Search

International audience

SUMMARYIn spite of growing evidence that many earthquakes are preceded by increased seismic activity, the nature of this activity is still poorly understood. Is it the result of a mostly random process related to the natural tendency of seismic events to cluster in time and space, in which case there is little hope to ever predict earthquakes? Or is it the sign that a physical process that will lead to the impending rupture has begun, in which case we should attempt to identify this process. With this aim we take a further look at the nucleation of two of the best recorded and documented strike-slip earthquakes to date, the 1999 Izmit and Düzce earthquakes which ruptured the North Anatolian Fault over ∼200 km. We show the existence of a remarkable mechanical logic linking together nucleation characteristics, stress loading, fault geometry and rupture speed. In both earthquakes the observations point to slow aseismic slip occurring near the ductile-to-brittle transition zone as the motor of their nucleation.

An increase of both shallow and intraslab intermediate-depth seismicity has been observed days to years before some great subduction earthquakes, as before Tohoku-oki (Mw 9.0, 2011), Maule (Mw 8.8, 2010) or Iquique (Mw 8.2, 2014) earthquakes (Bouchon et al., 2016, Jara et al,. 2017). These observations suggest that a link exists between these deep and shallow foreshocks, but it is still poorly understood and not characterized in a systematic manner. Some studies have attempted to address this lack of systematic characterization by using a statistical approach (Delbridge et al., 2017).The aim of this study is to systematically and statistically identify and characterize the potential correlations between deep and shallow seismicity. We want to assess whether or not such interactions exist. If they exist, we plan to characterize when and where they occur, at what frequency, their characteristic duration, and with what spatial pattern. For this purpose, we develop a statistical method to assess the relevance of deep-shallow interactions, that allows to identify statistically significant correlations between deep and shallow seismicity. We focused on the seismicity of the Japan trench subduction zone during the decade prior to the Tohoku-oki earthquake, because deep-shallow interactions were identified there, and because we can test the events picked by our method against the correlations highlighted in published papers (Bouchon et al., 2016). The correlation values between the deep and shallow events from the Japan Meteorological Agency catalog are calculated on various different sliding-windows with durations from month to week. These correlation values are then compared to the ones obtained using synthetic series of shallow events that meet the spectral properties of the real series, and the significance of the correlation is calculated.Some windows show a strong correlation. The dependence of our results to different parameters, such as the completeness magnitude, the length of the window, the lag, the smoothing etc… are evaluated. The spatio-temporal analysis of the seismicity on maps for these windows is also explored. While the results are still preliminary, we believe that this method has the potential to systematically and quantitatively assess the current presumptions on the link between deep and shallow seismicity, that would lead to a better understanding of the mechanisms leading to megathrust earthquakes. Bouchon, M., Marsan, D., Durand, V., Campillo, M., Perfettini, H., Madariaga, R., & Gardonio, B. (2016). Potential slab deformation and plunge prior to the Tohoku, Iquique and Maule earthquakes. Nature Geoscience, 9(5), 380.Delbridge, B. G., Kita, S., Uchida, N., Johnson, C. W., Matsuzawa, T., & Bürgmann, R. (2017). Temporal variation of intermediate‐depth earthquakes around the time of the M9. 0 Tohoku‐oki earthquake. Geophysical Research Letters, 44(8), 3580-3590.Jara, J., Socquet, A., Marsan, D., & Bouchon, M. (2017). Long-Term Interactions Between Intermediate Depth and Shallow Seismicity in North Chile Subduction Zone. Geophysical Research Letters, 44(18), 9283-9292.

International audience; The observation of a transient slip 1 month before the rupture of the 2011 Tohoku earthquake is a conandrum since the area was supposedly fully coupled. A better understanding of the mechanisms at work during the preseismic phase is thus fundamental. However, the configuration of the Pacific plate and the location of the Tohoku rupture zone 200 km from the coast make it difficult to detect microseismic events. In this study, we use a multidimensional template matching (MDTM) technique to detect earthquakes that are hidden in the noise. The temporal distribution of these 395 newly detected earthquakes provides new insights on the slip history of the megathrust earthquake epicentral zone. The detected events can be separated into two groups: 187 low-frequency detections (below 5 Hz) that well recorded the episodes of earthquake migration prior to the Tohoku earthquake and 208 high-frequency detections (above 10 Hz) that occurred close to the rupture zones of the M ≥ 4.8-6 earthquakes that struck between the 9 March 2011 M7.3 foreshock and the 30 November 2010 Tohoku-Oki earthquake. The seismic rate of these high frequency detection events starts to increase on 30 November 2010 until the Tohoku earthquake.

It has long been observed that periods of intense seismic activity in a region alternate with periods of relative quiescence, but establishing whether this is the expected result of purely random f...

A magnitude 7.5 earthquake hit the city of Palu in Sulawesi, Indonesia on 28 September 2018 at 10:02:43 (coordinated universal time). It was followed a few minutes later by a 4–7-m-high tsunami. Palu is situated in a narrow pull-apart basin surrounded by high mountains of up to 2,000 m altitude. This morphology has been created by a releasing bend in the Palu-Koro fault, a rapidly moving left-lateral strike-slip fault. Here we present observations derived from optical and radar satellite imagery that constrain the ground surface displacements associated with the earthquake in great detail. Mapping of the main rupture and associated secondary structures shows that the slip initiated on a structurally complex and previously unknown fault to the north, extended southwards over 180 km and passed through two major releasing bends. The 30 km section of the rupture south of Palu city is extremely linear, and slightly offset from the mapped geological fault at the surface. This part of the rupture accommodates a large and smooth surface slip of 4–7 m, with no shallow slip deficit. Almost no aftershock seismicity was recorded from this section of the fault. As these characteristics are similar to those from known supershear segments, we conclude that the Palu earthquake probably ruptured this segment at supershear velocities. The devastating 2018 magnitude 7.5 earthquake in Palu, Indonesia, ruptured at supershear speeds according to evidence from space geodesy.

