Search

The 3 September 2016 M w 5.8 Pawnee earthquake in northern Oklahoma is the largest earthquake ever recorded in Oklahoma. The coseismic deformation was measured with both Interferometric Synthetic Aperture Radar and Global Positioning System (GPS), with measureable signals of order 1 cm and 1 mm, respectively. We derive a coseismic slip model from Sentinel‐1A and Radarsat 2 interferograms and GPS static offsets, dominated by distributed left‐lateral strike slip on a primary west‐northwest–east‐southeast‐trending subvertical plane, whereas strike slip is concentrated near the hypocenter (5.6 km depth), with maximum slip of ∼1 m located slightly east and down‐dip of the hypocenter. Based on systematic misfits of observed interferogram line‐of‐sight (LoS) displacements, with LoS based on shear‐dislocation models, a few decimeters of fault‐zone collapse are inferred in the hypocentral region where coseismic slip was the largest. This may represent the postseismic migration of large volumes of fluid away from the high‐slip areas, made possible by the creation of a temporary high‐permeability damage zone around the fault.

Understanding fault slip rates in the eastern California shear zone (ECSZ) using GPS geodesy is complicated by potentially overlapping strain signals due to many sub-parallel strike-slip faults and by inconsistencies with geologic slip rates. The role of fault system geometry in describing ECSZ deformation may be investigated with total variation regularization, which algorithmically determines a best-fitting geometry from an initial model with numerous faults, constrained by a western United States GPS velocity field. The initial dense model (1) enables construction of the first geodetically constrained block model to include all ECSZ faults with geologic slip rates, allowing direct geologic-geodetic slip rate comparisons, and (2) permits fault system geometries with many active faults that are analogous to distributed interseismic deformation. Beginning with 58 ECSZ blocks, a model containing 10 ECSZ blocks is most consistent with geologic slip rates, reproducing five of 11 within their reported uncertainties. The model fits GPS observations with a mean residual velocity of 1.5 mm/yr. Persistent geologic-geodetic slip rate discrepancies occur on the Calico and Garlock faults, on which we estimate slip rates of 7.6 mm/yr and

The M w 6.0 South Napa earthquake, which occurred at 10:20 UTC 24 August 2014 was the largest earthquake to strike the greater San Francisco Bay area since the M w 6.9 1989 Loma Prieta earthquake. The rupture from this right‐lateral earthquake propagated mostly unilaterally to the north and up‐dip, directing the strongest shaking toward the city of Napa, where peak ground accelerations (PGAs) between 45% g and 61% g were recorded and modified Mercalli intensities (MMIs) of VII–VIII were reported. Tectonic surface rupture with dextral slip of up to 46 cm was observed on a 12.5 km long segment, some of which was along a previously mapped strand of the West Napa fault system, although the rupture extended to the north of the mapped Quaternary strand. Modeling of seismic and geodetic data suggests an average coseismic slip of 50 cm, with a maximum slip of about 1 m at depths of 10–11 km. We observed up to 35 cm of afterslip along the surface trace in the week following the mainshock, primarily along the southern half of the surface rupture that experienced relatively little coseismic offset. Relocation of the sparse aftershock sequence suggests en echelon southwest‐ and northeast‐dipping fault planes, reflective of the complex fault geometry in this region. The Napa basin and historic and late Holocene alluvial flood deposits in downtown Napa amplified the ground motions there. Few ground failures were mapped, reflecting the dry season (as well as a persistent drought that had lowered the groundwater table) and the short duration of strong shaking in the epicentral area. The South Napa fault rupture lies within an 80 km wide set of major north‐northwest‐trending faults of the San Andreas fault system, forming the boundary between the Pacific and North American tectonic …

Fault slip may involve slow aseismic creep and fast seismic rupture, radiating seismic waves manifested as earthquakes. These two complementary behaviors accommodate the long-term plate convergence of major subduction zones and are attributed to fault frictional properties. It is conventionally assumed that zones capable of seismic rupture on the subduction megathrust are confined to between about 10 to 50 km depth; however, the actual spatiotemporal distribution of fault mechanical parameters remains elusive for most subduction zones. The 2011 Tohoku Mw 9.0 earthquake ruptured with > 50 m slip up to the trench, thus challenging this conventional assumption, and provides a unique opportunity to probe the mechanical properties of the Japan subduction zone. Drawing on the inferred distribution of coseismic and postseismic slip, it has recently been suggested that portions of the megathrust are capable of switching between seismic and aseismic behavior. Kinematic models of the coseismic rupture and 15-month postseismic afterslip of this event suggest that the coseismic rupture triggered widespread frictional afterslip with equivalent moment magnitude of 8.17–8.53, in addition to viscoelastic relaxation in the underlying mantle. The identified linear relation between modeled afterslip, slip inferred from repeating earthquakes on the plate interface, and the cumulative number of aftershocks within 15 km distance of the subduction thrust suggests that most aftershocks are a direct result of afterslip. We constrain heterogeneous rate-state friction parameters of the subduction thrust from the computed coseismic stress changes and afterslip response. Our results indicate a variable pattern along dip and strike, characterizing areas down-dip and south of the main rupture zone as having velocity-strengthening properties. In agreement with seismic tomographic models of plate boundary elastic properties and geologic evidence for previous M > 8.5 megathrust earthquakes on this section of the plate boundary, we suggest that the obtained pattern of the frictional properties is characteristic of subducted material and thus persistent in time and space.

