Thèse (M.Sc.T.) -- Université du Québec à Chicoutimi, 2005. / Bibliogr.: f. 124-134. Document électronique également accessible en format PDF. CaQCU

The styles of deformation of the fore-arc wedges along the Chilean convergent margin are observed to vary significantly, despite similar plate kinematic conditions. Here, I focus on the analysis of fore-arc deformation on two regions along the Chilean convergent margin at 20°-24°S and 37°-42°S. Although both regions are subjected to the oblique subduction of the oceanic Nazca plate and backstopped by the Andes mountain chain; they display different patterns of deformation. The northern Chilean study area (20° - 24°S) is characterized by an exceptionally thick crust of about 60 km beneath the Altiplano – Puna plateau, lack of an accretionary wedge in the fore-arc due to hyperarid climate, and consequently a sediment starved trench. Two major margin parallel strike slip faults are observed in this area, the Atacama Fault Zone (AFZ) and the Precordilleran Fault System (PFS). Both strike-slip faults do not exhibit significant recent displacement. The southern study area (37° - 42°S), compared to the northern study area, is characterized by lower topography, high precipitation rates (~2000 mm/yr), and a younger subducted oceanic plate. An active strike-slip fault, the Liquiñe-Ofqui-Fault-Zone (LOFZ), shows ~1 cm/yr recent dextral movement and shapes the surface of this area. Thus, the southern Chilean study area exhibits localized strike-slip motion. Within this area the largest earthquake ever recorded, the 1960 Valdivia earthquake, occurred with a moment magnitude of MW=9.5. I have constructed 2D thermal models and 3D mechanical models for both Chilean study areas to study processes related to active subduction. The applied numerical method is the finite element technique by means of the commercial software package ABAQUS. The thermal models are focused on the thermal conditions along the plate interface. The thermal structure along the plate interface reveals the limits of coupling but also the type of transition from coupled to uncoupled and vice versa. The model results show that shear heating at the plate interface is an important mechanism that should be taken into account. The models also show that the thermal condition at the downdip limit of the coupling zone leads to a sharp decrease of friction along the interface. Due to the different geometries of the two Chilean study areas, such as the slab dip and the thickness of the continental crust, the downdip limit of the southern study area is slightly shallower than that of the northern study area. The results of the 2D thermal models are used to constrain the spatial extent of the coupling zone in the 3D mechanical models. 3D numerical simulations are used to investigate how geometry, rheology and mechanical parameters influence strain partitioning and styles of deformation in the Chilean fore-arc. The general outline of the models is based on the fore-arc geometry and boundary conditions as derived from geophysical and geological field data. I examined the influence of different rheological approaches and varying physical properties of the fore-arc to identify and constrain the parameters controlling the difference in surface deformation between the northern and southern study area. The results of numerical studies demonstrate that a small slab dip, a high coefficient of basal friction, a high obliquity of convergence, and a high Young’s modulus favour localisation of deformation in the fore-arc wedge. This parameter study helped me to constrain preferred models for the two Chilean study areas that fit to first order observations. These preferred models explain the difference in styles of deformation as controlled by the angle of obliquity, the dip of subducting slab, and the strength of wedge material. The difference in styles can be even larger if I apply stronger coupling between plates within the southern area; however, several independent observations indicate opposite tendency showing southward decrease of intensity of coupling. The weaker wedge material of the preferred model for the northern study