The precursory activity leading up to the Tohoku-Oki earthquake of 2011 has been suggested to feature both long- and short-term episodes of decoupling and suggests a particularly complex slow slip history. The analysis of the F3 solution of the Japanese GPS network suggested that an accelerated slip occurred in the deeper part of the seismogenic zone during the 10 years preceding the earthquake (Heki & Mitsui, EPSL 2013; Mavrommatis et al., GRL 2014; Yokota & Koketsu, Nat. Com. 2015). During the two months preceding the earthquake, no anomaly in the GPS position time series has been revealed so far, although several anomalous geophysical signals have been reported (an extended foreshock crisis near the future hypocenter (Kato et al., Science 2012), a synchronized increase of intermediate-depth background seismicity (Bouchon et al., Nat Geosc. 2016), a signal in ocean-bottom pressure gauges and on-land strainmeter time series (Ito et al., Tectonoph. 2013), and large scale gravity anomalies that suggest deep-seated slab deformation processes (Panet et al., Nat. Geosc. 2018 ; Wang & Burgmann, GRL 2019)).We present novel results based on an independent analysis of the Japanese GPS data set. We perform a full reprocessing of the raw data with a double-difference approach, a systematic analysis of the obtained time-series, including noise characterization and network filtering, and make a robust assessment of long- and short-term tectonic aseismic transients preceding the Tohoku-Oki earthquake. An accelerated slip on the lower part of the seismogenic zone over the last decade is confirmed, not only below the epicenter of Tohoku-Oki earthquake but also further south, offshore Boso peninsula, which is a worrying sign of an on-going slow decoupling east of Tokyo. At shorter time-scale, first results seem compatible with a slow slip close to the epicenter initiating ~ 2 months before the mainshock.

A massive template-matching approach is successfully applied in Marmara Sea region along the North Anatolian Fault, during the 2009-2014 period to enrich the description of the time and space evolution of the seismicity. Detection of events are performed on the continuous data recorded from 2009 to 2014 combining two types of catalogs as templates: a finely constructed catalog for the three first year (2009-2011) (Schmittbuhl et al, 2016) and a raw catalog from KOERI for the last three years (2012-2014). Magnitudes (Ml) are estimated for all detected events using relative amplitudes of the highly coherent waveforms between new events and template events. The template database provides a nearly threefold increase of the number of small events (more than 15000 earthquakes compare to the 4673 events of the initial catalog). Combined with a double-difference relocation based on cross-correlation differential travel-time data, the database is shown to be a relevant framework for the long term monitoring of specific remanent structures like seismic swarms or repeating earthquakes. The obtained catalog confirms the strong contrast of behaviors along the Main Marmara Fault (MMF): deep creeping to the west (Central Basin), fully locked in the center (Kumburgaz Basin) and dominated by fluid and off-fault activity to the east (Cinarcik Basin).

We document interactions between intermediate depth and interplate seismicity in the North Chile subduction zone, over a 25-year period (1990 - 2015). We show that the 2005 Mw 7.8 Tarapaca slab-pull earthquake was followed by 9 years of enhanced deep and shallow seismicity, together with the decrease of eastward average GPS velocities and associated interplate coupling, eventually leading to the 2014 Mw 8.1 Iquique megathrust earthquake. In contrast, megathrust ruptures (e.g. Mw 8.0 Antofagasta in 1995, or Mw 8.1 Iquique in 2014) initiate several years of silent background seismicity in the studied area, both at shallow and intermediate depths. The plunge of a rigid slab into a viscous asthenospheric mantle could explain the observed synchronization between deep and shallow seismicity, and their long-term interactions.

We investigate the temporal evolution of the background seismicity rate related to the subduction of the Pacific plate in northeast Japan, at latitudes 34o to 42o, for the 1/1/1990 - 9/3/2011 period. Two declustering methods are used to identify robust features. We find that the dominant behaviour is a lowering down of activity, especially in the northern half of our studied area, where changes appear related to the cycle of M7.5+ earthquakes, in particular the 1968 Tokachi and the 1994 Sanriku earthquakes. Acceleration of background seismicity is observed offshore the Kanto region, and could mark a long term decoupling of the Pacific and the Philippine Sea plates over a 100 km-long segment along the Sagimi Trough, which cannot be simply explained by the earthquake cycle model. Our analysis further suggests that changes in background seismicity are plausibly related to changes in seismic coupling, and thus further strengthens the recent observation that seismic coupling does vary at the time scales of tens of years, for the Japanese subduction zone.

