Search

The tsunami associated with the giant 9.5 M-w 1960 Chile earthquake deposited an extensive sand layer above organic-rich soils near Queule (39.3 degrees S, 73.2 degrees W), south-central Chile. Using the 1960 tsunami deposits, together with eye-witness observations and numerical simulations of tsunami inundation, we tested the tsunami inundation sensitivity of the site to different earthquake slip distributions. Stratigraphically below the 1960 deposit are two additional widespread sand layers interpreted as tsunami deposits with maximum ages of 4960-4520 and 5930-5740 cal BP. This >4500-year gap of tsunami deposits preserved in the stratigraphic record is inconsistent with written and geological records of large tsunamis in south-central Chile in 1575, 1837, and possibly 1737. We explain this discrepancy by: (1) poor preservation of tsunami deposits due to reduced accommodation space from relative sea-level fall during the late Holocene; (2) recently evolved coastal geomorphology that increased sediment availability for tsunami deposit formation in 1960; and/or (3) the possibility that the 1960 tsunami was significantly larger at this particular location than other tsunamis in the past >4500 years. Our research illustrates the complexities of reconstructing a complete stratigraphic record of past tsunamis from a single site for tsunami hazard assessment. Puget Sound Energy Graduate Fellowship at Central Washington University; Earthquake Hazards Program of the U.S. Geological Survey; Chilean National Fund for Development of Science and Technology (FONDECYT)Comision Nacional de Investigacion Cientifica y Tecnologica (CONICYT)CONICYT FONDECYT [1190258, 1181479]; Millennium Scientific Initiative (ICM) of the Chilean government [NC160025]; National Science Foundation (NSF)National Science Foundation (NSF)National Research Foundation of Korea [EAR-1624533, EAR-1624542]; ANID PIA Anillo [ACT192169] Published version Puget Sound Energy Graduate Fellowship at Central Washington University; Earthquake Hazards Program of the U.S. Geological Survey; Chilean National Fund for Development of Science and Technology (FONDECYT), Grant/Award Numbers: 1190258, 1181479; Millennium Scientific Initiative (ICM) of the Chilean government, Grant/Award Number: NC160025; National Science Foundation (NSF), Grant/Award Numbers: EAR-1624533, EAR-1624542; ANID PIA Anillo, Grant/Award Number: ACT192169 Public domain – authored by a U.S. government employee

We present a comprehensive relative sea-level (RSL) database for north, central, and south-central Chile (18.5 degrees S - 43.6 degrees S) using a consistent, systematic, and internationally comparable approach. Despite its latitudinal extent, this coastline has received little rigorous or systematic attention and details of its RSL history remain largely unexplored. To address this knowledge gap, we re-evaluate the geological context and age of previously published sea-level indicators, providing 78 index points and 84 marine or terrestrial limiting points spanning from 11 ka to the present day. Many data points were originally collected for research in other fields and have not previously been examined for the information they provide on sea-level change. Additionally, we describe new sea-level data from four sites located between the Gulf of Arauco and Valdivia. By compiling RSL histories for 11 different regions, we summarise current knowledge of Chilean RSL. These histories indicate mid Holocene sea levels above present in all regions, but at highly contrasting elevations from similar to 30 m to

On September 16, 2015, a Mw 8.3 earthquake struck the north-central Chile coast, triggering a tsunami observed along 500 km of coastline, between Huasco (28.5°S) and San Antonio (33.5°S). This tsunami provided a unique opportunity to examine the nature of tsunami deposits in a semi-arid, siliciclastic environment where stratigraphic and sedimentological records of past tsunamis are difficult to distinguish. To improve our ability to identify such evidence, we targeted one of the few low-energy, organic-rich depositional environments in north-central Chile: Pachingo marsh in Tongoy Bay (30.3°S). We found sedimentary evidence of the 2015 and one previous tsunami as tabular sand sheets. Both deposits are composed of poorly to moderately sorted, gray-brown, fine-to medium-grained sand and are distinct from underlying and overlying organic-rich silt. Both sand beds thin (from ∼20 cm to

An uncommon coastal sedimentary record combines evidence for seismic shaking and coincident tsunami inundation since AD 1000 in the region of the largest earthquake recorded instrumentally: the giant 1960 southern Chile earthquake (Mw 9.5). The record reveals significant variability in the size and recurrence of megathrust earthquakes and ensuing tsunamis along this part of the Nazca-South American plate boundary. A 500-m long coastal outcrop on Isla Chiloé, midway along the 1960 rupture, provides continuous exposure of soil horizons buried locally by debris-flow diamicts and extensively by tsunami sand sheets. The diamicts flattened plants that yield geologically precise ages to correlate with well-dated evidence elsewhere. The 1960 event was preceded by three earthquakes that probably resembled it in their effects, in AD 898–1128, 1300–1398 and 1575, and by five relatively smaller intervening earthquakes. Earthquakes and tsunamis recurred exceptionally often between AD 1300 and 1575. Their average recurrence interval of 85 years only slightly exceeds the time already elapsed since 1960. This inference is of serious concern because no earthquake has been anticipated in the region so soon after the 1960 event, and current plate locking suggests that some segments of the boundary are already capable of producing large earthquakes. This long-term earthquake and tsunami history of one of the world's most seismically active subduction zones provides an example of variable rupture mode, in which earthquake size and recurrence interval vary from one earthquake to the next.

