Thesis advisor: John E. Ebel / Degradation within an earth embankment structure is often unobservable from the surface. In order to evaluate the structural integrity of earth embankment dams and levees and identify subsurface zones of weakness that may result in future failures, various geophysical methods have been proposed as effective subsurface imaging tools. This study presents the results of using the horizontal-to-vertical spectral ratio (HVSR), seismic refraction analysis, and multi-channel analysis of surface waves (MASW) methods to estimate subsurface seismic structures for two earth embankment dams located in Chestnut Hill, MA, and Franklin Falls, NH. The estimated seismic velocity structures from the seismic refraction analysis and MASW performed in this study confirm the HVSR method is able to effectively estimate depth to bedrock at sites atop earth embankments using estimated fundamental frequencies. The MASW was found to resolve a low-velocity zone in the subsurface at the Chestnut Hill reservoir embankment that the seismic refraction method was unable to image, and this low-velocity zone is required to best fit a theoretical HVSR to an observed spectrum. Furthermore, the variation and uncertainty in fundamental frequency estimation were investigated by making repeated HVSR measurements at the Chestnut Hill embankment. / Thesis (BS) — Boston College, 2023. / Submitted to: Boston College. College of Arts and Sciences.. / Discipline: Departmental Honors. / Discipline: Earth and Environmental Sciences.

This dissertation details work aimed at better understanding topographic effects in earthquake ground motions. The experiment, conducted in Central-Eastern Utah, used frequent and predictable seismicity produced by underground longwall coal mining as a source of low-intensity ground motions. Locally-dense arrays of seismometers deployed over various topographic features were used to passively monitor seismic energy produced by mining-induced implosions and/or stress redistribution in the subsurface. The research consisted of two separate studies: an initial feasibility experiment (Phase I) followed by a larger-scale main study (Phase II). Over 50 distinct, small-magnitude (M[subscript 'L'] < 1.6) seismic events were identified in each phase. These events were analyzed for topographic effects in the time domain using the Peak Ground Velocity (PGV), and in the frequency domain using the Standard Spectral Ratio (SSR) method, the Median Reference Method (MRM), and the Horizontal-to-Vertical Spectral Ratio (HVSR) method. The polarities of the horizontal ground motions were also visualized using directional analyses. The various analysis methods were compared to assess their ability to estimate amplification factors and determine the topographic frequencies of interest for each feature instrumented. The MRM was found to provide the most consistent, and presumably accurate, estimates of the amplification factor and frequency range for topographic effects. Results from this study clearly indicated that topographic amplification of ground motions does in fact occur. These amplifications were very frequency dependent, and the frequency range was correctly estimated in many, but not all, cases using simplified, analytical methods based on the geotechnical and geometrical properties of the topography. Amplifications in this study were found to generally range from 2 to 3 times a reference/baseline site condition, with some complex 3D features experiencing amplifications as high as 10. Maximum amplifications occurred near the crest of topographic features with slope angles greater than approximately 15 degrees, and the amplifications were generally oriented in the direction of steepest topographic relief, with some dependency on wave propagation direction. / text

Six earthquake acceleration time histories were used to evaluate the groundmotion response of two sites, VSAP and VSAS, near the New Madrid Seismic Zone. These earthquakes ranged in magnitude from Mw 3.6 to Mw 5.2 and were located 46 to 173 km away from the recording instruments. These two sites are underlain by thick sequences (100 and 590 m) of unlithified soil that have been shown to greatly influence earthquake ground motions. Near-surface soil dynamic properties were characterized at the two sites using seismic SH-wave refraction, P-S suspension logging, borehole electrical logs, and geotechnical logging methods. The soil properties were developed into a soil model for each site and the soil models were used to compare theoretical ground-motion models to the actual strong-motion time histories. An 1-D ground-motion simulation program (EERA) was used to complete the theoretical ground-motion analysis. The results of the model indicated that the soils underlying VSAP generated amplification factors of 0.9 to 2.9 at about 6 and 9 Hz. Soils underlying VSAS generated amplification factors of 1.8 to 4.2 at about 5 Hz. These values correlated well with the observations at the two sites.