[1] In this study, we investigate the extent to which viscoelastic velocity perturbations (or “ghost transients”) from individual fault segments can affect elastic block model-based inferences of fault slip rates from GPS velocity fields. We focus on the southern California GPS velocity field, exploring the effects of known, large earthquakes for two end-member rheological structures. Our approach is to compute, at each GPS site, the velocity perturbation relative to a cycle average for earthquake cycles on particular fault segments. We then correct the SCEC CMM4.0 velocity field for this perturbation and invert the corrected field for fault slip rates. We find that if asthenosphere viscosities are low (3 × 1018 Pa s), the current GPS velocity field is significantly perturbed by viscoelastic earthquake cycle effects associated with the San Andreas Fault segment that last ruptured in 1857 (Mw = 7.9). Correcting the GPS velocity field for this perturbation (or “ghost transient”) adds about 5 mm/a to the SAF slip rate along the Mojave and San Bernardino segments. The GPS velocity perturbations due to large earthquakes on the Garlock Fault (most recently, events in the early 1600s) and the White Wolf Fault (most recently, the Mw = 7.3 1952 Kern County earthquake) are smaller and do not influence block-model inverted fault slip rates. This suggests that either the large discrepancy between geodetic and geologic slip rates for the Garlock Fault is not due to a ghost transient or that un-modeled transients from recent Mojave earthquakes may influence the GPS velocity field.

Synthetic seismicity simulations have been explored by the Southern California Earthquake Center (SCEC) Earthquake Simulators Group in order to guide long‐term forecasting efforts related to the Unified California Earthquake Rupture Forecast (Tullis et al. , 2012a). In this study I describe the viscoelastic earthquake simulator (ViscoSim) of Pollitz, 2009. Recapitulating to a large extent material previously presented by Pollitz (2009, 2011) I describe its implementation of synthetic ruptures and how it differs from other simulators being used by the group. ### Generation of Synthetic Slip Events Model faults are assumed to reside in an elastic crust that overlies a viscoelastic crust and mantle with a Maxwell viscoelastic rheology. The evolving physical variable is the Coulomb failure function, which is a linear combination of the shear stress and normal stress resolved upon an a priori slip direction along the considered fault surface (Simpson and Reasenberg, 1994). Following Ben‐Zion et al. (2003), a static stress threshold σ s determines the initiation of a slip event. During an event, stress on an initially failing patch drops by an amount Δ σ (a stress‐reduction parameter assigned to a given patch) to the arrest stress level σ a (more precisely, the slip is that which would reduce the stress to σ a in the absence of slip on other patches). After a set of slip events so determined, stress is transferred from slipped patches to neighboring patches, which may fail first at the static stress threshold σ s and subsequently (during the same event) at the dynamic stress threshold σ d . It is useful to define a dynamic overshoot coefficient D =Δ σ /( σ s − σ d ), where Δ σ = σ s − σ a (i.e., the reduction in stress on one initially failing patch). The parameter D characterizes how far the stress in each subevent …

SUMMARY In 2001 November a magnitude 7.8 earthquake ruptured a 400 km long portion of the Kunlun fault, northeastern Tibet. In this study, we analyse over five years of post-seismic geodetic data and interpret the observed surface deformation in terms of stress relaxation in the thick Tibetan lower crust. We model GPS time-series (first year) and InSAR line of sight measurements (years two to five) and infer that the most likely mechanism of post-seismic stress relaxation is time-dependent distributed creep of viscoelastic material in the lower crust. Since a single relaxation time is not sufficient to model the observed deformation, viscous flow is modelled by a lower crustal Burgers rheology, which has two material relaxation times. The optimum model has a transient viscosity 9 × 1017 Pa s, steady-state viscosity 1 × 1019 Pa s and a ratio of long term to Maxwell shear modulus of 2:3. This model gives a good fit to GPS stations south of the Kunlun Fault, while displacements at stations north of the fault are over-predicted. We attribute this asymmetry in the GPS residual to lateral heterogeneity in rheological structure across the southern margin of the Qaidam Basin, with thinner crust/higher viscosities beneath the basin than beneath the Tibetan Plateau. Deep afterslip localized in a shear zone beneath the fault rupture gives a reasonable match to the observed InSAR data, but the slip model does not fit the earlier GPS data well. We conclude that while some localized afterslip likely occurred during the early post-seismic phase, the bulk of the observed deformation signal is due to viscous flow in the lower crust. To investigate regional variability in rheological structure, we also analyse post-seismic displacements following the 1997 Manyi earthquake that occurred 250 km west of the Kokoxili rupture. We find that viscoelastic properties are the same as for the Kokoxili area except for the transient viscosity, which is 5 × 1017 Pa s. The viscosities estimated for the Manyi and Kokoxili areas are consistent with constraints obtained from other earthquakes in the northwest and south central parts of the Tibetan Plateau.