area is associated with advanced development of the adjacent orogen, the Central Andes. Analysis of world-wide examples of oblique subduction zones supports the conclusion that more mature subduction zones demonstrate less pronounced localization of strike-slip motion. / Die Deformationsmuster der Fore-Arc Keile entlang des chilenischen konvergenten Plattenrandes variieren beachtlich, trotz ähnlicher plattenkinematischer Randbedingungen. In dieser Arbeit konzentriere ich mich auf die Analyse der Deformation des Fore-Arcs in zwei Gebieten entlang des chilenischen konvergenten Plattenrandes zwischen 20°-24°S und 37°-42°S. Obwohl beide Gebiete durch schiefe Subduktion der ozeanischen Nazca Platte und der östlichen Begrenzung durch die Andine Gebirgskette gekennzeichnet sind, zeigen sie unterschiedliche Deformationsmuster an der Oberfläche. Das nördliche chilenische Gebiet (20° - 24°S) ist gekennzeichnet durch eine außergewöhnliche Krustendicke von ~ 60 km unterhalb des Altiplano - Puna Plateaus, dem Fehlen eines akkretionären Prismas im Fore-Arc aufgrund des trockenen Klimas und somit einer nahezu sedimentfreien Tiefseerinne. Zwei große Plattenrand-parallele Strike-Slip Störungen werden in diesem Gebiet beobachtet, die Atacama Fault Zone (AFZ) und das Precordilleran Fault System (PFS). Beide Strike-Slip Störungen zeigen keine signifikanten aktuellen Bewegungsraten. Das südliche Gebiet (37° - 42°S) ist im Vergleich zum nördlichen Gebiet durch eine niedrigere Topographie, hohe Niederschlagsraten (~2000 mm/a) und eine jüngere abtauchende ozeanische Platte gekennzeichnet. Die aktive Strike-Slip Störung, Liquiñe-Ofqui-Fault-Zone (LOFZ), ist gekennzeichnet durch aktuelle dextrale Bewegungsraten von 1 cm/a und prägt die Oberflächenstruktur in dieser Region. Folglich ist der südliche Arbeitsbereich durch lokalisierte Strike-Slip Bewegung charakterisiert. Innerhalb dieses Gebietes ereignete sich das größte instrumentell aufgezeichnete Erdbeben, das 1960 Valdivia Erdbeben, mit einer Stärke von MW=9.5. 2D thermische Modelle und 3D mechanische Modelle wurden für die beiden chilenischen Gebiete konstruiert, um Prozesse im Zusammenhang mit aktiver Subduktion zu untersuchen. Als numerisches Verfahren wurde die Finite Elemente Methode mit Hilfe des kommerziellen Softwarepakets ABAQUS angewandt. Die thermischen Modelle sind auf die thermischen Konditionen entlang der Plattengrenzfläche fokussiert. Die thermische Struktur entlang der Plattengrenzfläche zeigt die Grenzen der Kopplung an aber auch die Art des Überganges von gekoppelt zu nicht gekoppelt und umgekehrt. Die Modellergebnisse zeigen, dass Heizen infolge der Scherung an der Plattengrenzfläche ein wichtiger Faktor ist, der in Betracht gezogen werden sollte. Die Modelle zeigen auch, dass die thermische Struktur an der unteren Begrenzung der Koppelzone zu einer deutlichen Abnahme der Reibung entlang der Grenzfläche führt. Aufgrund der unterschiedlichen Geometrien der zwei chilenischen Untersuchungsgebiete, z.B. Abtauchwinkel der ozeanischen Platte und Krustendicke, ist die untere Begrenzung der Koppelzone des südlichen Untersuchungsgebietes in geringerer Tiefe als die des nördlichen Gebietes. Die Ergebnisse der thermischen 2D Modelle werden genutzt, um die räumliche Ausdehnung der Koppelzone in den mechanischen 3D Modellen festzulegen. Numerische 3D Simulationen werden genutzt, um zu verstehen, wie Geometrien, Rheologien und mechanische Parameter die Verformungspartitionierung und das Deformationsmuster im chilenischen Fore-Arc beeinflussen. Ich habe den Einfluss unterschiedlicher rheologischer Ansätze und unterschiedlicher physikalischer Eigenschaften auf den Fore-Arc untersucht, um Parameter zu identifizieren und zu bestimmen, die den Unterschied des Deformationsmusters zwischen dem nördlichen und südlichen Gebiet steuern. Die Ergebnisse der numerischen Studien stellen heraus, dass ein kleinerer Abtauchwinkel der ozeanischen Platte, ein hoher basaler Reibungskoeffizient, eine hohe Konvergenzschiefe und ein großer Elastizitätsmodul die Lokalisierung der Deformation im Fore-Arc Keil begünstigen. Basierend auf dieser Parameterstudie habe ich Modelle für die beiden chilenischen