International audience; The 30 May 2015 M w 7.9 Bonin Islands earthquake, Japan, is one of the largest deep-focus earthquakes ever recorded. Its occurrence, close to 700-km depth, in an area without any known historical seismicity, along with its magnitude, was a surprise to scientists. Deep earthquakes are generally believed to have few aftershocks and no foreshocks. Here, we explore the earthquake productivity in the hypocentral surroundings and detect 49 not previously identified earthquakes, 28 of which occurred during an accelerating preseismic phase that started 3 months prior to the main shock. This is the first time that such foreshock activity has been observed for a deep earthquake. The preseismic and postseismic activity suggests transformational faulting within a metastable olivine wedge (MOW) inside the slab at depth as the triggering principal mechanism for this deep earthquake sequence, the seismicity starting where the backward bending of the subducting Pacific plate is maximum. Plain Language Summary Deep earthquakes have been puzzling seismologists for almost a century, since their discovery in the 1920s. Earthquakes deeper than 50 km represent about 25% of the global seismicity, and deep-focus earthquakes are defined as those located at 300-km depth or more. They can be exceedingly large and occur at temperatures and pressures where sliding and fracture are inhibited, thus the brittle fracture/friction mechanism, valid in Earth's crust, cannot hold. Despite their abundance, the physical mechanism behind deep-focus earthquakes is still a subject of ongoing debate. The Bonin Islands earthquake (30 May 2015, M w 7.9) occurred at 680-km depth in a previously quiet area. In order to better understand why and how it occurred there, we searched for previously unknown earthquakes in the surroundings. For the first time, we identify a preseismic phase preceding such a very deep-focus earthquake. The location of the newly detected earthquakes reveals the geometry of the slab at depth, and their spatiotemporal distribution supports transformation of metastable olivine as the rupture initiation process.

The Main Marmara Fault (MMF) which crosses the whole Marmara Sea is a significant seismic gap along the North Anatolian Fault. Here we show that nine long-lasting strike-slip seismic repeaters exist below the Central basin within the seismogenic zone, in a 10 km region where deep creep was previously suggested from the analysis of the local seismicity. The typical recurrence time is 8 months during the 2008-2015 period. The cumulative slip of the repeating sequence appears to be compatible with the regional geodetic slip rate if they are assumed to be part of a large single asperity (10 km). The repeaters also exhibit short term crises and are possibly related to bursts of creep.

The seismicity along the Main Marmara Fault (MMF) below the Marmara Sea is analyzed during the 2007–2012 period to provide insights on the recent evolution of this important regional seismic gap. High precision locations show that seismicity is strongly varying along strike and depth providing fine details of the fault behavior that are inaccessible from geodetic observations. The activity strongly clusters at the regions of transition between basins. The Central basin shows significant seismicity located below the shallow locking depth inferred from GPS measurements. Its b-value is low and the average seismic slip is high. All observations are consistent with a deep creep of this segment. On the contrary, the Kumburgaz basin at the center of the fault shows sparse seismicity with the hallmarks of a locked segment. In the eastern Marmara Sea, the seismicity distribution along the Princes Island segment in the Cinarcik basin, is consistent with the geodetic locking depth of 10 km and a low contribution to the regional seismic energy release. The assessment of the locked segment areas provide an estimate of the magnitude of the main forthcoming event to be about 7.3 assuming that the rupture will not enter significantly within creeping domains.

Seismic activity has increased in the Kanto region, Japan, following the 2011 M9.0 Tohoku earthquake. We here reassess this increase up to June 2014, to show that normal, Omori-like relaxation characterizes the activity on crustal faults as well as on the Philippine Sea plate, but not on the deeper Pacific plate. There repeating earthquakes display a twofold rate of occurrence (still ongoing in June 2014) as compared to the pre-Tohoku rate, suggesting enhanced creep. We compute the Coulomb stress changes on the upper locked portion of the Philippine Sea plate, which last ruptured in 1923. We find that this fault was little affected by either the coseismic, the postseismic, the accelerated creep, or the 2011 Boso silent slip event.

Between 2010 and 2015 three giant earthquakes occurred in the Chilean subduction where the oceanic Nazca plate plunges under South America. These were the largest events there since the gigantic M9.5 1960 earthquake so their close occurrences raise the question of a possible link between them. We show here that two-and-a-half days after the M8.2 Iquique earthquake, seismic activity started to increase downdip below (depth similar to 100km) the future Illapel epicenter. This increase, which began with the largest intermediate-depth earthquake in the Chilean subduction after Iquique, lasted until the M8.3 Illapel earthquake, 18months later. The mechanisms involved suggest that the Iquique earthquake started a tear in the slab directly downdip from the future epicenter. This study relies on seismicity which occurs in the cold core of the slab and which is the only direct information we have on processes occurring at these depths. The results support that giant earthquakes interact at the scale of a subducting plate and suggest that this interaction occurs through the deep slab.

Foreshock activity sometimes precedes the occurrence of large earthquakes, but the nature of this seismicity is still debated, and whether it marks transient deformation and/or slip nucleation is still unclear. We here study at the worldwide scale how foreshock occurrence affects the postseismic phase and find a significant positive correlation between foreshock and aftershock activities: earthquakes preceded by accelerating seismicity rates produce 40% more aftershocks on average, and the length of the aftershock zone after 20 days is 20% larger. These observations cannot be reproduced by standard earthquake clustering models that predict the accelerating pattern of foreshock occurrence but not its impact on aftershock activity. This strongly suggests that slow deformation transients, possibly related to episodic creep, could initiate prior to the main shock and extend past the coseismic phase, resulting in compound ruptures that include a very long period (up to tens of days) component.

SUMMARY Slip-rate function and the rupture velocity are two important parameters that are critical in understanding the physics of earthquakes. When conventional objective functions are used, the slip-rate function is not well resolved from seismic data. Here, we propose a new method to obtain the slip-rate function by utilizing the near-field phases recorded near the fault rupture. First we illustrate the sensitivity of near-field phases to the moment accumulation and modify the objective function in order to take advantage of this sensitivity. By utilizing near-field P waves along with S pulses on the near-source records and using a Bayesian approach, we show that we can constrain the average slip-rate function as well as the average rupture velocity for a strike-slip earthquake. As a case example, we apply this technique to the record of the 2003, Mw6.6 Bam Earthquake. Our results indicate an asymmetric slip-rate function, with acceleration duration of about 0.4s, and deceleration duration of 1.4s. The slip-rate function obtained from kinematic modelling of the 2003 Bam earthquake is consistent with those predicted by dynamic rupture simulations. The rupture velocity is about 82–90 per cent of the shear wave velocity, implying a sub-Rayleigh rupture velocity close to the Rayleigh wave speed. In future cases where abundant near-source strong-motion data exist and slip is well constrained, the method described in this study can be applied to obtain the variation of the slip-rate function along the fault which would improve our understanding of earthquake rupture physics.