The generation of seismicity catalogs on synthetic fault networks holds the promise of providing key inputs into probabilistic seismic-hazard analysis, for example, the coefficient of variation, mean recurrence time as a function of magnitude, the probability of fault-to-fault ruptures, and conditional probabilities for foreshock–mainshock triggering. I employ a seismicity simulator that includes the following ingredients: static stress transfer, viscoelastic relaxation of the lower crust and mantle, and vertical stratification of elastic and viscoelastic material properties. A cascade mechanism combined with a simple Coulomb failure criterion is used to determine the initiation, propagation, and termination of synthetic ruptures. It is employed on a 3D fault network provided by Steve Ward (unpublished data, 2009) for the Southern California Earthquake Center (SCEC) Earthquake Simulators Group. This all-California fault network, initially consisting of 8000 patches, each of ∼12 square kilometers in size, has been rediscretized into patches, each of ∼1 square kilometer in size, in order to simulate the evolution of California seismicity and crustal stress at magnitude M ∼5–8. Resulting synthetic seismicity catalogs spanning 30,000 yr and about one-half million events are evaluated with magnitude-frequency and magnitude-area statistics. For a priori choices of fault-slip rates and mean stress drops, I explore the sensitivity of various constructs on input parameters, particularly mantle viscosity. Slip maps obtained for the southern San Andreas fault show that the ability of segment boundaries to inhibit slip across the boundaries (e.g., to prevent multisegment ruptures) is systematically affected by mantle viscosity.

The 26 December 2004 Sumatra earthquake produced static offsets at continuously operating GPS stations at distances of up to 4500 kilometers from the epicenter. We used these displacements to model the earthquake and include consideration of the Earth's shape and depth-varying rigidity. The results imply that the average slip was >5 meters along the full length of the rupture, including the ∼650-kilometer-long Andaman segment. Comparison of the source derived from the far-field static offsets with seismically derived estimates suggests that 25 to 35% of the total moment release occurred at periods greater than 1 hour. Taking into consideration the strong dip dependence of moment estimates, the magnitude of the earthquake did not exceed M w = 9.2.

Earthquake data from the Scotia Arc to early 2002 are reviewed in the light of satellite gravity and other data in order to derive a model for the motion of plates in the Scotia Sea region. Events with magnitude greater than or equal to5, which occurred on or near the boundaries of the Scotia and Sandwich plates, and for which Centroid Moment Tensor (CMT) solutions are available, are examined. The newer data fill some of the previous sampling gaps along the boundaries of the Scotia and Sandwich plates, and provide tighter constraints on relative motions. Variations in the width of the Brunhes anomaly on evenly spaced marine magnetic profiles over the East Scotia Ridge provide new estimates of Scotia-Sandwich plate spreading rates. Since there are no stable fracture zones in the east Scotia Sea, the mean azimuth of sea floor fabric mapped by sidescan is used to constrain the direction of spreading. 18 newrate estimates and four azimuths from the East Scotia Ridge are combined with 68 selected earthquake slip vectors from the boundaries of the Scotia Sea in a least-squares inversion for the best-fitting set of Euler poles and angular rotation rates describing the 'present-day' motions of the Scotia and Sandwich plates relative to South America and Antarctica. Our preferred model (TLP2003) gives poles that are similar to previous estimates, except for Scotia Plate motion with respect to South America, which is significantly different from earlier estimates; predicted rates of motion also differ slightly. Our results are much more robust than earlier work. We examine the implications of the model for motion and deformation along the various plate boundaries, with particular reference to the North and South Scotia Ridges, where rates are obtained by closure.

The M 7.3 June 28, 1992 Landers and M 7.1 October 16, 1999 Hector Mine earthquakes, California, both right lateral strike-slip events on NNW-trending subvertical faults, occurred in close proximity in space and time in a region where recurrence times for surface-rupturing earthquakes are thousands of years. This sug- gests a causal role for the Landers earthquake in triggering the Hector Mine earth- quake. Previous modeling of the static stress change associated with the Landers earthquake shows that the area of peak Hector Mine slip lies where the Coulomb failure stress promoting right-lateral strike-slip failure was high, but the nucleation point of the Hector Mine rupture was neutrally to weakly promoted, depending on the assumed coefficient of friction. Possible explanations that could account for the 7-year delay between the two ruptures include background tectonic stressing, dissi- pation of fluid pressure gradients, rate- and state-dependent friction effects, and post- Landers viscoelastic relaxation of the lower crust and upper mantle. By employing a viscoelastic model calibrated by geodetic data collected during the time period between the Landers and Hector Mine events, we calculate that postseismic relaxa- tion produced a transient increase in Coulomb failure stress of about 0.7 bars on the impending Hector Mine rupture surface. The increase is greatest over the broad surface that includes the 1999 nucleation point and the site of peak slip further north. Since stress changes of magnitude greater than or equal to 0.1 bar are associated with documented causal fault interactions elsewhere, viscoelastic relaxation likely con- tributed to the triggering of the Hector Mine earthquake. This interpretation relies on the assumption that the faults occupying the central Mojave Desert (i.e., both the Landers and Hector Mine rupturing faults) were critically stressed just prior to the Landers earthquake.