La configuration géologique des couches de sol proches de la surface peut modifier fortement le mouvement sismique en surface (effets de site lithologiques). Ces effets de sites peuvent être évalués numériquement ou empiriquement. Ce travail s'inscrit dans la thématique de l'évaluation des effets de site lithologiques et de la variabilité de la réponse sismique des sites. Il se base principalement sur l'analyse d'enregistrements de séismes. Nous avons étudié dans un premier temps la variabilité de la réponse sismique, par classe de sites et pour des mouvements sismique faibles (variabilité inter-sites), puis une partie de la variabilité inter-évènements en un site, due aux effets du comportement non-linéaire des sols. Enfin, nous avons inversé les courbes de réponse du site afin de préciser à quelles profondeurs le comportement du sol est non-linéaire. La base de données accélérométriques KiK-net (Japon) est constituée de 688 forages instrumentés avec un capteur en surface et un autre en fond de puits. Cette configuration en réseau vertical permet de calculer en chacun des sites la courbe de réponse empirique du forage. Ce réseau a été choisi compte tenu du nombre important de sites instrumentés, du nombre de séismes enregistrés (plus de 46 000 enregistrements ont été recueillis et analysés) et de l'existence de caractérisation géotechnique des sites (profils de vitesse de propagation des ondes de cisaillement et de compression).L'analyse de la variabilité de la réponse sismique par classe de site nous a permis de préciser les paramètres qui seraient à mesurer afin d'améliorer l'évaluation des effets de site. En régime linéaire, il s'agit du gradient du profil de vitesse calculé jusqu'à 30 m (B30) ou 100 m (B100) et de la fréquence de résonance fondamentale (f0). Ces paramètres pourraient être utilisés afin de réduire la variabilité inter-sites du mouvement sismique en surface (i.e. variabilité dans les GMPEs Equations prédictives du mouvement sismique).L'analyse de la variabilité inter-évènement associée au comportement non-linéaire des sols nous a fourni les paramètres pertinents pour l'analyse des sites en régime non-linéaire : B30, l'amplification maximale (Apred) et la fréquence associée (fpred). Cette analyse a également montré : - que le PGA (Peak Ground Acceleration, accélération maximale du sol) est un paramètre pertinent pour l'étude des effets du comportement non-linéaire des sols sur la réponse des sites. - que, quelque soit le site, le comportement non-linéaire des sols affecte la courbe de réponse du site à partir d'accélérations modérées (75 cm/s2 en fond de puits). - que la caractérisation non-linéaire d'un site, en vue de l'évaluation des effets de sites, pourrait être réduite à la caractérisation des couches de sol superficielles. Cette dernière assertion peut avoir une influence importante pour la caractérisation non-linéaire des sites. Elle a été confirmée par l'inversion comparée des fonctions de transfert forages linéaires et non-linéaires. Finalement, notre travail a également montré :• que l'analyse de sensibilité de la colonne de sol « à priori » est un bon outil pour déterminer la résolution de l'inversion compte tenu des informations disponibles ainsi que les paramètres du sol contrôlant les pics d'amplification.• que l'utilisation conjointe de l'analyse de sensibilité et de la comparaison des fonctions de transfert, sans étape d'inversion, pouvait être suffisante pour évaluer les profondeurs où le sol peut avoir un comportement non-linéaire important / Local geology can strongly affect seismic ground motion at the surface. These so-called site-effects can be evaluated either numerically by simulating the seismic wave propagation or empirically using earthquake recordings analyses or statistical correlations between site parameters and site effects. This thesis concerns the improvement of site effect evaluation and the consideration of the seismic site response variability. This work is mainly based on the analysis of earthquake recordings. First, we analyzed the site response variability between sites for similar ground motion incident intensity (weak motion) and then, we analysed the site response variability between events in one site caused by non-linear soil behaviour. Finally, we used inversion method to find the depths where soil non-linear behaviour mostly occurs.We analyzed various earthquake recordings from the KiK-net database in Japan (more than 46 000), which is composed of more than 688 surface/borehole instruments. The vertical array configuration allows the computation of the empirical site response in borehole condition. This database was chosen because of its large amount of instrumented sites located at sediments, its large amount of accelerometric data and the existence of characterisation of the shear and compressive wave velocity profiles down to the borehole depth. The analysis of the seismic response variability per site classes indicated which parameters must be measured to improve the site-effect assessment. In the linear range, the parameters are the Vs profile gradient calculated until 30 m (B30) or 100 m depth (B100) and the fundamental resonance frequency of the site (f0). These additional parameters to Vs30 can be used to reduce the between-sites surface motion variability (such as in GMPEs, Ground Motion Prediction Equations).Besides, the analysis of inter-event site response variability caused by non-linear soil behaviour showed the relevant parameters for analysis of site effects in non-linear range: B30, The maximal amplification (Apred) and the associated frequency (fpred). This analysis showed as well: - that the PGA (Peak Ground Acceleration) is a relevant parameter for non-linear site effect assessment. - that, whatever the site, non-linear soil behaviour affects the site response curves from moderate acceleration (75 cm/s2 at the down-hole station). - that only information on the non-linear soil behaviour of the superficial layers is enough to fully assess the non-linear site responses. This last conclusion may have a large impact for non-linear soil characterisation. It has been confirmed by inversion of linear and non-linear borehole site responses and comparison of the obtained Vs profiles. At the same time, the present work showed : - that the depth from which non-linear soil behaviour has no influence on site response depends on the site and on the intensity of the seismic input motion. - that careful attention needs to be taken when inverting data from vertical arrays. Sensitivity analyses are a powerful tool to evaluate the resolution of the inversion considering the available information and the soil parameters that are well solved during the inversion. - That the combined used of sensitivity analysis with comparison of transfer function, without inversion, could be enough to assess the depth where the non-linear soil behaviour mostly take place