[1] We investigate the effect of dynamic and static stress changes produced by the 1999 Izmit earthquake, on four preexisting seismic clusters located in the eastern Marmara sea, beyond the western termination of the earthquake rupture. These four clusters show long-lasting modifications in their seismicity rate. We observe that these seismic activity variations are related to stress changes. Dynamic stress pulses activate strike-slip faulting instantaneously, but in the absence of a concomitant static Coulomb stress increase, this activation is short lived. Indeed, a large dynamic stress combined with a negative static Coulomb stress may result in an immediate activation followed by the occurrence of a seismicity shadow. In contrast, the activation of extensional clusters begins slowly and takes a few days to fully develop. It is also remarkably long lasting and does not follow a classical Omori decay. More than 10 years after the earthquake, the extensional clusters located near the termination of the rupture, where static stress and pressure changes were high, are still activated.

We propose that a controlling parameter of static stress drop during an earthquake is related to the scaling properties of the fault-surface topography. Using high resolution laser distance meters, we have accurately measured the roughness scaling properties of two fault surfaces in different geological settings (the French Alps and Nevada). The data show that fault-surface topography is scale dependent and may be accurately described by a self-affine geometry with a slight anisotropy characterized by two extreme roughness exponents ( H R ), H ||=0.6 in the direction of slip and H ⊥=0.8 perpendicular to slip. Disregarding plastic processes like rock fragmentation and focusing on elastic deformation of the topography, which is the dominant mode at large scales, the stress drop is proportional to the deformation, which is a spatial derivative of the slip. The evolution of stress-drop fluctuations on the fault plane can be derived directly from the self-affine property of the fault surface, with the length scale ( λ ) as std Δσ ( λ )∝ λ H R -1. Assuming no characteristic length scale in fault roughness and a rupture cascade model, we show that as the rupture grows, the average stress drop, and its variability should decrease with increasing source dimension. That is for the average stress drop Δσ ( r )∝ r H R -1, where r is the radius of a circular rupture. This result is a direct consequence of the elastic squeeze of fault asperities that induces the largest spatial fluctuations of the shear strength before and after the earthquake at local (small) scales with peculiar spatial correlations.

SUMMARY We present analysis of the spatial correlations of seismological slip maps and fault topography roughness, illuminating their identical self-affine exponent. Though the complexity of the coseismic spatial slip distribution can be intuitively associated with geometrical or stress heterogeneities along the fault surface, this has never been demonstrated. Based on new measurements of fault surface topography and on statistical analyses of kinematic inversions of slip maps, we propose a model, which quantitatively characterizes the link between slip distribution and fault surface roughness. Our approach can be divided into two complementary steps: (i) Using a numerical computation, we estimate the influence of fault roughness on the frictional strength (pre-stress). We model a fault as a rough interface where elastic asperities are squeezed. The Hurst exponent Hτ, characterizing the self-affinity of the frictional strength field, approaches Hτ = H// −1, where H// is the roughness exponent of the fault surface in the direction of slip. (ii) Using a quasi-static model of fault propagation, which includes the effect of long-range elastic interactions and spatial correlations in the frictional strength, the spatial slip correlation is observed to scale as Hs = Hτ + 1, where Hs represents the Hurst exponent of the slip distribution. Under the assumption that the origin of the spatial fluctuations in frictional strength along faults is the elastic squeeze of fault asperities, we show that self-affine geometrical properties of fault surface roughness control slip correlations and that Hs = H//. Given that H// = 0.6 for a wide range of faults (various accumulated displacement, host rock and slip movement), we predict that Hs = 0.6. Even if our quasi-static fault model is more relevant for creeping faults, the spatial slip correlations observed are consistent with those of seismological slip maps. A consequence is that the self-affinity property of slip roughness may be explained by fault geometry without considering dynamical effects produced during an earthquake.

Megathrust earthquakes rupture the shallow plate boundary in subduction zones. Analysis of seismic activity preceding megathrust quakes in Japan and Chile reveals deep seismicity that may mark plunging of the slabs prior to main fault rupture. Megathrust earthquakes rupture hundreds of kilometres of the shallow plate interface in subduction zones, typically at depths of less than 50 km. Intense foreshock activity preceded the 2011 Mw 9 Tohoku-oki (Japan) and 2014 Mw 8.2 Iquique (Chile) megathrust earthquakes. This pre-earthquake activity was thought to be generated1,2,3,4,5,6 by slow slip in the seismogenic zone before rupture, but where this slow slip originated and how it spread rapidly over long distances are unknown. Here we analyse seismic activity deep in the subduction zone before the Tohoku-oki and Iquique ruptures, as well as before the 2010 Mw 8.8 Maule earthquake in Chile. We find that, before each of these megathrust earthquakes, shallow seismicity occurred synchronously with bursts of seismic activity deep (∼100 km) in the subducting slab. The extensional mechanism of these deep shocks suggests that the slab was stretched at depth. We therefore propose that, before these megathrust quakes, the slab might have started to plunge into the mantle below part of the future rupture zone. We speculate that synchronization between deep and shallow seismicity may have marked the nucleation phase for these three giant earthquakes.