Great megathrust earthquakes arise from the sudden release of energy accumulated during centuries of interseismic plate convergence. The moment deficit (energy available for future earthquakes) is commonly inferred by integrating the rate of interseismic plate locking over the time since the previous great earthquake. But accurate integration requires knowledge of how interseismic plate locking changes decades after earthquakes, measurements not available for most great earthquakes. Here we reconstruct the post-earthquake history of plate locking at Guafo Island, above the seismogenic zone of the giant 1960 (M-w = 9.5) Chile earthquake, through forward modeling of land-level changes inferred from aerial imagery (since 1974) and measured by GPS (since 1994). We find that interseismic locking increased to similar to 70% in the decade following the 1960 earthquake and then gradually to 100% by 2005. Our findings illustrate the transient evolution of plate locking in Chile, and suggest a similarly complex evolution elsewhere, with implications for the time- and magnitude-dependent probability of future events., Postprints der Universität Potsdam : Mathematisch-Naturwissenschaftliche Reihe, 678

The stratigraphy within coastal river valleys in south-central Chile clarifies and extends the region’s history of large, earthquakes and accompanying tsunamis. Our site at Quidico (38.1°S, 73.3°W) is located in an overlap zone between ruptures of magnitude 8–9 earthquakes in 1960 and 2010, and, therefore, records tsunamis originating from subduction-zone ruptures north and south of the city of Concepción. Hand-dug pits and cores in a 3-m-thick sequence of freshwater peat in an abandoned meander (a little-examined depositional environment for tsunami deposits) and exposures along the Quidico River show five sand beds that extend as much as 1.2 km inland. Evidence for deposition of the beds by tsunamis includes tabular sand beds that are laterally extensive (>100 m), well sorted, fine upward, have sharp lower contacts, and contain diatom assemblages dominated by brackish and marine taxa. Using eyewitness accounts of tsunami inundation,137Cs analyses, and14C dating, we matched the upper four sand beds with historical tsunamis in 2010, 1960, 1835, and 1751. The oldest prehistoric bed dates to 1445–1490 CE and correlates with lacustrine and coastal records of similar-aged earthquakes and tsunamis in south-central Chile.

The ∼500-year historical record of seismicity along the central Chile coast (30–34°S) is characterized by a series of ∼M 8.0–8.5 earthquakes followed by low tsunamis ( 10 m), but the frequency of such large events is unknown. We extend the seismic history of central Chile through a study of a lowland stratigraphic sequence along the metropolitan coast north of Valparaiso (33°S). At this site, higher relative sea level during the mid Holocene created a tidal marsh and the accommodation space necessary for sediment that preserves earthquake and tsunami evidence. Within this 2600-yr-long sequence, we traced six laterally continuous sand beds probably deposited by high tsunamis. Plant remains that underlie the sand beds were radiocarbon dated to 6200, 5600, 5000, 4400, 3800, and 3700 cal yr BP. Sediment properties and diatom assemblages of the sand beds—for example, anomalous marine planktonic diatoms and upward fining of silt-sized diatom valves—point to a marine sediment source and high-energy deposition. Grain-size analysis shows a strong similarity between inferred tsunami deposits and modern coastal sediment. Upward fining sequences characteristic of suspension deposition are present in five of the six sand beds. Despite the lack of significant lithologic changes between the sedimentary units under- and overlying tsunami deposits, we infer that the increase in freshwater siliceous microfossils in overlying units records coseismic uplift concurrent with the deposition of five of the sand beds. During our mid-Holocene window of evidence preservation, the mean recurrence interval of earthquakes and tsunamis is ∼500 years. Our findings imply that the frequency of historical earthquakes in central Chile is not representative of the greatest earthquakes and tsunamis that the central Chilean subduction zone has produced.

The Arauco Peninsula (37°-38°S) in south-central Chile has been proposed as a possible barrier to the along-strike propagation of megathrust ruptures, separating historical earthquakes to the south (1960 AD 1837, 1737, and 1575) and north (2010 AD, 1835, 1751, 1657, and 1570) of the peninsula. However, the 2010 (Mw 8.8) earthquake propagated into the Arauco Peninsula, re-rupturing part of the megathrust that had ruptured only 50 years earlier during the largest subduction zone earthquake in the instrumental record (Mw 9.5). To better understand long-term slip variability in the Arauco Peninsula region, we analyzed four coastal sedimentary sections from two sites (Tirua, 38.3°S and Quidico, 38.1°S) located within the overlap of the 2010 and 1960 ruptures to reconstruct a ∼600-year record of coseismic land-level change and tsunami inundation. Stratigraphic, lithologic, and diatom results show variable coseismic land-level change coincident with tsunami inundation of the Tirua and Quidico marshes that is consistent with regional historical accounts of coseismic subsidence during earthquakes along the Valdivia portion of the subduction zone (1960 AD and 1575) and coseismic uplift during earthquakes along the Maule portion of the subduction zone (2010 AD, 1835, 1751). In addition, we document variable coseismic land-level change associated with three new prehistoric earthquakes and accompanying tsunamis in 1470–1570 AD, 1425–1455, and 270–410. The mixed record of coseismic subsidence and uplift that we document illustrates the variability of down-dip and lateral slip distribution at the overlap of the 2010 and 1960 ruptures, showing that ruptures have repeatedly propagated into, but not through the Arauco Peninsula and suggesting the area has persisted as a long-term impediment to slip through at least seven of the last megathrust earthquakes (∼600 years).