Situated in an active tectonic region, Santiago de Chile, the country´s capital with more than six million inhabitants, faces tremendous earthquake hazard. Macroseismic data for the 1985 Valparaiso and the 2010 Maule events show large variations in the distribution of damage to buildings within short distances indicating strong influence of local sediments and the shape of the sediment-bedrock interface on ground motion. Therefore, a temporary seismic network was installed in the urban area for recording earthquake activity, and a study was carried out aiming to estimate site amplification derived from earthquake data and ambient noise. The analysis of earthquake data shows significant dependence on the local geological structure with regards to amplitude and duration. Moreover, the analysis of noise spectral ratios shows that they can provide a lower bound in amplitude for site amplification and, since no variability in terms of time and amplitude is observed, that it is possible to map the fundamental resonance frequency of the soil for a 26 km x 12 km area in the northern part of the Santiago de Chile basin. By inverting the noise spectral rations, local shear wave velocity profiles could be derived under the constraint of the thickness of the sedimentary cover which had previously been determined by gravimetric measurements. The resulting 3D model was derived by interpolation between the single shear wave velocity profiles and shows locally good agreement with the few existing velocity profile data, but allows the entire area, as well as deeper parts of the basin, to be represented in greater detail. The wealth of available data allowed further to check if any correlation between the shear wave velocity in the uppermost 30 m (vs30) and the slope of topography, a new technique recently proposed by Wald and Allen (2007), exists on a local scale. While one lithology might provide a greater scatter in the velocity values for the investigated area, almost no correlation between topographic gradient and calculated vs30 exists, whereas a better link is found between vs30 and the local geology. When comparing the vs30 distribution with the MSK intensities for the 1985 Valparaiso event it becomes clear that high intensities are found where the expected vs30 values are low and over a thick sedimentary cover. Although this evidence cannot be generalized for all possible earthquakes, it indicates the influence of site effects modifying the ground motion when earthquakes occur well outside of the Santiago basin. Using the attained knowledge on the basin characteristics, simulations of strong ground motion within the Santiago Metropolitan area were carried out by means of the spectral element technique. The simulation of a regional event, which has also been recorded by a dense network installed in the city of Santiago for recording aftershock activity following the 27 February 2010 Maule earthquake, shows that the model is capable to realistically calculate ground motion in terms of amplitude, duration, and frequency and, moreover, that the surface topography and the shape of the sediment bedrock interface strongly modify ground motion in the Santiago basin. An examination on the dependency of ground motion on the hypocenter location for a hypothetical event occurring along the active San Ramón fault, which is crossing the eastern outskirts of the city, shows that the unfavorable interaction between fault rupture, radiation mechanism, and complex geological conditions in the near-field may give rise to large values of peak ground velocity and therefore considerably increase the level of seismic risk for Santiago de Chile. / Aufgrund ihrer Lage in einem tektonisch aktiven Gebiet ist Santiago de Chile, die Hauptstadt des Landes mit mehr als sechs Millionen Einwohnern, einer großen Erdbebengefährdung ausgesetzt. Darüberhinaus zeigen makroseismische Daten für das 1985 Valparaiso- und das 2010 Maule-Erdbeben eine räumlich unterschiedliche Verteilung der an den Gebäuden festgestellten Schäden; dies weist auf einen starken Einfluss der unterliegenden Sedimentschichten und der Gestalt der Grenzfläche zwischen den Sedimenten und dem Festgestein auf die Bodenbewegung hin. Zu diesem Zweck wurde in der Stadt ein seismisches Netzwerk für die Aufzeichnung der Bodenbewegung installiert, um die auftretende Untergrundverstärkung mittels Erdbebendaten und seismischem Rauschen abzuschätzen. Dabei zeigt sich für die Erdbebendaten eine deutliche Abhängigkeit von der Struktur des Untergrunds hinsichtlich der Amplitude der Erschütterung und ihrer Dauer. Die Untersuchung der aus seismischem Rauschen gewonnenen horizontal-zu-vertikal-(H/V) Spektral-verhältnisse zeigt, dass diese Ergebnisse nur einen unteren Grenzwert für die Bodenverstärkung liefern können. Weil jedoch andererseits keine zeitliche Veränderung bei der Gestalt dieser Spektralverhältnisse festgestellt werden konnte, erlauben die Ergebnisse ferner, die Resonanzfrequenz des Untergrundes für ein 26 km x 12 km großes Gebiet im Nordteil der Stadt zu bestimmen. Unter Zuhilfenahme von Informationen über die Dicke der Sedimentschichten, welche im vorhinein schon durch gravimetrische Messungen bestimmt worden war, konnten nach Inversion der H/V-Spektralverhältnisse lokale Scherwellengeschwindigkeitsprofile und nach Interpolation zwischen den einzelnen Profilen ein dreidimensionales Modell berechnet werden. Darüberhinaus wurde mit den verfügbaren Daten untersucht, ob auf lokaler Ebene ein Zusammenhang zwischen der mittleren Scherwellengeschwindigkeit in den obersten 30 m (vs30) und dem Gefälle existiert, ein Verfahren, welches kürzlich von Wald und Allen (2007) vorgestellt wurde. Da für jede lithologische Einheit eine starke Streuung für die seismischen Geschwindigkeiten gefunden wurde, konnte kein Zusammenhang zwischen dem Gefälle und vs30 hergestellt werden; demgegenüber besteht zumindest ein tendenzieller Zusammenhang zwischen vs30 und der unterliegenden Geologie. Ein Vergleich der Verteilung von vs30 mit den MKS-Intensitäten für das 1985 Valparaiso-Erdbeben in Santiago zeigt, dass hohe Intensitätswerte vor allem in Bereichen geringer vs30-Werte und dicker Sedimentschichten auftraten. Weiterhin ermöglichte die Kenntnis über das Sedimentbeckens Simulationen der Bodenbewegung mittels eines spektralen-Elemente-Verfahrens. Die Simulation eines regionalen Erdbebens, welches auch von einem dichten seismischen Netzwerk aufgezeichnet wurde, das im Stadtgebiet von Santiago infolge des Maule-Erdbebens am 27. Februar 2010 installiert wurde, zeigt, dass das Modell des Sedimentbeckens realistische Berechnungen hinsichtlich Amplitude, Dauer und Frequenz erlaubt und die ausgeprägte Topographie in Verbindung mit der Form der Grenzfläche zwischen den Sedimenten und dem Festgestein starken Einfluss auf die Bodenbewegung haben. Weitere Untersuchungen zur Abhängigkeit der Bodenerschütterung von der Position des Hypozentrums für ein hypothetisches Erdbeben an der San Ramón-Verwerfung, welche die östlichen Vororte der Stadt kreuzt, zeigen, dass die ungünstige Wechselwirkung zwischen dem Verlauf des Bruchs, der Abstrahlung der Energie und der komplexen geologischen Gegebenheiten hohe Werte bei der maximalen Bodengeschwindigkeit erzeugen kann. Dies führt zu einer signifikanten Zunahme des seismischen Risikos für Santiago de Chile.

