, this segment could rupture during the next large interplate event along the Kuril Trench.

2010 Maule earthquake slip correlates with pre-seismic locking of Andean subduction zone

The magnitude-8.8 Maule (Chile) earthquake of 27 February 2010 ruptured a segment of the Andean subduction zone megathrust that has been suspected to be of high seismic potential. It is the largest earthquake to rupture a mature seismic gap in a subduction zone that has been monitored with a dense space-geodetic network before the event. This provides an image of the pre-seismically locked state of the plate interface of unprecedentedly high resolution, allowing for an assessment of the spatial correlation of interseismic locking with coseismic slip. Pre-seismic locking might be used to anticipate future ruptures in many seismic gaps, given the fundamental assumption that locking and slip are similar. This hypothesis, however, could not be tested without the occurrence of the first gap-filling earthquake. Here we show evidence that the 2010 Maule earthquake slip distribution correlates closely with the patchwork of interseismic locking distribution as derived by inversion of global positioning system (GPS) observations during the previous decade. The earthquake nucleated in a region of high locking gradient and released most of the stresses accumulated in the area since the last major event in 1835. Two regions of high seismic slip (asperities) appeared to be nearly fully locked before the earthquake. Between these asperities, the rupture bridged a zone that was creeping interseismically with consistently low coseismic slip. The rupture stopped in areas that were highly locked before the earthquake but where pre-stress had been significantly reduced by overlapping twentieth-century earthquakes. Our work suggests that coseismic slip heterogeneity at the scale of single asperities should indicate the seismic potential of future great earthquakes, which thus might be anticipated by geodetic observations.

Effects of Kinematic Constraints on Teleseismic Finite-Source Rupture Inversions: Great Peruvian Earthquakes of 23 June 2001 and 15 August 2007

Two great underthrusting earthquakes that occurred along the coast of Peru in 2001 and 2007 involve spatiotemporal slip distributions that differ from the predominantly unilateral or bilateral rupture expansion of many great events. Commonly used finite-source rupture model parameterizations, with specified rupture velocity and/or short duration of slip at each grid point applied to the seismic data for these two events, lead to incorrect slip-distributions or inaccurate estimation of rupture velocities as a result of intrinsic kinematic constraints imposed on the model slip distributions. Guided by large aperture array back projections of teleseismic broadband P-wave signals that image slip locations without imposing a priori kinematic constraints on the rupture process, we exploit the availability of large global broadband body and surface wave data sets to consider the effects of varying the kinematic constraints in teleseismic finite-source waveform inversions. By allowing longer than usual rupture durations at each point on the fault using a flexible subfault source-time function parameterization, we find that the anomalous attributes of the 2001 and 2007 Peru earthquake ruptures are readily recognized and accounted for by compound rupture models. The great 23 June 2001 (M_w 8.4 8.4) earthquake involved an initial modest-size event that appears to have triggered a much larger secondary event about 120 km away that developed an overall slip distribution with significant slip located back along the megathrust in the vicinity of the initial rupture. The great 15 August 2007 (M_w 8.0 8.0) earthquake was also a composite event, with a modest size initial rupture followed by a 60-sec delayed larger rupture that initiated 50–60 km away and spread up-dip and bilaterally. When back projections indicate greater rupture complexity than captured in a simple slip-pulse-type rupture model, one should allow for possible long-subfault slip-duration or composite triggered sequences, and not overly constrain the earthquake slip distribution.

Constraining the roughness degree of slip heterogeneity

This article investigates different approaches for assessing the degree of roughness of the slip distribution of future earthquakes. First, we analyze a database of slip images extracted from a suite of 152 finite‐source rupture models from 80 events (Mw = 4.1–8.9). This results in an empirical model defining the distribution of the slip spectrum corner wave numbers (kc) as a function of moment magnitude. To reduce the “epistemic” uncertainty, we select a single slip model per event and screen out poorly resolved models. The number of remaining models (30) is thus rather small. In addition, the robustness of the empirical model rests on a reliable estimation of kc by kinematic inversion methods. We address this issue by performing tests on synthetic data with a frequency domain inversion method. These tests reveal that due to smoothing constraints used to stabilize the inversion process, kc tends to be underestimated. We then develop an alternative approach: (1) we establish a proportionality relationship between kc and the peak ground acceleration (PGA), using a k−2 kinematic source model, and (2) we analyze the PGA distribution, which is believed to be better constrained than slip images. These two methods reveal that kc follows a lognormal distribution, with similar standard deviations for both methods.