Far-field tsunami records from the Japanese tide gauge network allow the reexamination of the moment magnitudes (Mw) for the 1906 and 1922 Chilean earthquakes, which to date rely on limited information mainly from seismological observations alone. Tide gauges along the Japanese coast provide extensive records of tsunamis triggered by six great (Mw >8) Chilean earthquakes with instrumentally determined moment magnitudes. These tsunami records are used to explore the dependence of tsunami amplitudes in Japan on the parent earthquake magnitude of Chilean origin. Using the resulting regression parameters together with tide gauge amplitudes measured in Japan we estimate apparent moment magnitudes of Mw 8.0–8.2 and Mw 8.5–8.6 for the 1906 central and 1922 north-central Chile earthquakes. The large discrepancy of the 1906 magnitude estimated from the tsunami observed in Japan as compared with those previously determined from seismic waves (Ms 8.4) suggests a deeper than average source with reduced tsunami excitation. A deep dislocation along the Chilean megathrust would favor uplift of the coast rather than beneath the sea, giving rise to a smaller tsunami and producing effects consistent with those observed in 1906. The 1922 magnitude inferred from far-field tsunami amplitudes appear to better explain the large extent of damage and the destructive tsunami that were locally observed following the earthquake than the lower seismic magnitudes (Ms 8.3) that were likely affected by the well-known saturation effects. Thus, a repeat of the large 1922 earthquake poses seismic and tsunami hazards in a region identified as a mature seismic gap.

We demonstrate a direct method for the calculation of the annual frequency of exceedance for earthquake losses (or the probability distribution for annual losses) to a portfolio.This method parallels the classic method of probabilistic seismic hazard analysis for the calculation of the annual frequency of exceedance for earthquake ground motions. The method assumes conditional independence of the random component of ground motions and losses at different sites for each earthquake, given magnitude, distance to the sites, and so-called interevent epsilon. Examples show that the method is realizable, and can take into account different loss functions and site conditions in the portfolio. The main advantage of this method is that it does not require a separate set of scenario earthquakes, as do Monte Carlo-based approaches, but can be calculated directly from the inputs used for hazard maps. DOI: 10.1193/1.3159475

Records of complete earthquake cycles are rare. Analysis of historical nautical charts and modern GPS data that record a full earthquake cycle in Chile show that great earthquakes can create small amounts of permanent uplift. Individual great earthquakes are posited to release the elastic strain energy that has accumulated over centuries by the gradual movement of tectonic plates1,2. However, knowledge of plate deformation during a complete seismic cycle—two successive great earthquakes and the intervening interseismic period—remains incomplete3. A complete seismic cycle began in south-central Chile in 1835 with an earthquake of about magnitude 8.5 (refs 4, 5) and ended in 2010 with a magnitude 8.8 earthquake6. During the first earthquake, an uplift of Isla Santa Maria by 2.4 to 3 m was documented4,5. In the second earthquake, the island was uplifted7 by 1.8 m. Here we use nautical surveys made in 1804, after the earthquake in 1835 and in 1886, together with modern echo sounder surveys and GPS measurements made immediately before and after the 2010 earthquake, to quantify vertical deformation through the complete seismic cycle. We find that in the period between the two earthquakes, Isla Santa Maria subsided by about 1.4 m. We simulate the patterns of vertical deformation with a finite-element model and find that they agree broadly with predictions from elastic rebound theory2. However, comparison with geomorphic and geologic records of millennial coastline emergence8,9 reveal that 10–20% of the vertical uplift could be permanent.

The distributions of insured losses to single-family housing following the 1994 Northridge, California, earthquake for 234 ZIP codes can be satisfactorily modeled with gamma distributions. Regressions of the parameters in the gamma distribution on estimates of ground motion, derived from ShakeMap estimates or from interpolated observations, provide a basis for developing curves of conditional probability of loss given a ground motion. Comparison of the resulting estimates of aggregate loss with the actual aggregate loss gives satisfactory agreement for several different ground-motion parameters. Estimates of loss based on a deterministic spatial model of the earthquake ground motion, using standard attenuation relationships and NEHRP soil factors, give satisfactory results for some groundmotion parameters if the input ground motions are increased about one and one-half standard deviations above the median, reflecting the fact that the ground motions for the Northridge earthquake tended to be higher than the median ground motion for other earthquakes with similar magnitude. The results give promise for making estimates of insured losses to a similar building stock under future earthquake loading. [DOI: 10.1193/1.1775238]

Bayesian inference provides a method to use seismic intensity data or instrumental locations, together with geologic and seismologic data, to make quan- titative estimates of the probabilities that specific past earthquakes are associated with specific faults. Probability density functions are constructed for the location of each earthquake, and these are combined with prior probabilities through Bayes' theorem to estimate the probability that an earthquake is associated with a specific fault. Results using this method are presented here for large, preinstrumental, his- torical earthquakes and for recent earthquakes with instrumental locations in the San Francisco Bay region. The probabilities for individual earthquakes can be summed to construct a probabilistic frequency-magnitude relationship for a fault segment. Other applications of the technique include the estimation of the probability of back- ground earthquakes, that is, earthquakes not associated with known or considered faults, and the estimation of the fraction of the total seismic moment associated with earthquakes less than the characteristic magnitude. Results for the San Francisco Bay region suggest that potentially damaging earthquakes with magnitudes less than the characteristic magnitudes should be expected. Comparisons of earthquake locations and the surface traces of active faults as determined from geologic data show sig- nificant disparities, indicating that a complete understanding of the relationship be- tween earthquakes and faults remains elusive.