I examine the behavior of a two-dimensional (2-D) strike-slip fault system embedded in a 1-D elastic layer (schizosphere) overlying a uniform viscoelastic half-space (plastosphere) and within the boundaries of a finite width shear zone. The viscoelastic coupling model of Savage and Prescott (1978) considers the viscoelastic response of this system, in the absence of the shear zone boundaries, to an earthquake occurring within the upper elastic layer, steady slip beneath a prescribed depth, and the superposition of the responses of multiple earthquakes with characteristic slip occurring at regular intervals. So formulated, the viscoelastic coupling model predicts that sufficiently long after initiation of the system, (1) average fault-parallel velocity at any point is the average slip rate of that side of the fault and (2) far-field velocities equal the same constant rate. Because of the sensitivity to the mechanical properties of the schizosphere-plastosphere system (i.e., elastic layer thickness, plastosphere viscosity), this model has been used to infer such properties from measurements of interseismic velocity. Such inferences exploit the predicted behavior at a known time within the earthquake cycle. By modifying the viscoelastic coupling model to satisfy the additional constraint that the absolute velocity at prescribed shear zone boundaries is constant, I find that even though the time-averaged behavior remains the same, the spatiotemporal pattern of surface deformation (particularly its temporal variation within an earthquake cycle) is markedly different from that predicted by the conventional viscoelastic coupling model. These differences are magnified as plastosphere viscosity is reduced or as the recurrence interval of periodic earthquakes is lengthened. Application to the interseismic velocity field along the Mojave section of the San Andreas fault suggests that the region behaves mechanically like a -600-km-wide shear zone accommodating 50 mrn/yr fault-parallel motion distributed between the San Andreas fault system and Eastern California Shear Zone. In the viscoelastic coupling model (Savage and Prescott, 1978), time-dependent deformation around a strike-slip fault zone is realized by loading, subsequent rupture, and relaxation of a fault present in a schizosphere (elastic upper crust) overlying a ductile plastosphere. Assuming a two-dimensional (2-D) geometry, periodic rupture of the fault, and identical slip, interseismic deformation is obtained as the superposition of viscoelastic relaxation of the plastosphere summed over the infinity of past periodic earthquakes. This model yields far- field interseismic motions equal to + 1/2 of the average slip rate v0 imposed on the fault. Thus, even when no external mechanism is invoked to load the fault (i.e., when the locked portion of the fault penetrates the entire schizosphere), the model satisfies the expectation of rigid plate-like behavior in the far field. In addition, when periodic coseismic offsets are factored in, the model predicts that the time-averaged fault- parallel velocity of each point is +vol2 (depending on location).

We have calculated shear strain rates in the northeastern part of the North Island of New Zealand, using geodetic data collected in the mid-1920s, and 1995. In addition we use observations from a part of the network across Raukumara Peninsula measured in 1976, to look for temporal variation in shear strain rates in that area. The strain rates were estimated using a simultaneous reduction method, assuming uniform strain in space and time within each subnetwork. We find the largest spatial variation in shear strain rates from the mid-1920s to 1995 occurs across Raukumara Peninsula, increasing from 0.06 ± 0.04 ppm/yr in the west to 0.18 ± 0.03 ppm/yr in the east (1σ confidence). The orientation of the azimuth of relative extension shows significant temporal variation across Raukumara Peninsula. From the mid-1920s to 1976 the maximum shear strain rate was 0.22 ± 0.05 ppm/yr with the azimuth of relative extension oriented at 129° ± 7°. From 1976 to 1995 the maximum shear strain rate was 0.14 ± 0.05 ppm/yr with the azimuth of relative extension oriented at 13° ± 12°. We show that a model where the subduction interface is locked down to 15 or 30 km depth, and slipping at NUVEL-IA plate rates below that depth, can explain the strain rates and orientation observed during 1976–1995. To fit the observations from the mid-1920s to 1976, we suggest that additional dip slip occurred on the shallow part of the subduction zone, above 30 km. From this we conclude that the plate boundary below Raukumara Peninsula is locked down to 15 or 30 km depth and the aseismic shallow slip episode that occurred during the mid-1920s–1976 was an isolated event, possibly induced by M ≥ 7 historical earthquakes in the area.

