The characteristics of earthquake shaking are affected by the local site conditions. The effects of the local soil conditions are often quantified via an amplification factor (AF), which is defined as the ratio of the ground motion at the soil surface to the ground motion at a rock site at the same location. Amplification factors can be defined for any ground motion parameter, but most commonly are assessed for acceleration response spectral values at different oscillator periods. Site amplification can be evaluated for a site by conducting seismic site response analysis, which models the wave propagation from the base rock through the site-specific soil layers to the ground surface. An alternative to site-specific seismic response analysis is site amplification models. Site amplification models are empirical equations that predict the site amplification based on general characteristics of the site. Most of the site amplification models that already used in ground motion prediction equations characterize a site with two parameters: the average shear wave velocity in the top 30 m (VS30) and the depth to bedrock. However, additional site parameters influence site amplification and should be included in site amplification models. To identify the site parameters that help explain the variation in site amplification, ninety nine manually generated velocity profiles are analyzed using seismic site response analysis. The generated profiles have the same VS30 and depth to bedrock but a different velocity structure in the top 30 m. Different site parameters are investigated to explain the variability in the computed amplification. The parameter Vratio, which is the ratio of the average shear wave velocity between 20 m and 30 m to the average shear wave velocity in the top 10 m, is identified as the site parameter that most affects the computed amplification for sites with the same VS30 and depth to bedrock. To generalize the findings from the analyses in which only the top 30 m of the velocity profile are varied, a suite of fully randomized velocity profiles are generated and site response analysis is used to compute the amplification for each site for a range of input motion intensities. The results of the site response analyses conducted on these four hundred fully randomized velocity profiles confirm the influence of Vratio on site amplification. The computed amplification factors are used to develop an empirical site amplification model that incorporates the effect of Vratio, as well as VS30 and the depth to bedrock. The empirical site amplification model includes the effects of soil nonlinearity, such that the predicted amplification is a function of the intensity of shaking. The developed model can be incorporated into the development of future ground motion prediction equations. / text

The Central United States Seismic Observatory (CUSSO), is a fifteen-element array of three-component accelerometers and seismometers in the New Madrid seismic zone, within the upper Mississippi embayment. Its location within the thick (up to 1 km) sequences of unlithified sediment comprising the embayment, make CUSSO a unique array in its ability to directly measure seismic wave propagation, including ground motion site effect; however, before the observational data can be used for analysis, the orientation and instrument response of the CUSSO array must be defined. This study used cross-correlation and direct comparisons to filter out the instrument response and determine the instrument orientation, making CUSSO data ready for analysis, and making CUSSO a viable calibration site for other free-field sensors in the area. The corrected waveforms from five far-field earthquakes with magnitudes ranging between 2.5 and 4.7, which were recorded at CUSSO, were used to validate the site’s proposed dynamic soil model. The corrected bedrock motions were also numerically propagated through the CUSSO soil profile (transfer function) and compared, in terms of both peak acceleration and amplitude spectra, to the recorded surface observations.