A combination of geological and historical records from south-central Chile provides a means to address general questions about the stability of megathrust rupture patches and the range of variation expected among earthquakes and tsunamis along a particular stretch of a subduction zone. The Tirua River estuary (38.3°S) records four large tsunamis and coseismic land-level changes over the past 450 years within the overlapping rupture zones of the great subduction-zone earthquakes of A.D. 1960 (Mw 9.5) and 2010 (Mw 8.8). Sand layers 2 km up the Tirua River represent the 2010 and 1960 tsunamis and two historical tsunamis, most likely in A.D. 1751 and 1575. Differing land-level changes during these earthquakes likely denote differences in the spatial distribution of slip on the megathrust in both the strike and dip directions within the overlapping rupture zone, with the uplift at Tirua in 1751 and 2010 probably caused by slip extending farther landward and to greater depth than in 1575 and 1960, which showed subsidence or little change.

Spatial correlation of annual earthquake ground motions and losses can be used to estimate the variance of annual losses to a portfolio of properties exposed to earthquakes. A direct method is described for the calculation of the spatial correlation of earthquake ground motions and losses. Calculations for the direct method can be carried out using either numerical quadrature or a discrete, matrix-based approach. Numerical results for this method are compared with those calculated from a simple Monte Carlo simulation. Spatial correlation of ground motion and loss is induced by the systematic attenuation of ground motion with distance from the source, by common site conditions, and by the finite length of fault ruptures. Spatial correlation is also strongly dependent on the partitioning of the variability, given an event, into interevent and intraevent components. Intraevent variability reduces the spatial correlation of losses. Interevent variability increases spatial correlation of losses. The higher the spatial correlation, the larger the variance in losses to a portfolio, and the more likely extreme values become. This result underscores the importance of accurately determining the relative magnitudes of intraevent and interevent variability in ground-motion studies, because of the strong impact in estimating earthquake losses to a portfolio. The direct method offers an alternative to simulation for calculating the variance of losses to a portfolio, which may reduce the amount of calculation required.

The seismic hazard and earthquake occurrence rates in Hawaii are locally as high as that near the most hazardous faults elsewhere in the United States. We have generated maps of peak ground acceleration (PGA) and spectral acceleration (SA) (at 0.2, 0.3 and 1.0 sec, 5% critical damping) at 2% and 10% exceedance probabilities in 50 years. The highest hazard is on the south side of Hawaii Island, as indicated by the M I 7.0, M S 7.2, and M I 7.9 earthquakes, which occurred there since 1868. Probabilistic values of horizontal PGA (2% in 50 years) on Hawaii's south coast exceed 1.75 g . Because some large earthquake aftershock zones and the geometry of flank blocks slipping on subhorizontal decollement faults are known, we use a combination of spatially uniform sources in active flank blocks and smoothed seismicity in other areas to model seismicity. Rates of earthquakes are derived from magnitude distributions of the modern (1959–1997) catalog of the Hawaiian Volcano Observatory's seismic network supplemented by the historic (1868–1959) catalog. Modern magnitudes are M L measured on a Wood-Anderson seismograph or M S. Historic magnitudes may add M L measured on a Milne-Shaw or Bosch-Omori seismograph or M I derived from calibrated areas of MM intensities. Active flank areas, which by far account for the highest hazard, are characterized by distributions with b slopes of about 1.0 below M 5.0 and about 0.6 above M 5.0. The kinked distribution means that large earthquake rates would be grossly underestimated by extrapolating small earthquake rates, and that longer catalogs are essential for estimating or verifying the rates of large earthquakes. Flank earthquakes thus follow a semicharacteristic model, which is a combination of background seismicity and an excess number of large earthquakes. Flank earthquakes are geometrically confined to rupture zones on the volcano flanks by barriers such as rift zones and the seaward edge of the volcano, which may be expressed by a magnitude distribution similar to that including characteristic earthquakes. The island chain northwest of Hawaii Island is seismically and volcanically much less active. We model its seismic hazard with a combination of a linearly decaying ramp fit to the cataloged seismicity and spatially smoothed seismicity with a smoothing half-width of 10 km. We use a combination of up to four attenuation relations for each map because for either PGA or SA, there is no single relation that represents ground motion for all distance and magnitude ranges. Great slumps and landslides visible on the ocean floor correspond to catastrophes with effective energy magnitudes M E above 8.0. A crude estimate of their frequency suggests that the probabilistic earthquake hazard is at least an order of magnitude higher for flank earthquakes than that from submarine slumps.