International audience; We present observations of ground accelerations recorded at a small array close to the fault during the Düzce earthquake and its early aftershocks. The records show the strong spatial variability of ground acceleration over distances of only a few hundred metres. During the main shock, the peak horizontal acceleration values ranged from 0.3 to about 1.0 g at stations distant of 1.5 km only. We attribute this spatial variability to a fault zone site effect as peak ground acceleration steadily increases as the distance to the fault trace decreases. The spectral ratio between the ground motion recorded near the fault and the one outside the fault zone shows a shift of the spectral peak to lower frequencies with increasing peak accelerations. Such an observation suggests a non-linear behaviour of the fault zone due to the strong ground shaking. As much as a 45 per cent reduction in the shear wave velocity is necessary for the observed shifts. The opening of pre-existing cracks throughout the fault zone is the proposed mechanism to account for the observed shear wave reductions. The observation that elastic fault zone properties are soon recovered following episodes of large strains shows that cracks and fissures close rapidly after the strong shaking is over.

Earthquakes: Nucleation, Triggering, and Relationships With Aseismic Processes; Cargèse, Corsica, France, 3–10 November 2014

International audience; We use the strong motion data recorded at near-fault accelerometers combined with Global Positioning System (GPS) measurements to infer the space and time evolution of the M w = 7.2 1999 November 12 Düzce (Turkey) earthquake. We confirm that the rupture is subshear towards the west whereas it overpassed the shear speed towards the east. The mechanism of the earthquake is predominantly right lateral on a 65 • north-dipping fault. The strike-slip we infer represents at least 75% of the total seismic moment. We stress that the use of strong motion data allows us to constrain a strong spatial variation of the slip direction during faulting: to the west of the hypocentral region, the slip is oblique accompanied by a significant normal component. In the central and eastern part of the activated fault the slip is almost pure right lateral. This result is consistent both with the surface offset observations and with the long-term morphology of the fault. Neither strong motion data nor GPS data can resolve the complexity of the eastern termination of the Düzce rupture.

SUMMARY A semi-analytical, semi-numerical method of seismogram synthesis is presented for piecewise heterogeneous media resulting from an arbitrary source. The method incorporates the discrete wavenumber Green’s function representation into the boundary‐volume integral equation numerical techniques. The presentation is restricted to 2-D antiplane motion (SH waves). To model different parts of the media to a necessary accuracy, the incident, boundary-scattering and volume-scattering waves are separately formulated in the discrete wavenumber domain and handled flexibly at various accuracies using approximation methods. These waves are accurately superposed through the generalized Lippmann‐Schwinger integral (GLSI) equation. The full-waveform boundary method is used for the boundary-scattering wave to accurately simulate the reflection/transmission across strong-contrast boundaries. Meanwhile for volume heterogeneities, the following four flexible approaches have been developed in the numerical modelling scheme present here, with a great saving of computing time and memory: (i) the solution implicitly for the volume-scattering wave with high accuracy to model subtle effects of volume heterogeneities; (ii) the solution semi-explicitly for the volume-scattering wave using the average Fresnelradius approximation to volume integrations to reduce numerical burden by making the coefficient matrix sparser; (iii)the solution explicitly for the volume-scattering wave using the first-order Born approximation for smooth volume heterogeneities; and (iv)the solution explicitly for the volume-scattering wave using the second-order/high-order Born approximation for practical volume heterogeneities. These solutions are tested for dimensionless frequency responses to a heterogeneous alluvial valley where the velocity is perturbed randomly in the range of ca 5‐20 per cent, which is not rare in most complex near-surface areas. Numerical experiments indicate that several times of site amplification can be expected as a result of heterogeneities introduced in a homogeneous valley. The test also confirms that the first-order Born approximation to the volume-scattering wave is strictly valid for velocity perturbation less than 10 per cent and approximately used for up to 15 per cent for general applications. The second-order Born approximation to the volume-scattering wave is strictly valid for velocity perturbation less than 15 per cent and approximately used for up to 20 per cent for general applications.

We use continuous recordings at four stations installed near the eastern termination of the Izmit rupture to analyze the seismic activity there during the 5 hr leading to the Duzce earthquake. The records clearly show the presence of three groups of events. One of the groups originates from the middle of the Izmit rupture and consists of aftershocks of the M 5.7 Spanca Lake earthquake that had occurred a day earlier and was itself, to date, the second largest aftershock in the Izmit sequence. A second group of events is spread over the 30-km-long eastern (Karadere) segment of the Izmit rupture, and a comparison with the number of aftershocks expected, based on decay laws for similar sequences in other regions, suggests that seismic activity was high on the eastern segment in the hours leading to the Duzce earthquake. The third group of events occurs beyond the termination of the Izmit rupture and clusters around the hypocenter of the Duzce earthquake. It shows that, in the little over 5 hr leading to the Duzce earthquake, this destructive event was preceded by six precursory shocks ranging in magnitude from 0.9 to 2.6. Manuscript received 30 August 2000.

We used seismic stations that we deployed after the Izmit earthquake along the shores and islands of the Izmit Bay-Cinarcik basin to study the seismic activity that took place after the earthquake in the eastern Marmara Sea. The aftershock distribution indicates the existence of three clusters of activity there in the days that followed the earthquake. One of the clusters shows the extension of the E-W-trending Izmit rupture into the eastern Marmara Sea. The seismic activity there clearly outlines the fault plane of the main rupture and shows its termination 35 km beyond the Hersek peninsula. Two other clusters of activity are present in the region. One is in the Armutlu peninsula and the other one is located a few kilometers southwest of the Tuzla peninsula beneath the northern slope of the Cinarcik basin. The focal mechanism solutions indicate strike-slip faulting along the main branch of the Izmit rupture and normal faulting within the two swarms. The presence of the different mechanisms suggests the existence of slip partitioning in the region. A remarkable feature of the aftershock data is the absence of seismicity above 4 km in the Marmara Sea. Manuscript received 30 August 2000.