Gebiete ausgewählt, die in Beobachtungen erster Ordnung übereinstimmen. Diese ausgewählten Modelle erklären die unterschiedlichen Deformationsmuster durch eine größere Konvergenzschiefe, einen kleineren Abtauchwinkel der ozeanischen Platte und ein härteres Keilmaterial für das südliche Untersuchungsgebiet. Der Unterschied bezüglich der Deformationsmuster kann sogar größer sein, wenn ich eine größere Reibung zwischen den Platten im südlichen Gebiet anwende; jedoch zeigen einige unabhängige Beobachtungen eine umgekehrte Tendenz: eine Abnahme der Intensität der Koppelung von Nord nach Süd. Das schwächere Keilmaterial des ausgewählten Modells für das nördliche Untersuchungsgebiet steht im Zusammenhang mit der fortgeschrittenen Entwicklung des angrenzenden Orogens, der zentralen Anden. Die Analyse weltweiter Beispiele von schiefen Subduktionzonen unterstützt die Schlussfolgerung, dass ältere Subduktionzonen weniger ausgeprägte Lokalisierung von Strike-Slip Bewegung aufzeigen.

The spatial patterns of bimaterial interfaces along the Parkfield section of the San Andreas Fault (SAF) and central section of the Calaveras Fault are systematically investigated with large data sets of near-fault waveforms. Different from the usage of direct P and S waves in traditional tomographic studies, a particular seismic phase named fault zone head wave (FZHW) is used to image the bimaterial fault interfaces. The results show clear variations of seismic velocities contrast both along-strike and along-depth directions in both regions, which is in general consistent with local geological setting at surface and existing 3D tomography results. In the Parkfield section of SAF, the result of velocity contrast is used to test the relationship between preferred rupture directions of M6 Parkfield earthquakes and bimaterial interface. Strong velocity contrast (~5-10%) near Middle Mountain (MM) could control the rupture directions of nearby earthquakes to SE, such as the case for 1966 M6 Parkfield earthquake. In comparison, weak velocity contrast (~0-2%) near the epicenter of the 2004 Parkfield M6 earthquake (i.e., Gold Hill) probably has no influence on controlling its rupture direction, which is consistent with the bilateral rupture of the 2004 Parkfield earthquake. In the central Calaveras Fault, a detailed analysis of the moveout between FZHWs and direct P waves revealed the existence of a complicated fault structure with velocity contrast increasing from NW to SE of station CCO. The high velocity contrast SE of station CCO could be caused by a low-velocity zone SE of station CCO. The spatio-temporal variations of seismic velocity around the central Calaveras Fault and its nearby region are investigated based on the waveform analysis of 333 repeating clusters following the 1984 ML6.2 Morgan Hill earthquake. Clear reduction of seismic velocity is shown for all repeating clusters immediately after the mainshock, followed by a logarithmic recovery. The coseismic change mostly occurs at shallow layers (top few hundred meters) for the region away from the rupture area of the mainshock, but extends much deeper around the rupture zone of the Morgan Hill earthquake. The estimated depth of the damage zone is up to 6 km in the fault based on the repeating clusters directly beneath station CCO. Finally, temporal changes around the Parkfield section of SAF are studied using recently developed ambient noise cross-correlation technique. The extracted daily empirical Green functions (EGFs) from 0.4-1.3 Hz noise records are used to estimate subtle temporal changes associated with large earthquakes from local to teleseismic distances. The results show clear coseismic reduction of seismic velocities after the 2004 M6 Parkfield earthquake, similar to the previous observation based on repeating earthquakes. However, no systematic changes have been detected for other four regional/teleseismic events that have triggered clear tremor activity in the same region. These results suggest that temporal changes associated with distance sources are very subtle or localized so that they could not be detected within the resolution of the current technique (~0.2%).