The proliferation of megacities in many developing countries, and their location in areas where they are exposed to a high risk from large earthquakes, coupled with a lack of preparation, demonstrates the requirement for improved capabilities in hazard assessment, as well as the rapid adjustment and development of land-use planning. In particular, within the context of seismic hazard assessment, the evaluation of local site effects and their influence on the spatial distribution of ground shaking generated by an earthquake plays an important role. It follows that the carrying out of earthquake microzonation studies, which aim at identify areas within the urban environment that are expected to respond in a similar way to a seismic event, are essential to the reliable risk assessment of large urban areas. Considering the rate at which many large towns in developing countries that are prone to large earthquakes are growing, their seismic microzonation has become mandatory. Such activities are challenging and techniques suitable for identifying site effects within such contexts are needed. In this dissertation, I develop techniques for investigating large-scale urban environments that aim at being non-invasive, cost-effective and quickly deployable. These peculiarities allow one to investigate large areas over a relative short time frame, with a spatial sampling resolution sufficient to provide reliable microzonation. Although there is a negative trade-off between the completeness of available information and extent of the investigated area, I attempt to mitigate this limitation by combining two, what I term layers, of information: in the first layer, the site effects at a few calibration points are well constrained by analyzing earthquake data or using other geophysical information (e.g., shear-wave velocity profiles); in the second layer, the site effects over a larger areal coverage are estimated by means of single-station noise measurements. The microzonation is performed in terms of problem-dependent quantities, by considering a proxy suitable to link information from the first layer to the second one. In order to define the microzonation approach proposed in this work, different methods for estimating site effects have been combined and tested in Potenza (Italy), where a considerable amount of data was available. In particular, the horizontal-to-vertical spectral ratio computed for seismic noise recorded at different sites has been used as a proxy to combine the two levels of information together and to create a microzonation map in terms of spectral intensity ratio (SIR). In the next step, I applied this two-layer approach to Istanbul (Turkey) and Bishkek (Kyrgyzstan). A similar hybrid approach, i.e., combining earthquake and noise data, has been used for the microzonation of these two different urban environments. For both cities, after having calibrated the fundamental frequencies of resonance estimated from seismic noise with those obtained by analysing earthquakes (first layer), a fundamental frequency map has been computed using the noise measurements carried out within the town (second layer). By applying this new approach, maps of the fundamental frequency of resonance for Istanbul and Bishkek have been published for the first time. In parallel, a microzonation map in terms of SIR has been incorporated into a risk scenario for the Potenza test site by means of a dedicated regression between spectral intensity (SI) and macroseismic intensity (EMS). The scenario study confirms the importance of site effects within the risk chain. In fact, their introduction into the scenario led to an increase of about 50% in estimates of the number of buildings that would be partially or totally collapsed. Last, but not least, considering that the approach developed and applied in this work is based on measurements of seismic noise, their reliability has been assessed. A theoretical model describing the self-noise curves of different instruments usually adopted in microzonation studies (e.g., those used in Potenza, Istanbul and Bishkek) have been considered and compared with empirical data recorded in Cologne (Germany) and Gubbio (Italy). The results show that, depending on the geological and environmental conditions, the instrumental noise could severely bias the results obtained by recording and analysing ambient noise. Therefore, in this work I also provide some guidelines for measuring seismic noise. / Aufgrund des enormen Wachstums neuer Megastädte und deren Vordringen in gefährdete Gebiete auf der einen Seite sowie der mangelnden Erdbebenvorsorge in vielen Entwicklungsländern auf der anderen Seite sind verbesserte Verfahren für die Beurteilung der Gefährdung sowie eine rasche Umsetzung bei der