Stress transfer and its implication for earthquake hazard on the Kunlun Fault, Tibet

The 1600-km-long Kunlun Fault striking E–W to WNW–ESE had long been recognized as one of the major left-lateral strike-slip faults bounding the Tibetan Plateau, and ranked one of the most active faults in China continent. During the past hundred years, over twenty strong earthquakes occurred along and near the Kunlun Fault, including six large earthquakes ( M > 7). Since some major highly-populated and industrialized cities are close to the Kunlun Fault, understanding of stress transfer and earthquakes migration along the Kunlun Fault is most important for assessing seismic hazard in this region. In this study, by integrating coseismic effect, viscoelastic relaxation and tectonic loading, we studied the evolution of the regional Coulomb stress field by analyzing a sequence of strong earthquakes along the Kunlun Fault. We studied the stress evolution over one century by analysing a sequence of five earthquakes ( M ≥ 7) that occurred along the Kunlun Fault since 1937. The model of dislocation sources embedded in a mixed elastic/inelastic layered half-space was used, and the layered model and relevant parameters were constrained by seismic studies. Fault rupture locations and geometry, as well as slip distribution of earthquakes were taken from field observations and seismic studies. Numerical results showed a good correlation between stress transfer, accumulation and earthquakes occurrence. All four studied earthquakes occurred after the 1937 Tuosuo Lake quake were encouraged by the preceding earthquakes with positive stress loading. In subject to the choice of the earthquake source parameter two or three out of four events occurred in regions that experienced previous coseismic and postseismic stress changes of at least 0.01 MPa, suggesting that earthquake triggering due to stress transfer has occurred along the Kunlun Fault. The total stress change since 1937 of the Kunlun Fault region has lead to high levels of stress accumulated on the Xidatan–Dongdatan segment and Maqin–Maqu segment, which have not experienced any significant large earthquake over at least several hundred or several thousand years. The accumulated stress raises the potential earthquake hazard in these areas. Our study demonstrated the crucial importance of postseismic viscoelastic relaxation in the stress transfer and accumulation following large earthquakes.

A subsurface stress analysis and its possible relation with seismicity near the Itoiz Reservoir, Navarra, Northern Spain

We study the effects of surface water loads and moderate-magnitude earthquakes on the subsurface state of stress and its relation with seismicity, near the newly constructed Itoiz reservoir in the western Pyrenees, northern Spain. To do this, we initially compute the evolution of the stress changes in the subsoil due to the water load distribution during the impounding of the reservoir, and relate it with the main seismicity occurred after the beginning of impoundment. After this, we compute the coseismic and postseismic stress changes produced by the main events of the seismic series and study its influence on the triggering of the aftershocks. Stress changes due to the water loads are computed by means of the Boussinesq solution for a homogeneous elastic half-space whereas the coseismic stress changes due to earthquakes are obtained using the equations for an internal elastic dislocation in a half space. The effects over the possible seismogenic faults are in both cases, accounted by means of the Coulomb Failure Stress criterion (ΔCFS). We considered different threshold values and apparent friction coefficients of ΔCFS for the effective static triggering. Then, we calculate the pore pressure changes produced by the time evolution of the surface water loads, as also the pore pressure changes generated by the elastic dislocations due to the largest earthquake of the series (mainshock). The fault parameters for the computation of the ΔCFS of the perturbing seismic events were selected from published focal solutions. To define the target fault parameters we took into account the focal solutions of the earthquakes and aftershocks themselves, the parameters of mapped faults based on the tectonic setting and those of optimally oriented faults. Due to the absence of reliable estimations for the slip distributions of the main earthquakes, we assumed for stress computations a semi-elliptical distribution of slips over the fault. Results show the existence of a mainly positive stress influence of the surface water loads on the main earthquake of the series, in two different senses: the elastic Coulomb failure stress and the change in pore pressure. However, due to different uncertainties on the possible values of the apparent coefficient of friction ( µ') for the zone, there is not a clear indication of an effective earthquake triggering of the mainshock due to the elastic stress changes by the surface water loads. On the other hand, a more clear effect is found on the possible triggering of aftershocks due to the largest earthquakes of the series. It is also found a positive pore pressure influence of the two largest events on most of the aftershocks.