Strong site effects were observed during the two MW 5.7 and MW 6.0 main shocks of the Colfiorito seismic crisis which occured on September 26, 1997 in Umbria-Marche (Central Italy).The most obvious indications of these effects are the dramatic differences in damage shown by buildings of similar construction in neighboring villages.Such observations were specifically made in the Verchiano valley in the fault area, 15 km south of Colfiorito where the Verchiano village and the Colle and Camino hamlets were heavily damaged (MCS intensity IX-X) since the first main shock of 1997/09/26,while in contrast, the Curasci village located 2 km eastwards remains almost intact.In order to study the anomalous ground motion amplifications in this area, an array of 11, 3-components seismo- and accelero-meters was set up during the 1997/10/20-24 period, extending from the western side of the valley, up to the top of Mount San Salvatore, going accross the Colle and Curasci hamlets.During the experiment, 67 aftershocks enlightened the valley from the Colfiorito (10 km north) and the Sellano (6 km south) active swarms.Seismic refraction experiments were conducted at the same time in the 500 m wide, 1500 m long Verchiano valley in order to determine the thickness and main characteristics of the alluvial infilling.The main results are: (i) compared to the valley side ground motion, and for all the events, recordings in the central part of the valley (piana di Verchiano) show relative amplification of ∼10 with a clear lengthening of the seismogram duration by a factor of ∼2 – (ii) broad band relative amplification of ∼6–8 is also clearly identified at the top of the Mount San Salvatore overhanging the valley – (iii) any of the site effect measurements done explains by itself the strongly contrasted damage observed at Colle and Curasci: i.e. the modification of the near-field radiation pattern by interaction with the free heterogeneous surface may have induced local shadow zones that saved Curasci.

We found extensive evidence that the vertical ground accelerations produced during the largest shock (M = 6.0) of the 1997 Umbria-Marche earthquake sequence exceeded 1g in two areas close to the heavily-damaged villages of Annifo and Colle Croce. This evidence comes from the striking observation of thousands of freshly fractured and broken rocks and stones in these areas. Some of the broken stones lie isolated on soft detritic soil while others had been previously piled up, probably a long time agoto clear the fields for farming. The freshness of the cuts and fractures and the consistency of the observations for thousands of rocks and stones in these areas indicate that these rocks were thrown upwards during the earthquake, with breakage occurring at the time of impact. Ground motion calculations consistent with the static deformation inferred from GPS and interferometry data, show that the broken stones and rocks are found in the zone where the strongest shaking took place during the earthquake and that most of the shaking there was vertical.

We use the slip model of the Bolivian earthquake obtained by Ihmle [1998] to investigate the spatial distribution of stress drop and calculate the frictional heat generated locally on the fault. We find that the stress drop associated with the earthquake averages between 5 and 10 M Pa over the fault. The total frictional energy is estimated at between 3 × 1017 and 2 × 1018J. If the fault width is as thin as the one produced by transformational faulting in laboratory experiments, where it is about two orders of magnitude smaller than the local shear displacement, melting likely occurred on a significant portion of the fault. In this event, as proposed by Kanamori et al. [1998], once rupture is triggered, shear melting can quickly promote extensive slip.

We try to characterize the fracturing of the anisotropic shallow granite (

We study the propagation of P waves through media containing open fractures by performing numerical simulations. The important parameter in such problems is the ratio between crack length and incident wavelength. When the wavelength of the incident wavefield is close to or shorter than the crack length, the scattered waves are efficiently excited and the attenuation of the primary waves can be observed on synthetic seismograms. On the other hand, when the incident wavelength is greater than the crack length, we can simulate the anisotropic behaviour of fractured media resulting from the scattering of seismic waves by the cracks through the time delay of the arrival of the transmitted wave. The method of calculation used is a boundary element method in which the Green’s functions are computed by the discrete wavenumber method. For simplicity, the 2-D elastodynamic diffraction problem is considered. The rock matrix is supposed to be elastic, isotropic and homogeneous, while the cracks are all empty and have the same length and strike direction. An iterative method of calculation of the diffracted wavefield is developed in the case where a large number of cracks are present in order to reduce the computation time. The attenuation factor Q−1 of the direct waves passing through a fractured zone is measured in several frequency bands. We observe that the attenuation factor Q−1 of the direct P wave peaks around kd = 2, where k is the incident wavenumber and d the crack length, and decreases proportionally to (kd )−1 in the high-wavenumber range. In the long-wavelength domain, the velocity of the direct P wave measured for two different crack realizations is very close to the value predicted by Hudson’s theory on the overall elastic properties of fractured materials.

We present the results of an experiment about the effects on the seismic ground motion of a hill of large dimensions (6km long, 3km large and 700m high) near Corinth (Greece). We installed 7 seismometers across the hill, and analyzed the ground motion with different methods: the classical spectral ratios (CSR) and the horizontal to vertical spectral ratios calculated both on noise (HVNR) and earthquake data (RF). The amplification at the resonant frequency (around 0.7Hz) is low, as expected, but one of the two stations installed at the top exhibits large amplification (up to a factor of 10) at 3Hz. The comparison between the H/V results and the CSR ones suggests that the former method is able to point out the fundamental frequencies of a hill. Amplifications and spatial localization predicted by numerical 3D modeling are consistent with the experimental data only at frequencies lower than 1Hz.