Raumplanung erforderlich. Im Rahmen der seismischen Gefährdungsabschätzung spielt insbesondere die Beurteilung lokaler Standorteffekte und deren Einfluss auf die durch ein Erdbeben verursachte räumliche Verteilung der Bodenerschütterung eine wichtige Rolle. Es ist daher unabdingbar, mittels seismischer Mikrozonierungsstudien diejenigen Bereiche innerhalb dicht besiedelter Gebiete zu ermitteln, in denen ein ähnliches Verhalten im Falle seismischer Anregung erwartet wird, um daraus eine zuverlässige Basis bei der Risikoabschätzung großer städtischer Gebiete zu erhalten. Aufgrund des schnellen Wachstums vieler Großstädte in Entwicklungsländern ist eine seismische Mikrozonierung zwingend erforderlich, stellt aber auch eine große Herausforderung dar; insbesondere müssen Verfahren verfügbar sein, mit deren Hilfe rasch eine Abschätzung der Standorteffekte durchgeführt werden kann. In der vorliegenden Arbeit entwickle ich daher Verfahren für die Untersuchung in Großstädten, die darauf abzielen, nicht-invasiv, kostengünstig und schnell durchführbar zu sein. Damit lassen sich innerhalb eines relativ kurzen Zeitraums große Gebiete untersuchen, falls der räumlichen Abstand zwischen den Messpunkten klein genug ist, um eine zuverlässige Mikrozonierung zu gewährleisten. Obwohl es eine gegenläufige Tendenz zwischen der Vollständigkeit aller Informationen und der Größe des untersuchten Gebiets gibt, versuche ich, diese Einschränkung durch Verknüpfung zweier Informationsebenen zu umgehen: In der ersten Ebene werden die Standorteffekte für einige Kalibrierungspunkte durch die Analyse von Erdbeben oder mittels anderer geophysikalischer Datensätze (z.B. Scherwellengeschwindigkeitsprofile) bestmöglich abgeschätzt, in der zweiten Ebene werden die Standorteffekte durch Einzelstationsmessungen des seismischen Rauschens für ein größeres Gebiet bestimmt. Die Mikrozonierung erfolgt hierbei mittels spezifischer, fallabhängiger Parameter unter Berücksichtigung eines geeigneten Anknüpfungspunktes zwischen den beiden Informationensebenen. Um diesen Ansatz der Mikrozonierung, der in dieser Arbeit verfolgt wurde, zu präzisieren, wurden in Potenza (Italien), wo eine beträchtliche Menge an Daten verfügbar war, verschiedene Verfahren untersucht. Insbesondere kann das Spektralverhältnis zwischen den horizontalen und vertikalen Seismometerkomponenten, welche für das seismische Rauschen an mehreren Orten aufgenommen wurde, als eine erste Näherung für die relative Verstärkung der Bodenbewegung verwendet werden, um darauf aufbauend die beiden Informationsebenen zu verknüpfen und eine Mikrozonierung hinsichtlich des Verhältnisses der spektralen Intensität durchzuführen. Anschließend führte ich diesen Zwei-Ebenen-Ansatz auch für Istanbul (Türkei) und Bischkek (Kirgisistan) durch. Für die Mikrozonierung dieser beiden Städte habe ich denselben Hybridansatz, der Daten von Erdbeben und von seismischem Rauschen verbindet, verwendet. Für beide Städte wurde nach Gegenüberstellung der Resonanzfrequenz des Untergrunds, die zum einen mit Hilfe des seismischen Rauschens, zum anderen durch Analyse von Erdbebendaten bestimmt worden ist (erste Ebene), eine Karte der Resonanzfrequenz unter Verwendung weiterer Messungen des seismischen Rauschens innerhalb des Stadtgebiets erstellt (zweite Ebene). Durch die Anwendung dieses neuen Ansatzes sind vor kurzem zum ersten Mal auch Karten für die Resonanzfrequenz des Untergrunds für Istanbul und Bischkek veröffentlicht worden. Parallel dazu wurde für das Testgebiet in Potenza eine auf dem spektralen Intensitätsverhältnis (SIR) basierende Mikrozonierungskarte in ein Risikoszenario mittels der Regression zwischen SIR und makroseismischer Intensität (EMS) integriert. Diese Szenariostudie bestätigt die Bedeutung von Standorteffekten innerhalb der Risikokette; insbesondere führt deren Einbeziehung in das Szenario zu einem Anstieg von etwa 50% bei der Zahl der Gebäude, für die ein teilweiser oder gar vollständiger Zusammenbruch erwartet werden kann. Abschließend wurde der im Rahmen dieser Arbeit entwickelte und angewandte Ansatz auf seine Zuverlässigkeit geprüft. Ein theoretisches Modell, das zur Beschreibung des Eigenrauschens verschiedener Instrumente, die in der Regel in Mikrozonierungsstudien (z. B. in Potenza, Istanbul und Bischkek) zum Einsatz kommen, wurde untersucht, und die Ergebnisse wurden mit Daten verglichen, die vorher bereits in Köln (Deutschland) und Gubbio (Italien) aufgenommen worden waren. Die Ergebnisse zeigen, dass abhängig von den geologischen und umgebenden Bedingungen das Eigenrauschen der Geräte die Ergebnisse bei der Analyse des seismischen Rauschens stark verzerren kann. Deshalb liefere ich in dieser Arbeit auch einige Leitlinien für die Durchführung von Messungen des seismischen Rauschens.