The U.S. Geological Survey (USGS) recently completed new probabilistic seismic hazard maps for the United States, including Alaska and Hawaii. These hazard maps form the basis of the probabilistic component of the design maps used in the 1997 edition of the NEHRP Recommended Provisions for Seismic Regulations for New Buildings and Other Structures, prepared by the Building Seismic Safety Council and published by FEMA. The hazard maps depict peak horizontal ground acceleration and spectral response at 0.2, 0.3, and 1.0 sec periods, with 10%, 5%, and 2% probabilities of exceedance in 50 years, corresponding to return times of about 500, 1000, and 2500 years, respectively. In this paper we outline the methodology used to construct the hazard maps. There are three basic components to the maps. First, we use spatially smoothed historic seismicity as one portion of the hazard calculation. In this model, we apply the general observation that moderate and large earthquakes tend to occur near areas of previous small or moderate events, with some notable exceptions. Second, we consider large background source zones based on broad geologic criteria to quantify hazard in areas with little or no historic seismicity, but with the potential for generating large events. Third, we include the hazard from specific fault sources. We use about 450 faults in the western United States (WUS) and derive recurrence times from either geologic slip rates or the dating of pre-historic earthquakes from trenching of faults or other paleoseismic methods. Recurrence estimates for large earthquakes in New Madrid and Charleston, South Carolina, were taken from recent paleoliquefaction studies. We used logic trees to incorporate different seismicity models, fault recurrence models, Cascadia great earthquake scenarios, and ground-motion attenuation relations. We present disaggregation plots showing the contribution to hazard at four cities from potential earthquakes with various magnitudes and distances.

We present a summary of the data and analyses leading to the revision of the time-independent probabilistic seismic hazard maps of Alaska and the Aleutians. These maps represent a revision of existing maps based on newly obtained data, and reflect best current judgments about methodology and approach. They have been prepared following the procedures and assumptions made in the preparation of the 2002 National Seismic Hazard Maps for the lower 48 States, and will be proposed for adoption in future revisions to the International Building Code. We present example maps for peak ground acceleration, 0.2 s spectral amplitude (SA), and 1.0 s SA at a probability level of 2% in 50 years (annual probability of 0.000404). In this summary, we emphasize issues encountered in preparation of the maps that motivate or require future investigation and research.

We report on a time-dependent seismic hazard analysis for Alaska and the Aleutians to complement our recently completed time-independent map. Whereas the time-independent map treats all sources as statistically independent, the time-dependent analysis is based on calculations of the conditional probability of occurrence for the next 50 years by using a Brownian Passage Time model for the seismic sources judged to be characteristic. We then consider how those probabilities are modified by coseismic and postseismic stress changes resulting from large regional earthquakes occurring from 1938 to 2002. Recombining the time-dependent probabilities with time-independent truncated Gutenberg-Richter and smoothed seismicity sources leads to our time-dependent probabilistic seismic hazard results. We find that when accounting for time dependence without stress changes, earthquake probabilities can be significantly altered, reducing probabilities to near zero or increasing them to several times the time-independent values. In addition, accounting for coseismic stress changes tends to have a local influence on earthquake probabilities, whereas postseismic effects can be far-reaching in both time and space. In sum, however, since we combine time-dependent and time-independent sources, the modification to seismic hazard is relatively minor, increasing or decreasing hazard adjacent to characteristic faults by about 10%. Most cities, located far from characteristic faults, are little affected.

Aftershocks of the 1971 San Fernando and 1994 Northridge earthquakes were relocated using a three-dimensional velocity model that was derived from inverting P-wave travel time data. The hypocenters show clear orientations of the dipping fault planes. The San Fernando aftershocks form a plane extending from a depth of 15 km to the surface, dipping toward the northeast at about 40°. The Northridge aftershocks delineate a fault extending from a depth of 18 km up to about 5 km, dipping toward the southwest at about 40°. In the region the aftershocks overlap in map view, the San Fernando plane cuts off the Northridge plane at a depth of 5 to 8 km, preventing it from reaching the surface. The similar but oppositely dipping fault planes suggest a pair of conjugate planes reflecting a horizontal northeast-southwest compression.

Splay faults are thrusts that emerge from the plate boundaries of subduction zones. Such structures have been mapped at several convergent margins and their activity commonly ascribed to large megathrust earthquakes. However, the behavior of splay faults during the earthquake cycle is poorly constrained because typically these structures are located offshore and are difficult to access. Here we use geologic mapping combined with space and land geodesy, as well as offshore sonar data, to document surface-fault ruptures and coastal uplift at Isla Santa Maria in south-central Chile (37 degrees S) caused by the 27 February 2010 Maule earthquake (M-w 8.8). During the earthquake, the island was tilted parallel to the margin, and normal faults ruptured the surface and adjacent ocean bottom. We associate tilt and crestal normal faulting with growth of an anticline above a blind reverse fault rooted in the Nazca-South America plate boundary, which slipped during the Maule earthquake. The splay fault system has formed in an area of reduced coseismic plate-boundary slip, suggesting that anelastic deformation in the upper plate may have restrained the 2010 megathrust rupture. Surface fault breaks were accompanied by prominent discharge of fluids. Our field observations support the notion that splay faulting may frequently complement and influence the rupture of subduction-zone earthquakes.