We investigate the characteristics of the stress field associated with the 1995 Hyogo-ken Nanbu (Kobe) earthquake. We use for the study a tomographic model of the fault slip inferred from the numerous near-field recordings, and we calculate the space and time evolution of shear stress on the fault during the earthquake. We show that the spatial distribution of stress drop is very heterogeneous. The apparent strength of the fault at the onset of the earthquake is low (of the order of 1 MPa or less), which indicates that the pre-earthquake tectonic shear stress was close to the static friction over most of the fault. At many locations the stress drop rotates significantly during sliding. As was originally proposed by Spudich [1992], we use the requirement of colinearity between the directions of maximum shear stress and instantaneous slip to determine the initial stress on the fault at the onset of the earthquake. The pre-earthquake tectonic stress varies greatly over the fault and ranges from about 1 to 10 MPa. Average values of the initial and final shear stresses over the fault are 3.3 and 1.6 MPa, respectively, indicating that about half of the pre-earthquake tectonic stress was released during the earthquake. There is a relatively strong correlation between the initial and final stress distributions, which suggests that intrinsic fault properties, not modified by the earthquake, control the spatial distribution of tectonic stress over the fault. We show that on average over the fault the dynamic friction coefficient is equal to about 40% of the static friction coefficient. Diagrams depicting the evolution of shear stress as a function of slip are consistent with slip-weakening behavior, but their interpretation is questionable because of the poor resolution of the data at high frequency and because of the constraints imposed on the model to perform the inversion.

We investigate the space and time history of the shear stress produced on the fault during the 1992 Landers earthquake. The stress is directly calculated from the tomographic image of slip on the fault derived from near-source strong motion data. The results obtained shed some light on why the earthquake rupture cascaded along a series of previously distinct fault segments to produce the largest earthquake in California in over 40 years. Rupture on the 30 km long northernmost segment of the fault was triggered by a large dynamic increase of the stress field, of the order of 20 to 30 MPa, produced by the rupturing of the adjacent fault segments. Such a large increase was necessary to overcome the static friction on this strand of the fault, unfavorably oriented in today's tectonic stress field. This misorientation eventually led to the arrest of rupture. The same mechanism explains why rupture broke only a small portion of the Johnson Valley fault on which the earthquake originally started, before jumping to an adjacent fault more favorably oriented. We conclude from these results that the dynamic stress field could not sustain and drive the rupture along the strongly misoriented NW-SE strands of the preexisting fault system. Instead, the dynamic stress field produced new fractures favorably oriented in a N-S direction and connecting parts of the old fault system.

International audience; Over the past two decades, geophysical observations have shown that earthquakes can trigger other earthquakes, raising the possibility that earthquake interaction plays an important role in the earth's deformation. We analyze here a "storm" of earthquakes in Greece and show that their interaction provides remarkable insight into the mechanics of one of the fastest deforming continental region in the world. A rupture of the African slab initiates a cascade of large earthquakes and a long episode of slow slip marking the downward plunge of the slab, the concomitant rollback of the subduction, and the subsequent detachment of southern Greece from the Eurasian plate. Intense crustal deformation, indicative of the resulting plate stretching, follows. This slow deformation which spreads in a few months over more than 500 km lasts~3 years and triggers earthquakes. The observations also show that the retreat of the African subduction is the motor of the Aegean deformation.

A new hybrid two-step method for computation of P-SV seismic motion at inhomogeneous viscoelastic topographic structure is presented. The method is based on a combination of the discrete-wavenumber (DW), finite-difference (FD), and finite-element (FE) methods. In the first step, the DW method is used to calculate the source radiation and wave propagation in the background 1D medium. In the second step, the FD-FE algorithm is used to compute the wave propagation along the topographic structure. The accuracy of the method has been separately tested for inclusion of the attenuation and for inclusion of the free-surface topography through numerical comparisons with analytical and independent numerical methods. The method is a generalization of the hybrid DW-FD method of Zahradník and Moczo (1996) for localized structures with a flat free surface. Numerical computations for a ridge, sediment valley, and the ridge neighboring the sediment valley show that a ridge can considerably influence the response of the neighboring sediment valley. This means that the neighboring topographic feature should be taken into account even when we are only interested in the valley response.

We present a simple method to calculate the stress produced on an earthquake fault during rupture. This method allows the complete evaluation of the stress spatio-temporal history over the fault. We apply this approach to study the changes in shear stress produced during four of the largest earthquakes which occurred along the San Andreas fault system over the last 20 years: the Imperial Valley earthquake of 1979, the Morgan Hill earthquake of 1984, the Loma Prieta earthquake of 1989, and the Northridge earthquake of 1994. We use for this study the tomographic models of the fault rupture obtained from the inversion of the near-field seismic data recorded during these earthquakes. The results obtained show that the static and the dynamic stress drops vary greatly over the fault. The peak values obtained for the four earthquakes studied range from about 20 to 100 MPa. These high values imply that the initial shear stress level on the fault at the onset of the earthquake was high on at least a significant portion of the fault. The regions of the fault which break with a high stress drop are also the regions where slip is large. This suggests that most of the slip produced in a large earthquake takes place over the “strong” areas of the fault. Low slip regions tend to break with low stress drops. After the earthquake, the shear stress is increased over a significant portion of the fault, which corresponds to low slip regions. Aftershock activity tends to be concentrated in these areas of stress increase. The apparent strength of the fault before the earthquake (that is the local shear stress increase which is required for rupture) is also extremely heterogeneous. The rupture velocity seems to be inversely correlated with this apparent fault strength, the rupture accelerating over the “weak” areas of the fault and slowing down over the high strength areas.