Les séismes sont connus comme des catastrophes naturelles destructrices et meurtrières. Particulièrement dans les régions montagneuses, les effets des séismes sont encore beaucoup plus aggravés à cause des effets de site topographiques. La présence d'une pente est à l'origine d'une amplification du signal sismique notamment dans le voisinage de la crête. De nombreux séismes anciens tels que le séisme de Lambesc (1909, Ms=6.2) qui a conduit à la destruction du village de Rognes (Bouches-du-Rhône, France), ou plus récemment le séisme d'Athènes en Grèce (1999, Ms=5.9) qui a endommagé la ville d'Adames ainsi que les séismes de Chichi à Taiwan (1999, Mw=7.6), du Salvador (2001, Mw=7.6) et du Sichuan (2008, Mw=7.9) responsables de très nombreux glissements de terrain particulièrement catastrophiques sont des exemples représentatifs. Les investigations post-sismiques ont montré une contribution importante des effets de site topographiques au bilan des dommages humains et matériels. Dans le but d'améliorer les connaissances de ce phénomène, de nombreuses simulations numériques ont été réalisées sur des modèles de versant isolés en utilisant le logiciel FLAC 2D (Itasca). Les résultats obtenus ont été analysés en considérant plusieurs critères comme les facteurs d'amplification, les surfaces et les dimensions des zones d'amplification ainsi que la distribution spatiale de ces zones dans le massif de versant. Des études paramétriques ont permis de mettre en évidence le rôle principal de la fréquence adimensionnelle, rapport de la hauteur du versant à la longueur d'onde du signal sismique, dans l'évaluation des effets de site topographiques. Ces résultats montrent également que la pente est le deuxième paramètre important, suivie du coefficient de Poisson et de la géométrie de la crête et du pied de la pente.Ces résultats numériques ont également permis de définir des relations empiriques, nommées ANS, qui permettent d'estimer les effets de site topographiques selon les différents critères d'interprétation. Ces formules sont adaptées pour les signaux sismiques avec une bande de fréquences prédominantes relativement étroite. Pour les autres cas, la Méthode de Décomposition des ondes (MD) a été développé dans ce travail. Cette méthode est basée sur la décomposition du signal complexe multi-fréquentiel en plusieurs ondes mono-fréquentielles par transformée de Fourier. L'effet du signal complexe est alors évalué en faisant une combinaison des effets individuels de chaque onde élémentaire. Ces approches développées dans ce travail (ANS et MD) permettent d'évaluer les effets de site topographiques, en tenant compte de plusieurs paramètres géomorphologiques et sismiques du massif de versant ainsi que du contenu fréquentiel du signal sismique, sans avoir recours à des simulations numériques. / Earthquakes are known as destructive and murderous natural catastrophe. Particularly in the mountainous regions, the effects of earthquakes are still much aggravated due to the topographic site effects. The presence of a slope causes an amplification of the seismic signal, particularly in the vicinity of the crest. Numerous earthquakes in the past such as the 1909 Lambesc earthquake (Ms=6.2) which led to the destruction of the Rognes village (Bouches-du-Rhône, France), or more recently the 1999 Athens, Greece earthquake (Ms=5.9) which damaged the Adames city as well as the 1999 Chichi, Taiwan earthquake (MW=7.6), the 2001 El Salvador earthquake (MW=7.6) and the 2008 Sichuan earthquake (MW=7.9) responsible for numerous catastrophic landslides are representative examples. Post-seismic investigations demonstrated a significant contribution of topographic site effects on the human and material damage assessment.In order to improve the knowledge of this phenomenon, numerous numerical simulations were performed on the step-like slope models by using the FLAC 2D software (Itasca). The obtained results were analyzed by considering various criteria such as amplification factors, dimensions and area of amplification zones as well as spatial distribution of these zones inside the slope mass. The parametric analyses allow underlining the principal role of the dimensionless frequency, ratio of the slope height to the wavelength of the seismic signal, in the evaluation of topographic site effects. These results also show that the slope angle is the second important parameter, followed by the Poisson's ratio and the geometry of the crest and of the toe of slope.These numerical results allow pointing out empirical equations, called ANS, which can be used to estimate the topographic site effects according to different interpretation criteria. These formulas are suitable for seismic signals with a relatively narrow band of predominant frequencies. For other cases, the method concerning decomposition of seismic incident wave (MD) was developed in this work. This method is based on the decomposition of the complex multi-frequency seismic signal into multiple mono-frequency waves by using Fourier transform. The effect of the complex signal is then evaluated through a combination of the effects of elementary waves. The approaches developed in this work (ANS and MD) allow evaluating the topographic site effects, by taking into account geomorphologic and seismic parameters of the slope as well as the frequency content of the seismic signal, without recourse to numerical simulations.