Permettant de protéger les populations des inondations, de canaliser l'eau à des fins d'irrigation ou de production d'électricité, les digues en terre sont des ouvrages stratégiques dont la rupture peut avoir des conséquences dramatiques. La stabilité de ces ouvrages doit ainsi pouvoir être assurée, notamment en cas de séisme. Cependant, étant donné le très grand linéaire qu'ils représentent, la mise en œuvre de vastes campagnes de reconnaissances et de modèles numériques complexes n'est pas toujours envisageable dans les zones de sismicité modérée. L'utilisation d'outils simplifiés, faisant appel à des données facilement mesurables, se révèle indispensable pour fournir une indication sur la stabilité de tronçons de digue soumis à un chargement sismique. Les méthodes simplifiées existantes permettant d'estimer la réponse dynamique des remblais sont inadaptées au contexte des digues. En effet, alors que les digues sont pour la plupart construites dans des vallées sédimentaires, la majorité de ces méthodes, développées pour l’étude des barrages, supposent que l’ouvrage est fondé directement au rocher. Seules deux approches permettent de prendre en compte l’interaction entre l’ouvrage et son sol d’assise (effets de site) : la méthode de Sarma(1979) et celle de Papadimitriou(2014). Cependant, la première est basée sur des hypothèses très fortes (elle considère notamment un rocher rigide et un amortissement uniforme, qui plus est très fort) et la seconde se limite à des ouvrages ayant une hauteur supérieure à 20 si{m}. L'approche développée au cours de cette thèse repose sur des hypothèses plus réalistes, en considérant notamment des géométries adaptées à celles des digues, en prenant en compte les effets de site et en adaptant la dissipation d'énergie au niveau de sollicitation. Une étude paramétrique numérique (calcul de la réponse visco-élastique par éléments spectraux en deux dimensions) est utilisée pour couvrir un ensemble de configurations géométriques et mécaniques de digues. Pour prendre en compte la non-linéarité, les valeurs d'amortissement et de dégradation du module de cisaillement sont adaptées en chaque point des modèles à partir des résultats fournis par une série de calculs en linéaire équivalent 1D. Les réseaux de neurones artificiels, dont l'apprentissage est basé sur les résultats des simulations numériques, sont utilisés pour le développement d'une nouvelle méthode simplifiée. Des abaques sont réalisés, à partir des réseaux de neurones établis, pour offrir une alternative visuelle à leur utilisation "mathématiques". Les outils simplifiés auxquels aboutit cette thèse font appel à des paramètres facilement mesurables sur le terrain. Des méthodes géophysiques sont mises en œuvre au niveau de quatre sites de digues afin d'identifier le meilleur compromis de mesures permettant d'estimer les paramètres nécessaires à l'évaluation de la réponse dynamique de l'ouvrage. / Embankments are strategic facilities that, among other things, protect population from flooding and canalize rivers for energy production or for irrigation concerns. Since their rupture can lead to terrible consequences, they must remain stable, especially in case of earthquake. However, given the very large length of embankments, it is rarely possible to conduct complete investigations and complex numerical models in regions of moderate seismicity. Therefore, simplified tools to estimate the dynamic response of embankments using accessible embankments’ features are necessary to estimate their capacity to resist seismic loadings. Existing simplified methods, generally developed to assess the dynamic response of embankment-dams, turn out to be inappropriate to study the particular case of embankments along rivers – or other types of embankment, with small height and large length. The majority of these simplified methods do not take into account site effects: they assume the presence of rock directly at the base of the embankment, whereas most of large length embankments are located in alluvial valleys. Two methods only - Sarma(1979) and Papadimitriou(2014) - take into account the interaction between the embankment and its soil foundation. However, the first one relies on strong assumptions (among others, that the bedrock is considered rigid and that the viscous damping is supposed to be both spatially uniform and very large) and the second one is limited to tall embankments (higher than 20 si{m}).The method developed in this thesis relies on assumptions that are more realistic: it considers geometries adapted to large length embankments, it accounts for the presence of a soil foundation and includes an energy dissipation process consistent with expected levels of strain. A numerical parametric study is performed based on the computation with the spectral element method of the viscoelastic response of a large set of embankments spanning a wide range of geometrical and mechanical properties. In order to take into account nonlinearity, damping and shear modulus reduction are adapted in each point of the numerical models according to the results of a set of 1D linear equivalent computations. A new simplified method is derived from the obtained numerical results using artificial neural networks. Abacuses are produced from the neural networks so that engineers can have at their disposal a visual tool. This thesis leads to a simplified method that uses some accessible features of embankments as input parameters. The ability of classical geophysical methods to provide those input parameters is further estimated on four sites corresponding to different realistic configurations.