Effect of a compliant fault zone on the inferred earthquake slip distribution

We present a new semi‐analytic method to evaluate the deformation due to a screw dislocation in arbitrarily heterogeneous and/or anisotropic elastic half plane. The method employs integral transformations to reduce the governing partial differential equations to the integral Fredholm equation of the second kind. Dislocation sources, as well as spatial perturbations in the elastic properties are modeled using equivalent body forces. The solution to the Fredholm equation is obtained in the Fourier domain using a method of successive over‐relaxation, and is mapped into the spatial domain using the inverse Fast Fourier Transform. We apply this method to investigate the effect of a soft damage zone around an earthquake fault on the co‐seismic displacement field, and on the earthquake slip distribution inferred from inversions of geodetic data. In the presence of a kilometer‐wide damage zone with a reduction of the effective shear modulus of a factor of 2, inversions that assume a laterally homogeneous model tend to underestimate the amount of slip in the middle of the seismogenic layer by as much as 20%. This bias may accentuate the inferred maxima in the seismic moment release at depth between 3–6 km suggested by previous studies of large strike‐slip earthquakes.

Distribution of slip from 11 Mw > 6 earthquakes in the northern Chile subduction zone

We use interferometric synthetic aperture radar, GPS, and teleseismic data to constrain the relative location of coseismic slip from 11 earthquakes on the subduction interface in northern Chile (23°–25°S) between the years 1993 and 2000. We invert body wave waveforms and geodetic data both jointly and separately for the four largest earthquakes during this time period (1993 Mw 6.8; 1995 Mw 8.1; 1996 Mw 6.7; 1998 Mw 7.1). While the location of slip in the teleseismic‐only, geodetic‐only, and joint slip inversions is similar for the small earthquakes, there are differences for the 1995 Mw 8.1 event, probably related to nonuniqueness of models that fit the teleseismic data. There is a consistent mislocation of the Harvard centroid moment tensor locations of many of the 6 < Mw < 8 earthquakes by 30–50 km toward the trench. For all models, the teleseismic data are better able to resolve fine details of the earthquake slip distribution. The 1995 earthquake did not rupture to the maximum depth of the seismogenic zone (as defined by the other earthquakes). In addition to the above events, we use only teleseismic data to determine the rupture characteristics of four other Mw > 6 earthquakes, as well as three Mw > 7 events from the 1980s. All of these earthquakes appear to rupture different portions of the fault interface and do not rerupture a limited number of asperities.

Spatial relation between the 1908 Messina Straits earthquake slip and recent earthquake distribution

We re‐analyze coseismic surface displacement data of the 1908 Messina earthquake, taking into account both correlated and uncorrelated uncertainties, to re‐determine the slip distribution on the fault plane. Using a data set of earthquakes occurred at crustal depth in the Messina Straits between July 1988 and December 2004, we evidence a peculiar recent seismicity distribution inside the fault volume of the 1908 Messina Straits earthquake with respect to other regions in the area. An unusually large number of earthquakes is found in regions smoothly ringing slipped areas. Our results seem consistent with static Coulomb stress variations inside the fault volume and regional strain rate from GPS data, but could also suggest that in the last 15–20 years the 1908 earthquake source has been releasing strain accumulated in portions of it which did not rupture during the 1908 event.

High precision slip distribution of the 1944 Tonankai earthquake inferred from tsunami waveforms: Possible slip on a splay fault