Global Positioning System (GPS) data from campaigns carried out over the 5 years following the 1989 Loma Prieta earthquake and leveling data measured in 1990 and 1992 define the postseismic velocity field around the Loma Prieta rupture zone. Subtraction of a background velocity field yields a residual velocity pattern which we interpret as the product of two physical processes: (1) slow afterslip along distinct planes in the upper cru, st and (2) viscoelastic relaxation of the lower crust and upper mantle. Biirgmann et al. (1997) previously derived an afterslip model involving uniform afterslip on two optimally determined planes, in- cluding oblique reverse slip on the coseismic rupture and reverse slip on a shallow thrust fault to the northeast of the San Andreas fault. We further consider models of distributed slip on these two fault planes plus a viscoelastic relaxation pattern which depends on a suitable cose- ismic rupture model and crust and mantle viscosities. Several fault models from the literature were considered for the 1989 coseismic rupture, with nearly identical impact on the results. Simultaneous maximum likelihood inversion of the GPS and leveling data for afterslip distri- bution and viscosity yields the following results: (1) A good fit to the data is obtained by smooth afterslip distributions without any viscoelastic relaxation being required. (2) Tangible broad-scale viscoelastic relaxation of the lower crust and upper mantle are present in this data set at 97% confidence, and a lower crustal viscosity of- 1029 Pa s is obtained; however, the viscosity of both the lower crust and upper mantle are poorly constrained. (3) For a given misfit, 20% less integrated afterslip and smoother afterslip distributions result when viscoelas- tic relaxation is included. (4) Maximum slip rates on the slip-distributed models are 3-5 crn/yr, the dominant patches estimated on the two planes fill in the entire depth range 4-13 km without significant overlap, and deeper afterslip is not required. The afterslip distribution on the coseismic rupture plane is strongly dominated by reverse slip immediately southeast of the main center of coseismic reverse slip.

The 1995 Kobe earthquake produced predominantly right-lateral slip along the northeast-southwest-trending Nojima fault approximately 200 km landward of the Nankai trough, southwest Japan. The last major earthquakes along the Nankai trough, the M8.0 1944 Tonanakai and M8.2 1946 Nankaido earthquakes, involved about 4 m of interplate slip, each releasing about 100 years of accumulated stress from the northwest-southeast convergence of the Philippine Sea (PHS) plate. By calculating the change in Coulomb failure stress on the Nojima fault due to 1944/46 to 1995 postseismic relaxation of a thin low-viscosity channel beneath southwest Japan, we show that stress diffusion from the 1940s interplate earthquakes increased the likelihood of future failure on the Nojima fault by slowly but progressively unloading the normal stress on the fault over a 50-yr period.

SUMMARY We propose the existence of a major right-lateral transform fault which we call the Boso transform fault. It is related to the Sagami trough, a portion of the Philippine Sea plate boundary south of the Kanto area (central Japan). This Boso transform fault is the result of shear partitioning due to oblique subduction and has delimited a Boso sliver for 2 Myr. The rate of motion is estimated at 16 mm yr-' and the total offset at 30 km. The fault cuts through the Miura and Boso peninsulas onland, where it has a multiple surface expression roughly along the limit of a steeply dipping Miocene ophiolitic body. These subaerial faults have been identified as active, and their cumulated rate of slip across the Miura peninsula can be estimated to be greater than 12 mm yr-', in reasonable agreement with the above estimate. We propose that the slip on the Boso transform fault was responsible for two large (M = 7.0 and 7.5) aftershocks which occurred on the second day after the 1923 great Kanto earthquake. This explains the unusual duration of the aftershock sequence, and the large magnitudes of some of the aftershocks.

SUMMARY A method for calculating the static displacement field following earthquake faulting in a layered spherical earth is presented. At shallow levels, the Earth’s layering is characterized by sharp jumps in bulk and shear moduli at the Conrad discontinuity and the Moho and is therefore important to consider when evaluating crustal deformation. The solution to the equations of static equilibrium is represented as a superposition of spheroidal and toroidal components that each depend on spherical harmonic degree and the moment tensor. A method that has recently been applied to the problem of wave propagation on a layered spherical earth is here applied to the static deformation field. By representing the point source in terms of discontinuities in the displacement-stress vector, the Green’s function for a particular source geometry is derived directly. Numerical tests are presented to verify the accuracy of the method and to illustrate the effects of sphericity and layering on the calculated deformation fields. The effect of sphericity is generally less than about 2 per cent (of maximum deformation) within 100 km of an earthquake source at crustal depths. Comparisons between the deformation calculated on a spherical homogeneous earth and spherical layered earth show that up to 20 per cent errors would be introduced if the Earth’s layered structure were ignored. The effect of layering is strongest for sources with a strong horizontal slip component.

The faulting associated with the 1891 Nobi earthquake, central Honshu, is re-examined on the basis of horizontal displacement vectors determined between the periods 1882/1899-1950/1961. These vectors are modelled in terms of two components: (1) strain accumulation and release along the Nankai Trough and Suruga Trough, and (2) postseismic relaxation following the 1891 earthquake. We assume a rheological model of the central Honshu crust and upper mantle consisting of a 33-km thick elastic lithosphere overlying a 130 km thick viscoelastic sublithosphere. The modelling demonstrates that the observed displacement vectors are dominated by post-seismic displacements following the Nobi earthquake. The Nukumi-Neodani faults are inferred to have accommodated 7-8 m of left lateral strike-slip displacement, in good agreement with those estimates based on the observed surface rupture. In addition, we identify two northeast-southwest trending blind thrust faults, together comprising a 70-km-long region of compressional deformation, each of which accommodated several meters of reverse slip during the Nobi earthquake. These thrust faults consist of a 20-km-long western segment and a 50-km-long eastern segment. The western thrust begins near the southern termination of the strike-slip faulting component, has shallow (31°) dip, is buried by 15 km and has a prominent coseismic signal in the leveling data. Examination of the topography east of Nagoya suggests that the eastern thrust is buried 5-10 km below the surface and represents the subterranean continuation of the exposed thrust fault on the eastern flank of the Kiso Range to the northeast. The model of strain accumulation and release includes the various effects of the 1854 Ansei and 1944 Tonankai earthquakes, as well as the effect of steady state Philippine Sea interplate convergence. A model consisting of only tectonic strain accumulation and release along the Philippine Sea interplate boundary provides a poor fit to the data. The modelling suggests that the 1944 rupture did not extend beyond the Atsumi Peninsula. The ob-served displacement vectors and an independent set of Suruga Bay uplift data contain a moderate signal due to 1854 postseismic displacements. Considering that this signal has tradeoffs with the slip accumulation rate in Suruga Bay, this enables us to constrain the central Honshu sublithospheric viscosity to lie near 2×1019 Pa·s.