Earthquakes can be triggered by any significant perturbation of the hydrologic regime. In areas where potentially active faults are already close to failure, the increased pore pressure resulting from fluid injection, or, alternatively, the massive extraction of fluid or gas, can induce sufficient stress and/or strain changes that, with time, can lead to sudden catastrophic failure in a major earthquake. Injection-induced earthquakes typically result from the reduction in frictional strength along preexisting, nearby faults caused by the increased formation fluid pressure. Earthquakes associated with production appear to respond to more complex mechanisms of subsidence, crustal unloading, and poroelastic changes in response to applied strains induced by the massive withdrawal of subsurface material. As each of these different types of triggered events can occur up to several years after well activities have begun (or even several years after all well activities have stopped), this suggests that the actual triggering process may be a very complex combination of effects, particularly if both fluid extraction and injection have taken place locally. To date, more than thirty cases of earthquakes triggered by well activities can be documented throughout the United States and Canada. Based on these case histories, it is evident that, owing to preexisting stress conditions in the upper crust, certain areas tend to have higher probabilities of exhibiting such induced seismicity.

[1] Spatially averaged, absolute deviatoric stress tensors along the faults ruptured during the 2002 Denali fault earthquake, both before and after the event, are derived, using a new method, from estimates of the orientations of the principal stresses and the stress change associated with the earthquake. Stresses are estimated in three regions along the Denali fault, one of which also includes the Susitna Glacier fault, and one region along the Totschunda fault. Estimates of the spatially averaged shear stress before the earthquake resolved onto the faults that ruptured during the event range from near 1 MPa to near 4 MPa. Shear stresses estimated along the faults in all these regions after the event are near zero (0 ± 1 MPa). These results suggest that deviatoric stresses averaged over a few tens of km along strike are low, and that the stress drop during the earthquake was complete or nearly so.

Vitaly Khalturin, a pioneering seismologist with encyclopedic knowledge of regional seismic signals and a mentor and friend of seismologists around the world, died on 17 April 2007 in Stanford, California, four days after suffering a major stroke.⇓ ![][1] Vitaly Khalturin (second from left) in 1976, working at home with graduate students in Garm, Tajikistan. Photo by T. Rautian. Born on 15 August 1927 in Leningrad, U.S.S.R., Vitaly grew up in war-torn and politically tumultuous surroundings. His father was a writer well-known in Russian literary circles and, from childhood on, Vitaly knew many prominent Russian authors. While still a student at the University of Leningrad he married a fellow student, Tat'yana Glebovna Rautian. After receiving their masters' degrees in 1951, the two moved to Central Asia as part of the Complex Seismological Expedition (CSE) of the U.S.S.R.'s Academy of Sciences' Institute of Earth Physics, where they would work together for more than 40 years. Vitaly and Tat'yana spent their early years in Garm, Tajikistan, building CSE's network of seismographic stations. Far from any libraries or centers of science, they were forced to learn for themselves—from first principles in classical physics—how to design experiments and interpret seismograms. Under harsh economic circumstances, they installed and maintained an extended seismic network. They taught generations of students—from Russia, Armenia, Georgia, and each of the Central Asian republics—the rigors and joys of earthquake studies. It is remarkable that years later the quality of their work has turned out to be so informative to research groups operating with fewer obstacles and far greater … [1]: /embed/graphic-1.gif

Mean apparent focal depths of microearthquakes occurring along a 20-km stretch of the San Andreas fault southeast of Hollister, California, increased by 25 percent some 60 days before the magnitude 4.6 Stone Canyon earthquake of September 4, 1972. The shape of the time-depth anomaly is virtually identical to the time plots of Vp/Vs preceding moderate earthquakes at Garm, U.S.S.R. A less well-defined depth anomaly occurred in October-December 1971, preceding the Limekiln Road swarm in December 1971 and the magnitude 5.0 Melendy Ranch earthquake in February 1972. The observed depth anomalies can be attributed to dilatancy biasing of hypocenters, although true vertical migration of seismicity cannot be ruled out.

Preliminary measurements of the stress orientation at a depth of 2 km in the Cajon Pass Scientific Drillhole, about 3.6 km from the San Andreas fault in southern California, indicate that the local direction of maximum compression is nearly normal to the fault zone. The measurements have been interpreted to indicate that the regional orientation of the maximum compression is normal to the fault, and taken as evidence for a very weak fault. The orientation of the regional maximum compression expected from plate tectonic arguments is about 66° NE from the strike of the fault. Geodetic data indicate that the orientation of maximum compressive strain rate is about 43° NE from the strike of the fault, and show nearly pure right-lateral shear acting parallel to the fault. These apparent conflicts in the inferred orientation of the axis of maximum compression may be explained in part by a model in which the fault zone is locked over a depth interval in the range of 2-5 to 15 km, but is very weak above and below that interval. An analytic solution for this model predicts orientations of the principal stresses that are nearly parallel or normal to the fault, for observations made close to the fault and above and below the locked zone, but an orientation consistent with plate tectonic estimates at depths corresponding to that of the locked interval. This solution does require, however, a few mm/yr of creep at the surface on the San Andreas or nearby sub-parallel faults (such as the San Jacinto), which has not yet been observed, or a shallow zone near the faults of distributed deformation.