Inversion of tsunami waveforms is a well-established technique for estimating the slip distributions of subduction zone earthquakes, with some of the most detailed results having been obtained for earthquakes in the Nankai Trough, SW Japan. The present study, although it uses a method and tsunami waveform data set almost identical to previous study, aims to improve on previous work by using a more precise specification of initial conditions for the calculation of tsunami Green's functions. Specifically, we incorporated four improvements in the present study: (1) we used a realistic plate model based only on seismic survey results, and assumed it to be the fault plane of the 1944 Tonankai earthquake; (2) the smallest subfaults consistent with the long wavelength approximation were used in the tsunami inversion analysis; (3) we included the effect of horizontal displacement of the ocean bottom on tsunami generation; and (4) we performed a checkerboard resolution test. As obtained in previous studies, a zone of high slip (> 2.0 m) was resolved off the Shima Peninsula. However, the more precise calculation of tsunami Green's functions has revealed additional detail that was not evident in previous studies, which we demonstrate is resolvable and correlates with the position of known faults in the accretionary prism. While there was little or no slip near the trench axis in the eastern part of the rupture zone, there was up to 1.5 m of slip resolved within 30 km of the trough axis in the western part, along the coast of the Kii Peninsula. This troughward slip zone coincides with the position of a large splay fault mapped in multichannel reflection surveys. Furthermore, it is also clear that the upper edge of the Enshu fault off Shima and Atsumi peninsulas is consistent with the up-dip limit of slip in the eastern part of our model. We tested the possibility that slip occurred on the former splay fault instead of on the plate interface during the 1944 Tonankai earthquake, and find that slip on this splay fault is also consistent with the data, although we cannot distinguish whether slip was dominant on the splay fault or on the plate interface. We further suggest that the position of the Enshu fault may be determined by the subduction of topographic highs, and that such faults may have an important influence on the up-dip rupture limit of the 1944 Tonankai and, potentially, other subduction zone earthquakes.

Stochastic model of heterogeneity in earthquake slip spatial distributions

SUMMARY Finite-fault source inversions reveal the spatial complexity of earthquake slip or pre-stress distribution over the fault surface. The basic assumption of this study is that a stochastic model can reproduce the variability in amplitude and the long-range correlation of the spatial slip distribution. In this paper, we compute the stochastic model for the source models of four earthquakes: the 1979 Imperial Valley, the 1989 Loma Prieta, the 1994 Northridge and 1995 Hyogo-ken Nanbu (Kobe). For each earthquake (except Imperial Valley), we consider both the dip and strike slip distributions. In each case, we use a 1-D stochastic model. For the four earthquakes, we show that the average power spectra of the raw, that is, non-interpolated, data follow a power-law behaviour with scaling exponents that range from 0.78 to 1.71. For the four earthquakes, we have found that a non-Gaussian probability law, that is, the Levy law, is better suited to reproduce the main features of the spatial variability embedded in the slip amplitude distribution, including the presence and frequency of large fluctuations. Since asperities are usually defined as regions with large slip values on the fault, the stochastic model will allow predicting and modelling the spatial distribution of the asperities over the fault surface. The values of the Levy parameters differ from one earthquake to the other. Assuming an isotropic spatial distribution of heterogeneity for the dip and the strike slip of he Northridge earthquake, we also compute a 2-D stochastic model. The main conclusions reached in the 1-D analysis remain appropriate for the 2-D model. The results obtained for the four earthquakes suggest that some features of the slip spatial complexity are universal and can be modelled accordingly. If this is proven correct, this will imply that the spatial variability and the long-range correlation of the slip or pre-stress spatial distribution can be described with the help of five parameters: a scaling exponent controlling the spatial correlation and the four parameters of the Levy distribution constraining the spatial variability.

Evidence for an earthquake barrier model from Mw∼7.8 Kokoxili (Tibet) earthquake slip-distribution

Abstract The slip distribution of the Mw ∼ 7.8 Kokoxili (Tibet, 2001) earthquake has been measured at high resolution using optical correlation of satellite images and provides both the parallel and perpendicular components of the horizontal co-seismic slip. This reveals a variation of the horizontal slip at a scale of ∼ 20 km along-strike. Anti-correlation of slip parallel and perpendicular to the fault indicates transfer of slip from the horizontal to the vertical component at the ends of segments. These features suggest a rupture model with segments separated by strong persistent geometric barriers. The unexpected ending of the rupture south of the main fault can be explain by such a structure but bears important implications for the initiation and rupture directivity of the next earthquake.

Earthquake Simulation by Restricted Random Walks

This article simulates earthquake slip distributions as restricted random walks. Random walks offer several unifying insights into earthquake behaviors that physically based simulations do not. With properly tailored variables, random walks generate observed power law rates of earthquake number versus earthquake magnitude (the Gutenberg–Richter relation). Curiously, b -value, the slope of this distribution, not only fixes the ratio of small to large events but also dictates diverse earthquake scaling laws such as mean slip versus fault length and moment versus mean slip. Moreover, b -value determines the overall shape and roughness of earthquake ruptures. For example, mean random walk quakes with b = −½ have elliptical slip distributions characteristic of a uniform stress drop on a crack. Random walk earthquake simulators, tuned by comparison with field data, provide improved bases for statistical inference of earthquake behavior and hazard.