[1] Static offsets produced by the February 27, 2010 Mw = 8.8 Maule, Chile earthquake as measured by GPS and InSAR constrain coseismic slip along a section of the Andean megathrust of dimensions 650 km (in length) × 180 km (in width). GPS data have been collected from both campaign and continuous sites sampling both the near-field and far field. ALOS/PALSAR data from several ascending and descending tracks constrain the near-field crustal deformation. Inversions of the geodetic data for distributed slip on the megathrust reveal a pronounced slip maximum of order 15 m at ∼15–25 km depth on the megathrust offshore Lloca, indicating that seismic slip was greatest north of the epicenter of the bilaterally propagating rupture. A secondary slip maximum appears at depth ∼25 km on the megathrust just west of Concepcion. Coseismic slip is negligible below 35 km depth. Estimates of the seismic moment based on different datasets and modeling approaches vary from 1.8 to 2.6 × 1022 N m. Our study is the first to model the static displacement field using a layered spherical Earth model, allowing us to incorporate both near-field and far-field static displacements in a consistent manner. The obtained seismic moment of 1.97 × 1022 N m, corresponding to a moment magnitude of 8.8, is similar to that obtained by previous seismic and geodetic inversions.

[1] The three-dimensional crustal displacement field as sampled by GPS is used to determine the coseismic slip of the 2011 M9.0 Tohoku Earthquake. We employ a spherically layered Earth structure and use a combination of onland GPS, out to ∼4000 km from the rupture, and offshore GPS, which samples the high-slip region on the interplate boundary along the Japan trench. Inversion of the displacement field for dip slip, assuming an interplate boundary of variable dip and striking 195°, yields a compact slip maximum of about 33 m located 200 km east of Sendai. The geodetic moment is 4.06 × 1022 N m, corresponding to Mw = 9.0. The area of maximum slip is concentrated at a depth of about 10 km, is updip of the rupture areas of the M ≳ 7 Miyagi-oki earthquakes of 1933, 1936, 1937, and 1978, and roughly coincides with the rupture area of the M7.1 1981 Miyagi-oki earthquake. The overlap of the 2011 slip area with several preceding ruptures suggests that the same asperities may rupture repeatedly with M ≳ 7 events within several decades of one another.

SUMMARY We apply a viscoelastic cycle model to a compilation of GPS velocity fields in order to address the kinematics of deformation in the northwestern United States. A viscoelastic cycle model accounts for time-dependent deformation following large crustal earthquakes and is an alternative to block models for explaining the interseismic crustal velocity field. Building on the approach taken in Pollitz et al., we construct a deformation model for the entire western United States—based on combined fault slip and distributed deformation—and focus on the implications for the Mendocino triple junction (MTJ), Cascadia megathrust, and western Washington. We find significant partitioning between strike-slip and dip-slip motion near the MTJ as the tectonic environment shifts from northwest-directed shear along the San Andreas fault system to east–west convergence along the Juan de Fuca Plate. By better accounting for the budget of aseismic and seismic slip along the Cascadia subduction interface in conjunction with an assumed rheology, we revise a previous model of slip for the M∼ 9 1700 Cascadia earthquake. In western Washington, we infer slip rates on a number of strike-slip and dip-slip faults that accommodate northward convergence of the Oregon Coast block and northwestward convergence of the Juan de Fuca Plate. Lateral variations in first order mechanical properties (e.g. mantle viscosity, vertically averaged rigidity) explain, to a large extent, crustal strain that cannot be rationalized with cyclic deformation on a laterally homogeneous viscoelastic structure. Our analysis also shows that present crustal deformation measurements, particularly with the addition of the Plate Boundary Observatory, can constrain such lateral variations.