Earthquake prediction is at the cutting edge of basic science. Substantial research programs are under way in China, Japan and the U.S.S.R. In the U.S., meanwhile, some $17 million is being spent each year on prediction and another $43 million on a research program including work on earthquake-hazard assessment, the engineering of buildings to resist earthquakes and studies of the fundamental nature of earthquakes. Progress on short-term prediction has been slower than workers hoped it would be a decade ago, when early successes were reported by both Soviet and American researchers. On the other hand, progress in long-term prediction and in the understanding of how earthquakes occur has been substantial. Now more than ever the aim is to learn to interpret the signals of strain accumulating in the earth within the framework of fundamental understanding of the earthquake process. This article reviews progress to date.

Quantitative calculations for the effect of a fault creep event on observations of changes in water level in wells provide an approach to the tectonic interpretation of these phenomena. For the pore pressure field associated with an idealized creep event having an exponential displacement versus time curve, an analytic expression has been obtained in terms of exponential-integral functions. The pore pressure versus time curves for observation points near the fault are pulselike; a sharp pressure increase (or decrease, depending on the direction of propagation) is followed by more gradual decay to the normal level after the creep event. The time function of the water level change may be obtained by applying the filter—derived by A. G. Johnson and others to determine the influence of atmospheric pressure on water level—to the analytic pore pressure versus time curves. The resulting water level curves show a fairly rapid increase (or decrease) and then a very gradual return to normal. The results of this analytic model do not reproduce the steplike changes in water level observed by Johnson and others. If the procedure used to obtain the water level from the pore pressure is correct, these results suggest that steplike changes in water level are not produced by smoothly propagating creep events but by creep events that propagate discontinously, by changes in the bulk properties of the region around the well, or by some other mechanism. In addition, simplistic calculations show that significant pressure field variations and water level changes near the surface may be expected to accompany a propagating creep event on a buried fault. Water level changes of as much as several meters may be expected at the surface for a creep event having a dislocation amplitude of l m on a semi-infinite dislocation surface extending downward from a depth of 10 km and propagating horizontally at a rate of 10 km/day. The maximum near-surface effect should be observed at a horizontal distance from the fault about equal to the depth of the top of the dislocation surface. These results are consistent with the observations made in China of large water level changes preceding large earthquakes, if some sort of aseismic creep event at depth precedes these earthquakes.

The relatively high seismicity of the San Andreas fault zone in central California provides an excellent opportunity to search for seismic forerunners to moderate earthquakes. Analysis of seismic traveltime and earthquake location data has resulted in the identification of two possible seismic forerunners. The first is a period of apparently late (0.3 sec) P-wave arrival times lasting several weeks preceding one earthquake of magnitude 5.0. The rays for these travel paths passed through — or very close to — the aftershock volume of the subsequent earthquake. The sources for these P-arrival time data were earthquakes in the distance range 20–70 km. Uncertainties in the influence of small changes in the hypocenters of the source earthquakes and in the identification of small P-arrivals raise the possibility that the apparantly delayed arrivals are not the result of a decrease in P-velocity. The second possible precursor is an apparent increase in the average depth of earthquakes preceding two moderate earthquakes. This change might be only apparent, caused by a location bias introduced by a decrease in P-wave velocity, but numerical modeling for realistic possible changes in velocity suggests that the observed effect is more likely a true migration of earthquakes. To carry out this work — involving the manipulation of several thousand earthquake hypocenters and several hundred thousand readings of arrival time — a system of data storage was designed and manipulation programs for a large digital computer have been executed. This system allows, for example, the automatic selection of earthquakes from a specific region, the extraction of all the observed arrival times for these events, and their relocation under a chosen set of assumptions.

An analytic solution is not available for the diffraction of elastic waves by wedges; however, numerical solutions of finite-difference type are available for selected wedge angles. The P - to Rayleigh-wave conversion coefficients at wedge tips have been measured on two-dimensional seismic models for stress-free wedges with wedge angles, θ 0 , of 10, 30, 60, 90 and 120°. The conversion coefficients show two broad peaks and a minimum as a function of the angle between the wedge face and the direction of the incident P -wave. The minimum occurs for the P wave incident parallel to the wedge face and one maximum is near an incidence angle of 90° to the wedge face. The amplitude of this maximum, relative to the other, decreases as the wedge angle increases. The asymmetry of the conversion coefficients, C PR ( θ ; θ 0 ), relative to parallel incidence ( θ = 0) increases as the wedge angle increases. The locations of the maxima and the minimum as well as the asymmetry can be explained qualitatively. The conversion coefficients are measured with an accuracy of ±5% in those regions where there are no interfering waves. A comparison of the data for the 10° wedge with the theoretical results for a half plane (0° wedge) shows good correlation.

Intermediate-term observations preceding earthquakes of magnitude 5.7 or greater in California from 1975 through 1986 suggest that: (1) The sudden appearance of earthquakes in a previously inactive area indicates an increased likelihood of a significant earthquake in that area for a period from days to years; (2) these larger earthquakes tend to occur towards the ends of creeping fault segments; (3) one large earthquake in a region increases the likelihood of a subsequent significant event in the adjacent area; and (4) marginal evidence for the occurrence of a regional deformation event suggests that such events increase the probability of earthquake occurrence throughout the entire area. A common element in many of these observed patterns appears to be the transmission and amplification of tectonic stress changes by the mechanism of fault creep, and suggests that surface fault creep is a sensitive indicator of changes in stress. The preceding critieria are used to construct a preliminary ‘forecast’ of the likely locations of significant earthquakes over the next decade.