We present a method of calculation to simulate numerically the growth and propagation of shear cracks in an elastic medium. The problem is formulated in terms of integral equations written along the crack contour. These boundary integral equations are discretized in both horizontal wavenumber and space and are solved in the frequency domain, at each time step. The earthquake model that we consider starts with a nucleating crack where slip occurs instantaneously at some arbitrary time. The shear stress is calculated at the tip of the crack, and, if and when it reaches the static friction, the crack tip propagates and the shear stress drops to the dynamic friction level. The boundary integral equation system is then solved for the new crack, and this scheme is repeated until the rupture stops or reaches the extremity of the fault. The method is naturally free of singularities. It can be applied to the case of a nonplanar fault. A few examples of simulation are presented to illustrate the method.

The aim of the present article is to further check the use of the horizontal-to-vertical (h/v) spectral ratio, which has been recently suggested as an indicator of site effects. The data set consists of 110, three-component, high sensitivity accelerograms, recorded at five different depths by the Garner Valley Downhole Array (GVDA), in southern California, with peak ground accelerations 0.0002 g ≦ ag ≦ 0.04 g, magnitudes 3.0 ≦ ML ≦ 4.6, and hypocentral distances 16 km ≦ R ≦ 107 km. First, the stability of the (h/v) spectral ratio is investigated by computing the mean for the whole data set in different depths. The (h/v) spectral ratio on the surface is compared with the surface-to-depth standard spectral ratio, with theoretical S-wave transfer functions derived from the vertical geotechnical profile, as well as with the (h/v) spectral ratio of synthetic accelerograms generated by the discrete wavenumber method. Both theoretical and experimental data show a good stability of the (h/v) spectral ratio shape, which is in good agreement with the local geological structure and is insensitive to the source location and mechanism. However, the absolute level of the (h/v) spectral ratio depends on the wave field and is different from the surface-to-depth spectral ratio. Consequently the (h/v) spectral ratio technique provides only partially the information that can be obtained from a downhole array. But surface-to-depth ratios may also be misleading because they combine effects at surface and at depth.

We present a semianalytical, seminumerical method to calculate the diffraction of elastic waves by an irregular topography of arbitrary shape. The method is a straightforward extension to three dimensions of the approach originally developed to study the diffraction of SH waves [Bouchon, 1985] and P-SV waves [Gaffet and Bouchon, 1989] by two-dimensional topographies. It relies on a boundary integral equation scheme formulated in the frequency domain where the Green functions are evaluated by the discrete wavenumber method. The principle of the method is simple. The diffracted wave field is represented as the integral over the topographic surface of an unknown source density function times the medium Green functions. The Green functions are expressed as integrals over the horizontal wavenumbers. The introduction of a spatial periodicity of thc topography combined with the discretization of the surface at equal intervals results in a discretization of the wavenumber integrals and in a periodicity in the horizontal wavenumber space. As a result, the Green functions are expressed as finite sums of analytical terms. The writing of the boundary conditions of free stress at the surface yields a linear system of equations where the unknowns are the source density functions representing the diffracted wave field. Finally, this system is solved iteratively using the conjugate gradient approach. We use this method to investigate the effect of a hill on the ground motion produced during an earthquake. The hill considered is 120 m high and has an elliptical base and ratios of height-to-half-width of 0.2 and 0.4 along its major and minor axes. The results obtained show that amplification occurs at and near the top of the hill over a broad range of frequencies. For incident shear waves polarized along the short dimension of the hill the amplification at the top reaches 100% around 10 Hz and stays above 50% for frequencies between 1.5 Hz and 20 Hz. For incident shear waves polarized along the direction of elongation of the topography, the maximum amplification occurs between 2 Hz and 5 Hz with values ranging from 50% to 75%. The results also show that the geometry of the topography exerts a very strong directivity on the wave field diffracted away from the hill and that at some distance from the hill this diffracted wave field consists mostly of Rayleigh waves.

The Northridge, California, earthquake that strongly shook the city of Los Angeles in January 1994, produced one of the highest ground accelerations ever recorded in an earthquake, at a site located on top of a small hill in Tarzana, about 6 km south of the epicenter. The subsequent study of aftershock recordings obtained by a dense seismic array deployed on the hill a few days after the earthquake showed the existence of a strong amplification at stations located at the top of the hill, relative to stations near the base (Spudich et al., 1996). Resonances and polarization rotations were also observed. We investigate in this study the role that the topography of the site played on the observed ground motions and accelerations. To this aim, we perform numerical simulations and study the response of the three-dimensional topography of the site to incident shear waves polarized in different directions. The method used is a boundary integral equation scheme in which the Green's functions are calculated by the discrete wavenumber method. The results obtained show that the topography of the site, though quite gentle (the hill is less than 20-m high), strongly affects the ground motions in the frequency range between 2 and 15 Hz. Many of the observed characteristics of the seismic response at Tarzana are explained in part by its topography: the consistent amplification of ground motion at and near the top of the hill, the directional seismic response of the hill that results in a strong amplification of the ground motion transverse to the direction of elongation of the hill, the existence of a fundamental transverse oscillatory resonance mode of the hill at 3 to 5 Hz, the rotation of the polarization of ground motion, and the spatial variation of amplification over the hill at the fundamental resonance mode. The seismic response of the topography, however, does not fully explain the amplitude of the effects observed. The three-dimensional geological structure of the site must in some way amplify the effect of the topography to produce the observed seismic response. In spite of not being as strong as the observed effect, the topographic effect of the site is considerable. The ground motion is amplified by factors ranging from 30% to 100% at some locations in the frequency range from 2 to 15 Hz. Rapid spatial variations of ground-shaking intensities can take place over distance scales of a few tens of meters at high frequency. Finally, the results of the simulation indicate that the topography of the site amplified the large east-west accelerations recorded there during the Northridge mainshock by 30% to 40%.