Sensitivity of static stress calculations to the earthquake slip distribution

The apparent strong correlation between Coulomb stress changes and the spatial distribution of aftershocks suggests the possibility of making near‐real‐time estimations of areas at risk of experiencing off‐fault aftershocks. In order to do this in practice a number of issues must first be addressed, including the extent to which the main shock slip must be known in detail before a meaningful stress map can be constructed. Here we investigate this issue by constructing a time‐ordered sequence of slip solutions for the Landers earthquake, computing Coulomb stress changes for each solution, and quantitatively comparing the stress field with the observed aftershocks by (1) resolving the Coulomb stress change onto the aftershock nodal planes and calculating the percentage of events consistent with triggering and (2) constructing a two‐dimensional map of Coulomb stress and computing the correlation coefficient between the positive and negative areas and the locations of the aftershocks. We find that slip solutions based on empirical relations and either focal mechanism or moment tensor data produce stress fields inconsistent with the observed spatial distribution of aftershocks, whereas slip solutions incorporating the correct rupture geometry but greatly simplified slip produce stress fields consistent with the aftershock distribution when very near‐fault events are excluded. We further find that resolving stress perturbations onto earthquake nodal planes and computing the percentage of events experiencing positive Coulomb stress provides a poor measure of success because of the limited range of structures on which events occur and the compatibility of the main shock stress field with these structures. Our results support the hypothesis that Coulomb stress changes affect the spatial distribution of aftershocks and suggest that meaningful calculations of Coulomb stress can be made as soon as an earthquake's rupture geometry is well constrained.

Surface Rupture on the Denali Fault Interpreted from Tree Damage during the 1912 Delta River Mw 7.2-7.4 Earthquake: Implications for the 2002 Denali Fault Earthquake Slip Distribution

During the 3 November 2002 Denali fault earthquake, surface rupture propagated through a small, old-growth forest in the Delta River valley and damaged many trees growing on the fault. Damage was principally the result of fault offset of tree roots and tilting of trees. Some trees were split by surface faults that intersected the base of their trunks or large taproots. A few trees appear to have been damaged by strong shaking. Many of the older trees damaged in 2002 were deformed and scarred. Some of these scarred trees exhibit past damage indicative of surface faulting and have abrupt changes in their annual ring patterns that coincide with the past damage. Annual ring counts from several of these older scarred trees indicate the damage was caused by surface rupture on the Denali fault in 1912. The only earthquake of sufficient magnitude that fits the requirements for timing and general location as recorded by the damaged trees is a widely felt M s 7.2–7.4 earthquake on 6 July 1912 informally referred to as the 1912 Delta River earthquake. Seismologic data and intensity distribution for the 1912 Delta River earthquake indicate that its epicenter was within 60–90 km of the Delta River and that rupture probably propagated toward the west. Inferred fault length, displacement, and rupture direction suggest the 1912 rupture was probably largely coincident with the western, lower slip section of the 2002 rupture.

Rupture behavior of the 1999 Chi-Chi, Taiwan, earthquake—slips on a curved fault in response to the regional plate convergence

Simple kinematic modeling of particle motion along a curved fault similar to the ruptured Chelungpu fault, Taiwan, indicates a unique spatial slip pattern. Specifically, we find that the large convergent slip on the curve is a result of the minimum deformation scenario of the fault geometry and regional northwestern movement of the Philippine Sea Plate (PSP). The modeled deformation regime portrays an accumulation of deformation in the curved region, which coincides very well with a long-term observed NW-SE-trending seismogenic zone in the central Taiwan. This consistence suggests that the Chelungpu fault is a preexisting curved fault. This is further evidenced by geological and geophysical observations. Because the spatial slip pattern is locally and regionally tectonically controlled, it indicates that the rupture behavior of the Chi-Chi earthquake is repeatable. Better knowledge of the fault geometry and the regional plate motion may help us to predict the possible spatial slip distribution of large earthquakes. This discovery is important for avoiding large buildings and constructions near predicted large slip regions.