Local site conditions can greatly increase the intensity and character of earthquake shaking and, thus, the extent and type of structural damage. The removal and replacement of in-situ soils with compacted gravel-sized volcanic rock has been prevalent in the Icelandic foundation subgrade construction practice for decades, despite the unknown seismic site response characteristics of the fill (e.g., the predominant frequency and relative site amplification). To fill this knowledge gap, over 500 hours of microtremor measurements were made at six study sites located throughout the Reykjavík, Iceland, capital region. Measurements recorded at various construction stages (e.g., the in-situ or pre-excavation, post-excavation, intermediate grades, and final grade) reveal the change in site response characteristics before and after gravel fill placement. The data was analyzed using the horizontal-to-vertical spectral ratio (HVSR) technique over a bandwidth of 0.3 to 25 Hz. Generally, the pre-excavation condition had a predominant site frequency between 3.5 and 7 Hz with relative amplification between 3.8 and 3.9 times. The placement of gravel fill atop dense to very dense silty sand underlain by bedrock shifts the predominant frequency between 10 and 16 Hz with a relative peak amplification between 2.5 and 5.3 times, generally increasing with fill thickness. Fill underlain by undulating lava rock also results in a higher a predominant frequency between 9 and 10.5 Hz, but little change in relative site amplification occurred at these frequencies (between 0.95 and 1.2 times). This dissimilarity is due to the unique lava rock HVSR signatures which have large amplification values (between 2.6 to 3.9 times) throughout the high-band frequency range. Additional investigations of sites underlain by lava rock are required to draw stronger empirical trends. The data set produced by this study can serve as a useful tool for the local geotechnical and seismological communities to mitigate seismic risk for the capital region. / Master of Science / Problematic soil conditions can greatly increase the intensity and character of earthquake shaking and, thus, the extent and type of building damage. The removal of native soils and replacement with compacted gravel-sized fill has been the predominant building foundation subgrade construction method in Iceland for decades. The practice of removal and replacement is one of the oldest and conceptually simplest approaches of site improvement to reduce settlement and increase soil strength. However, the understanding of how compacted gravel fill responds to earthquake shaking was nonexistent in literature. To fill this knowledge gap, the response characteristics of compacted gravel fill were derived using the horizontal-to-vertical spectral ratio (HVSR) and standard spectral ratio (SSR) analysis techniques from a data set of over 500 hours of experimental in-field measurements. Measurements were recorded at various construction stages (e.g., pre-excavation or native soil, post-excavation, intermediate fill grades, and the final fill grade) to reveal the change in site response characteristics before and after gravel fill placement. The findings presented in this thesis can serve as useful information for the local geotechnical and seismological communities to mitigate seismic risk (e.g., the probability of building damage and/or loss of lives) of structures with compacted gravel fill subgrades in the Reykjavík, Iceland capital region.

Dans cette thèse, des simulations numériques ont d'abord été effectuées avec le logiciel FLAC (Itasca) sur un modèle de pente élastique linéaire homogène pour caractériser l'amplification de l'accélération le long de la surface située le long et à l’arrière de la crête d'une pente, et évaluer ainsi l'effet de la topographie sur l'amplification de l'accélération. L'interaction entre la fréquence du signal sismique appliqué à la base du modèle et l'angle et la hauteur de la pente a été particulièrement étudiée. Il a été constaté que le facteur d’amplification de l’accélération varie de manière significative avec l’angle de et la hauteur de la pente, la fréquence et la durée (nombre de cycle) du signal. De plus, l'amplification du mouvement du sol due à la topographie de la pente est influencée de manière significative par l'effet de couplage complexe entre les ondes incidentes et les ondes réfléchies sur la topographie, et qu’elle est fortement contrôlée par le rapport entre la longueur d'onde du signal incident et la hauteur de la pente.Les simulations numériques sont basées sur une étude géotechnique et une modélisation géomécanique nécessitant de valider les résultats par des comparaisons entre les résultats de modélisations et les données provenant des observations sur le terrain. L’analyse des domaines de fréquence, telle que la densité spectrale et la réponse en fréquence, est un moyen performant pour comprendre les caractéristiques des processus et les divers phénomènes qui ne peuvent pas être expliqués dans le domaine temporel. À cette fin, une étude de la crête du parc Xishan à Zigong au Sichuan en Chine a été réalisée. Les amplifications du site associées au mouvement du sol produit par le séisme de Wenchuan en 2008 ont été évaluées à l'aide de la technique du rapport spectral standard (SSR) et de la méthode d'accélération quadratique moyenne (arms) dans le domaine temporel. Une analyse numérique à 2D utilisant la méthode des différences finies au moyen du logiciel FLAC (Itasca) a été ensuite mise en œuvre et les résultats ont été comparés aux mesures de terrain. Les pics “simulés” des amplifications spectrales sont toujours inférieurs à ceux dérivés des enregistrements de terrain. L'effet d'atténuation important sur le mouvement d'entrée pour les hautes fréquences met en évidence le fait qu'un rapport d'amortissement du signal ne représente pas correctement la dissipation d'énergie dans les simulations numériques. Des amplifications significatives se sont produites à des fréquences élevées (> 10 Hz) et sont considérées comme résultant de conditions locales spécifiques telles que la fracturation des roches et les marches de faîtage; ils ne se produisent donc pas nécessairement au sommet de la colline.Enfin, des études paramétriques ont été réalisées avec des modèles élastiques en termes de diverses géométries de pente 2D et de couches géologiques de subsurface pour caractériser les amplifications du mouvement du sol. L’analyse paramétrique a pour but de comprendre le rôle joué par ces couches de surface, l’angle de la pente et la hauteur de la pente sur l’amplification du mouvement du sol,et donc