Regional seismicity data have been examined to determine the nature of seismicity in the years preceding these moderate earthquakes in California: 1952 Kern County (M = 7.7), 1963 Watsonville (M = 5.4), 1964 and 1967 Corralitos (M = 5.0, 5.3), 1966 Parkfield-Cholame (M= 5.1, 5.5), 1968 Borrego Mountain (M = 6.4), 1969 Santa Rosa (M = 5.6, 5.7), and 1971 San Fernando (M = 6.4). In each instance the moderate earthquakes occurred in areas characterized by a relatively high level of small earthquake activity. In most cases the preceding activity in the area immediately surrounding the epicenter was high relative to other segments of the same fault zone or other nearby fault zones that, on the basis of geologic criteria, would be considered equally likely locations for moderate earthquakes. Detailed observations for a moderate earthquake that occurred in 1972 near Bear Valley (M = 5.0), where seismograph station coverage is adequate to obtain reliable focal depths for small earthquakes, indicate that concentrations of small earthquakes were present in the immediate hypocentral region in the months before the earthquake. The regional observations have potential applications for the analysis of seismic risk in that fault zones or segments of fault zones characterized by a relatively large number of small earthquakes seem more likely to sustain moderate earthquakes than do adjacent fault zones or fault segments with equivalent geologic evidence for recency of movement. The detailed observations provide a potential tool for mapping stress concentrations or areas of anomalously low strength along the parts of fault zones capable of seismic slip.

Residuals for P-wave traveltimes at a seismnograph station near Bear Valley, California, for small, precisely located local earthquakes at distances of 20 to 70 kilometers show a sharp increase of nearly 0.3 second about 2 months before a magnitude 5.0 earthquake that occurred within a few kilometers of the station. This indicates that velocity changes observed elsewhere premonitory to earthquakes, possibly related to dilatancy, occur along the central section of the San Andreas fault system.

Observations of the water-level changes in a well drilled into the San Andreas fault zone have been under way since May 1971, with the objective of studying in situ pore-pressure changes in a zone of active tectonic creep and seismicity. Small water-level changes, characterized by a decrease and subsequent rise, have been followed by earthquakes of moderate size on the San Andreas fault zone. Compatibility of these observations with either a dilatancy-type behavior or a dislocation-type behavior for the pre-earthquake process can be demonstrated. Additional water-level observations at other sites in the fault zone are needed to examine the spatial and temporal extent of the actual preseismic process that is responsible for observed creep events, water-level changes, and tilt changes.

Observations along the surface traces of active faults in California and elsewhere indicate that tectonic displacement can occur as seismic slip or as aseismic fault creep. Fault creep occurs as secular displacement and as displacement episodes, called “creep events,” which last a few hours to days and include from a millimeter or less to a few tens of millimeters of displacement. Instrumental measurements of displacement versus time during creep events show that many events have a characteristic, simple form, including a steep beginning, followed by a gradual decay in the rate of displacement. Some creep events display small amounts of precursory slip, and many seem to be composed of several discrete “simple” events. Propagation of creep events along the San Andreas, Calaveras, and Imperial faults in California has been inferred from the nearly simultaneous observation of creep events at nearby creepmeters and from kinematic modeling of continuous measurements of strain during creep events. Fault creep can be viewed as arising from the interplay of three factors: stress applied to the fault from external sources; stress caused by the geometry and distribution of displacement on the fault, arising from the elastic response of the surrounding medium to the displacements within the anelastic fault zone itself; and the constitutive relations characterizing the resistance to slip on the fault. Analytic solutions are presented for a simple, rectangular dislocation, quasi-static model of nonpropagating fault creep assuming viscous and quasi-plastic (power law creep) fault zone rheology and for a multi-element (composed of several rectangular or striplike dislocation surfaces), quasi-static model assuming a viscous rheology. Viscous rheology implies solutions composed of exponentials that decay with time. The simple, one-element, quasi-plastic model is in agreement with the form of creep events observed at Melendy Ranch, California. A matrix formulation is presented to calculate dislocation stresses on a fault zone in a half-space or plate. The formulation involves the division of the dislocation surface into strips or rectangles and the use of published solutions to calculate the stress resulting from displacement on each individual element. Results from the matrix method agree with analytic results for equilibrium displacements on “stress-free” cracks. The matrix formulation is then used as the basis for a numerical method to simulate one- and two-dimensional propagating creep assuming a quasi-plastic rheology on the fault zone. Results from this method agree reasonably well with observations of afterslip from the 1966 Parkfield, 1975 Oroville, and 1979 Imperial Valley, California, earthquakes and of a propagating creep event observed along the southern Calaveras fault, central California, during July 1977. This method of analysis, and the assumption that creep at each point on the fault zone follows a power law rheology after the applied stress reaches a spatially varying yield stress, can explain the secular and episodic nature of fault creep; afterslip; the simple shape of some individual events and the multiple, composite shape of others; and the propagation of creep events. On the southern Calaveras fault, observations of propagating creep indicate a yield stress near zero at the surface and throughout much of the fault zone. However, they also indicate the presence of a zone at a depth of about 0.5 km, and about 5 km long, with a yield stress of about 15 bars (1.5 MPa). The methods of solution presented here should be applicable to many problems in the dynamics of faulting.