The central and eastern United States is subject to seismic hazards from both natural and induced earthquakes, as evidenced by the 1811-1812 New Madrid earthquake sequence, consisting of at least three magnitude 7 and greater earthquakes, and by four magnitude 5 and greater induced earthquakes in Oklahoma since 2011. To mitigate seismic hazards, both earthquake sources and their effects need to be characterized. Ground motion site response can cause additional damage to susceptible infrastructure and buildings. Recent studies indicate that Vs30, one of the primary site-response predictors used in current engineering practice, is not reliable. To investigate site response in the New Madrid Seismic Zone, ratios of surface-to-bedrock amplitude spectra, TFT, from S-wave recordings at the two deep vertical seismic arrays in the sediment-filled upper Mississippi Embayment (i.e., VSAP and CUSSO) were calculated. The mean TFT curves were compared with theoretical transfer functions; the results were comparable, indicating that TFT estimates of the empirical, linear SH-wave site responses at these sites. The suitability of surface S-wave horizontal-to-vertical spectral ratios, H/V, for estimating the empirical site transfer function was also evaluated. The results indicate that mean S-wave H/V curves are similar to TFT at low frequencies (less than the fifth natural frequencies) at both CUSSO and VSAP. SH-wave fundamental frequency, f0, and fundamental-mode amplification, A0, were evaluated as alternatives to the Vs30 proxy to estimate primary linear site-response characteristics at VSAP, CUSSO, and nine other seismic stations in the CEUS. In addition, calculated f0 and A0 were compared with the first peaks of S-wave H/V spectral ratios. The f0 and A0 were found to approximate the 1-D linear, viscoelastic, fundamental-mode responses at most stations. Also, S-wave H/V from weak-motion earthquakes can be used to measure f0. However, S-wave H/V does not reliably estimate A0 in the project area. S-wave H/V observations reveal site response within the frequency band of engineering interest from deeper, unmodeled geological structures. Because damaging or felt earthquakes induced by hydraulic fracturing and wastewater disposal have occurred in the CEUS, characterizing background seismicity prior to new large-scale subsurface fluid injection is important to identify cases of and the potential for induced seismicity. The Rogersville Shale in the Rome Trough of eastern Kentucky is being tested for unconventional oil and gas potential; production of this shale requires hydraulic fracturing, which has been linked to induced seismicity elsewhere in the CEUS. To characterize natural seismicity and to monitor induced seismicity during testing, a temporary seismic network was deployed in the Rome Trough near the locations of new, Rogersville Shale oil and gas test wells. Using the real-time recordings of this network and those of other regional seismic stations, three years of local seismicity were cataloged. Only three earthquakes occurred in the Rome Trough of eastern Kentucky, none of which was associated with the deep Rogersville Shale test wells that were stimulated during the time the network was in operation.

Une estimation précise de la magnitude est primordiale pour établir des catalogues de sismicité fiables. L’objectif de cette thèse est de développer une méthode d’estimation de la magnitude de moment à partir de la coda des ondes crustales applicable sur les sismogrammes numériques et analogiques. Afin d’étudier les propriétés de la coda en France, nous avons réalisé une analyse fréquentielle et régionale des enveloppes de coda. Nous avons pu estimer le moment sismique M0 et la magnitude de moment Mw pour des sismogrammes numériques, mais pas pour des enregistrements sur papier pour lesquels le filtrage n’est pas possible. Nous avons donc observé les propriétés de la coda dans le domaine temporel. Nous avons déterminé un modèle empirique afin de représenter les enveloppes de coda du signal brut et obtenir une magnitude de coda Mcoda. A partir de la relation linéaire Mcoda/Mw, nous avons déterminé la magnitude de moment directement sur les enregistrements analogiques en France. / Accurate magnitude determination is necessary to establish reliable seismicity catalogs in order to assess seismic hazard. The main goal of this PhD is to develop a method for estimating moment magnitude Mw from coda waves applicable to new digital seismograms as well as to old paper records in France. In order to analyze coda waves properties in France, a study of the regional and frequency properties of coda-wave envelopes has been performed. From this spectral analysis of coda waves, we can estimate seismic moment M0 and moment magnitude Mw from digital seismograms but not from paper records for which no frequency filtering is possible. Therefore, in a second step, we have analyzed the coda-waves properties directly in the time domain. We develop an empirical model which fits the coda envelopes of the raw signal and permits to obtain a coda magnitude Mcoda. From the Mcoda/Mw relationship, we determined the moment magnitude directly from the old paper records in France.