d’évaluer si l’amplification du site peut être responsable du déclenchement du glissement de terrain. Ensuite, l'analyse dynamique sur des modèles de pente pour différentes magnitudes a été effectuée et une rupture de pente basée sur le déplacement a été créée pour évaluer la stabilité de la pente du site de Las Colinas au Salvador. Les résultats numériques ont clairement montré que les effets de site peuvent avoir induit d'importantes amplifications du mouvement du sol qui ont contribué à déclencher des glissements de terrain. / In this research work, numerical simulations using the finite difference FLAC software (Itasca) were first conducted with a homogeneous linear elastic slope model in order to characterize the acceleration amplification along the slope surface and behind the slope crest, and then to evaluate the topographic effect on the acceleration amplification. The interaction between the frequency of the seismic input motion applied at the base of the model with the slope angle and height has been deeply investigated. It was found that significant changes in the acceleration amplification factor result from variations in the slope angle and height as well as the signal frequency and duration. In addition, it has been shown that the ground motion amplification due to slope topography result from complex coupling effects between the input waves and the reflected waves on the topographic features and is highly controlled by the ratio between the wavelength of the input signal and the slope height.Numerical simulations are based on geotechnical investigations and geotechnical modeling, and it is necessary to validate the results through comparisons between modeling results and field observations. Frequency domain analysis such as spectral density and frequency response are an effective way to understand process characteristics and the various phenomena that cannot be explained in the time domain. For this purpose, a case study at Xishan Park ridge in Zigong in China has been studied. Site amplifications associated with the ground motion produced by the 2008 Wenchuan earthquake have been evaluated using the Standard Spectral Ratio (SSR) technique and root-mean-square acceleration (arms) method in time domain. 2D numerical analysis using finite difference method using the FLAC software (Itasca) has been then performed and results have been compared with monitoring data. The “simulated” peaks of the spectral amplifications are always lower than those derived from the field records. The strong attenuation of input motion at high frequencies highlights the shortcoming that a signal damping ratio does not adequately represent the energy dissipation in numerical simulations. Significant amplifications occurred at high frequencies (>10 Hz) and are considered to result from local specific conditions such as rock fracturing and ridge steps; thus they do not necessarily occur at the top of the hill. Finally, parametric studies were performed with elastic models in terms of various 2D slope geometries and geological layers to characterize the ground motion amplifications. The purpose of the parametric analysis is to understand the role of the geological layer, slope angle and slope height on the ground motion amplification, and thus to estimate if site amplifications could be responsible for the triggering of landslide. Then, the dynamic analysis on the slope model for different seismic magnitudes was performed and a slope failure based on displacement was created to evaluate the Las Colinas slope stability in Salvador. The numerical results clearly showed that site effects can have induced significant ground motion amplifications that contributed to trigger landslides.

While site effects are accounted for in most modern U.S. seismic design codes for building structures, there exist no standardized procedures for the computationally efficient integration of nonlinear ground response analyses in broadband ground motion simulations. In turn, the lack of a unified methodology affects the prediction accuracy of site-specific ground motion intensity measures, the evaluation of site amplification factors when broadband simulations are used for the development of hybrid attenuation relations and the estimation of inelastic structural performance when strong motion records are used as input in aseismic structural design procedures. In this study, a set of criteria is established, which quantifies how strong nonlinear effects are anticipated to manifest at a site by investigating the empirical relation between nonlinear soil response, soil properties, and ground motion characteristics. More specifically, the modeling variability and parametric uncertainty of nonlinear soil response predictions are studied, along with the uncertainty propagation of site response analyses to the estimation of inelastic structural performance. Due to the scarcity of design level ground motion recording, the geotechnical information at 24 downhole arrays is used and the profiles are subjected to broadband ground motion synthetics. For the modeling variability study, the site response models are validated against available downhole array observations. The site and ground motion parameters that govern the intensity of nonlinear effects are next identified, and an empirical relationship is established, which may be used to estimate to a first approximation the error introduced in ground motion predictions if nonlinear effects are not accounted for. The soil parameter uncertainty in site response predictions is next evaluated as a function of the same measures of soil properties and ground motion characteristics. It is shown that the effects of nonlinear soil property uncertainties on the ground-motion variability strongly depend on the seismic motion intensity, and this dependency is more pronounced for soft soil profiles. By contrast, the effects of velocity profile uncertainties are less intensity dependent and more sensitive to the velocity impedance in the near surface that governs the maximum site amplification. Finally, a series of bilinear single degree of freedom oscillators are subjected to the synthetic ground motions computed using the alternative soil models, and evaluate the consequent variability in structural response. Results show high bias and uncertainty of the inelastic structural displacement ratio predicted using the linear site response model for periods close to the fundamental period of the soil profile. The amount of bias and the period range where the structural performance uncertainty manifests are shown to be a function of both input motion and site parameters.