[1] We construct a viscoelastic cycle model of plate boundary deformation that includes the effect of time-dependent interseismic strain accumulation, coseismic strain release, and viscoelastic relaxation of the substrate beneath the seismogenic crust. For a given fault system, time-averaged stress changes at any point (not on a fault) are constrained to zero; that is, kinematic consistency is enforced for the fault system. The dates of last rupture, mean recurrence times, and the slip distributions of the (assumed) repeating ruptures are key inputs into the viscoelastic cycle model. This simple formulation allows construction of stress evolution at all points in the plate boundary zone for purposes of probabilistic seismic hazard analysis (PSHA). Stress evolution is combined with a Coulomb failure stress threshold at representative points on the fault segments to estimate the times of their respective future ruptures. In our PSHA we consider uncertainties in a four-dimensional parameter space: the rupture peridocities, slip distributions, time of last earthquake (for prehistoric ruptures) and Coulomb failure stress thresholds. We apply this methodology to the San Francisco Bay region using a recently determined fault chronology of area faults. Assuming single-segment rupture scenarios, we find that future rupture probabilities of area faults in the coming decades are the highest for the southern Hayward, Rodgers Creek, and northern Calaveras faults. This conclusion is qualitatively similar to that of Working Group on California Earthquake Probabilities, but the probabilities derived here are significantly higher. Given that fault rupture probabilities are highly model-dependent, no single model should be used to assess to time-dependent rupture probabilities. We suggest that several models, including the present one, be used in a comprehensive PSHA methodology, as was done by Working Group on California Earthquake Probabilities.

[1] We report Gravity Recovery and Climate Experiment (GRACE) satellite observations of coseismic displacements and postseismic transients from the great Sumatra-Andaman Islands (thrust event; Mw ∼9.2) earthquake in December 2004. Instead of using global spherical harmonic solutions of monthly gravity fields, we estimated the gravity changes directly using intersatellite range-rate data with regionally concentrated spherical Slepian basis functions every 15-day interval. We found significant step-like (coseismic) and exponential-like (postseismic) behavior in the time series of estimated coefficients (from May 2003 to April 2007) for the spherical Slepian functions. After deriving coseismic slip estimates from seismic and geodetic data that spanned different time intervals, we estimated and evaluated postseismic relaxation mechanisms with alternate asthenosphere viscosity models. The large spatial coverage and uniform accuracy of our GRACE solution enabled us to clearly delineate a postseismic transient signal in the first 2 years of postearthquake GRACE data. Our preferred interpretation of the long-wavelength components of the postseismic gravity change is biviscous viscoelastic flow. We estimated a transient viscosity of 5 × 1017 Pa s and a steady state viscosity of 5 × 1018–1019 Pa s. Additional years of the GRACE observations should provide improved steady state viscosity estimates. In contrast to our interpretation of coseismic gravity change, the prominent postearthquake positive gravity change around the Nicobar Islands is accounted for by seafloor uplift with less postseismic perturbation in intrinsic density in the region surrounding the earthquake.

[1] Coseismic slip associated with the M7.9, 1923 Kanto earthquake is fairly well understood, involving slip of up to 8 m along the Philippine Sea–Honshu interplate boundary under Sagami Bay and its onland extension. Postseismic deformation after the 1923 earthquake, however, is relatively poorly understood. We revisit the available deformation data in order to constrain possible mechanisms of postseismic deformation and to examine the consequences for associated stress changes in the surrounding crust. Data from two leveling lines and one tide gage station over the first 7–8 years postseismic period are of much greater amplitude than the corresponding expected interseismic deformation during the same period, making these data suitable for isolating the signal from postseismic deformation. We consider both viscoelastic models of asthenosphere relaxation and afterslip models. A distributed coseismic slip model presented by Pollitz et al. (2005), combined with prescribed parameters of a viscoelastic Earth model, yields predicted postseismic deformation that agrees with observed deformation on mainland Honshu from Tokyo to the Izu peninsula. Elsewhere (southern Miura peninsula; Boso peninsula), the considered viscoelastic models fail to predict observed deformation, and a model of � 1 m shallow afterslip in the offshore region south of the Boso peninsula, with equivalent moment magnitude Mw = 7.0, adequately accounts for the observed deformation. Using the distributed coseismic slip model, layered viscoelastic structure, and a model of interseismic strain accumulation, we evaluate the post-1923 stress evolution, including both the coseismic and accumulated postseismic stress changes and those stresses contributed by interseismic loading. We find that if account is made for the varying tectonic regime in the region, the occurrence of both immediate (first month) post-1923 crustal aftershocks as well as recent regional crustal seismicity is consistent with the predicted stress pattern. This suggests that the influence of the 1923 earthquake on regional seismicity is fairly predictable and has persisted for at least seven decades following the earthquake.

Summary We present a relationship between the long-term fault slip rates and instantaneous velocities as measured by Global Positioning System (GPS) or other geodetic measurements over a short time span. The main elements are the secularly increasing forces imposed by the bounding Pacific and Juan de Fuca (JdF) plates on the North American plate, viscoelastic relaxation following selected large earthquakes occurring on faults that are locked during their respective interseismic periods, and steady slip along creeping portions of faults in the context of a thin-plate system. In detail, the physical model allows separate treatments of faults with known geometry and slip history, faults with incomplete characterization (i.e. fault geometry but not necessarily slip history is available), creeping faults, and dislocation sources distributed between the faults. We model the western United States strain-rate field, derived from 746 GPS velocity vectors, in order to test the importance of the relaxation from historic events and characterize the tectonic forces imposed by the bounding Pacific and JdF plates. Relaxation following major earthquakes (Mγ 8.0) strongly shapes the present strain-rate field over most of the plate boundary zone. Equally important are lateral shear transmitted across the Pacific–North America plate boundary along ∼1000 km of the continental shelf, downdip forces distributed along the Cascadia subduction interface, and distributed slip in the lower lithosphere. Post-earthquake relaxation and tectonic forcing, combined with distributed deep slip, constructively interfere near the western margin of the plate boundary zone, producing locally large strain accumulation along the San Andreas fault (SAF) system. However, they destructively interfere further into the plate interior, resulting in smaller and more variable strain accumulation patterns in the eastern part of the plate boundary zone. Much of the right-lateral strain accumulation along the SAF system is systematically underpredicted by models which account only for relaxation from known large earthquakes. This strongly suggests that in addition to viscoelastic-cycle effects, steady deep slip in the lower lithosphere is needed to explain the observed strain-rate field.

[1] The slip distribution associated with the 1923 M = 7.9 Kanto, Japan, earthquake is reexamined in light of new data and modeling. We utilize a combination of first-order triangulation, second-order triangulation, and leveling data in order to constrain the coseismic deformation. The second-order triangulation data, which have not been utilized in previous studies of 1923 coseismic deformation, are associated with only slightly smaller errors than the first-order triangulation data and expand the available triangulation data set by about a factor of 10. Interpretation of these data in terms of uniform-slip models in a companion study by Nyst et al. shows that a model involving uniform coseismic slip on two distinct rupture planes explains the data very well and matches or exceeds the fit obtained by previous studies, even one which involved distributed slip. Using the geometry of the Nyst et al. two-plane slip model, we perform inversions of the same geodetic data set for distributed slip. Our preferred model of distributed slip on the Philippine Sea plate interface has a moment magnitude of 7.86. We find slip maxima of ∼8–9 m beneath Odawara and ∼7–8 m beneath the Miura peninsula, with a roughly 2:1 ratio of strike-slip to dip-slip motion, in agreement with a previous study. However, the Miura slip maximum is imaged as a more broadly extended feature in our study, with the high-slip region continuing from the Miura peninsula to the southern Boso peninsula region. The second-order triangulation data provide good evidence for ∼3 m right-lateral strike slip on a 35-km-long splay structure occupying the volume between the upper surface of the descending Philippine Sea plate and the southern Boso peninsula.

The M7.9 2002 Denali earthquake, Alaska, is one of the largest strike-slip earthquakes ever recorded. The postseismic GPS velocity field around the 300-km-long rupture is characterized by very rapid horizontal velocity up to ∼300 mm/yr for the first 0.1 years and slower but still elevated horizontal velocity up to ∼100 mm/yr for the succeeding 1.5 years. I find that the spatial and temporal pattern of the displacement field may be explained by a transient mantle rheology. Representing the regional upper mantle as a Burghers body, 1 infer steady state and transient viscosities of η 1 = 2.8 x 10 18 Pa s and η 2 = 1.0 x 10 17 Pa s, respectively, corresponding to material relaxation times of 1.3 and 0.05 years. The lower crustal viscosity is poorly constrained by the considered horizontal velocity field, and the quoted mantle viscosities assume a steady state lower crust viscosity that is 7η 1 . Systematic bias in predicted versus observed velocity vectors with respect to a fixed North America during the first 3-6 months following the earthquake is reduced when all velocity vectors are referred to a fixed site. This suggests that the post-Denali GPS time series for the first 1.63 years are shaped by a combination of a common mode noise source during the first 3-6 months plus viscoelastic relaxation controlled by a transient mantle rheology.

[1] A cluster of exceptionally large earthquakes in the interior of Asia occurred from 1905 to 1967: the 1905 M7.9 Tsetserleg and M8.4 Bolnai earthquakes, the 1931 M8.0 Fu Yun earthquake, the 1957 M8.1 Gobi-Altai earthquake, and the 1967 M7.1 Mogod earthquake (sequence). Each of the larger (M ≥ 8) earthquakes involved strike-slip faulting averaging more than 5 m and rupture lengths of several hundred kilometers. Available geologic data indicate that recurrence intervals on the major source faults are several thousands of years and distances of about 400 km separate the respective rupture areas. We propose that the occurrences of these and many smaller earthquakes are related and controlled to a large extent by stress changes generated by the compounded static deformation of the preceding earthquakes and subsequent viscoelastic relaxation of the lower crust and upper mantle beneath Mongolia. We employ a spherically layered viscoelastic model constrained by the 1994–2002 GPS velocity field in western Mongolia [Vergnolle et al., 2003]. Using the succession of twentieth century earthquakes as sources of deformation, we then analyze the time-dependent change in Coulomb failure stress (Δσf). At remote interaction distances, static Δσf values are small. However, modeled postseismic stress changes typically accumulate to several tenths of a bar over time intervals of decades. Almost all significant twentieth century regional earthquakes (M ≥ 6) with well-constrained fault geometry lie in positive Δσf lobes of magnitude about +0.5 bar. Our results suggest that significant stress transfer is possible among continental faults separated by hundreds of kilometers and on timescales of decades.