The Valle de Catacamas forms a major basin along the central portion of the Guayape fault, the most prominent tectonic element of the Chortís block. The Guayape fault extends 290 km southwest from the Caribbean coast to the region of El Paraíso, Honduras, and may continue to the Pacific coast along a related prominent topographic feature, the Choluteca linear. Basins presently forming along the Guayape fault indicate that the fault is currently experiencing right-slip. The active features of the Valle de Catacamas displace older folds and reverse faults which apparently formed during an earlier period of sinistral shear. Thus, the Guayape fault has undergone at least two phases of movement, post-Cenomanian left-slip followed by the present right-slip. The geology of the valley suggests multiple stages of evolution. These include at least one period of thrust and reverse faulting, possibly associated with sinistral shear along the Guayape fault, and a recent episode of normal faulting associated with dextral shear on the Guayape fault. Thrusting of basement rocks over Jurassic strata on the south side of the valley was the earliest deformation to affect Mesozoic or Cenozoic rocks. The event can only be dated as post-Jurassic in age. The Cretaceous rocks of the Sierra de Agalta on the north side of the Valle de Catacamas are much more strongly deformed than similar rocks in central Honduras. In this range, the Aptian-Albian Atima Limestone commonly has a pervasive pressure solution cleavage which has not been reported from other locations on the Chortís block. The cleavage is apparently axial planar to the folds. The age of this deformation is constrained only as post-Cenomanian. SIR data indicate that these folds are deflected in sinistral shear near the Guayape fault. In addition, a major structural contact has a large left-lateral separation. The folds in the Sierra de Agalta are cut by the range-bounding normal fault of the Sierra de Agalta. Younger rocks are placed on older rocks by this normal fault, and fault slip data from small fault planes in the footwall block indicate normal faulting. The N 65° E strike of this normal fault, the N 35° E strike of the Guayape fault, and stress orientations inferred from fault slip data indicate that the present movement on the Guayape fault is right-slip. Fault slip data from the Guayape fault zone is heterogeneous as would be expected if two stage slip has occurred. / text

L’objectif principal de cette thèse est de tirer de données LiDAR de télédétection à très haute résolution afin d’extraire une partie du traces tectono-géomorphiques imprimées dans la morphologie de grands tremblements de terre préhistoriques. Les informations consultées dans ces traces constituent l'historique des glissements cumulés de grands tremblements paléoséismique successifs le long d'une faille donnée. L'historique des glissements permet de déterminer le nombre d'événements et les glissements les plus importants produits par ces événements. La connaissance des plus grandes glissades produites par des grands séismes historiques et préhistoriques permettra de déduire l'ampleur potentielle des événements futurs. La caractérisation de la distribution du glissement superficiel fournit des informations importantes sur la mécanique des failles, les contrôles de la propagation de la rupture et la répétabilité de la rupture à certains points le long de la faille. Cependant, la caractérisation et la mesure correctes de la distribution des glissements à partir de formes de relief géomorphologiques déplacées par tectonisme sont accompagnées d'incertitudes considérables, résultant principalement de processus d'érosion et de dépôt. Ces incertitudes pourraient entraîner à la fois une sous-estimation et une surestimation du glissement, ainsi que des résultats contradictoires issus d'enquêtes différentes sur le même défaut. Par conséquent, nous avons développé une nouvelle technique basée sur MATLAB, 3D_Fault_Offsets, pour caractériser mathématiquement, et donc automatiquement, la géométrie 3D de marqueurs géomorphiques décalés (définie par 9 entités géométriques situées de part et d'autre de la faille), puis calculer composants latéraux et verticaux du glissement. Nous estimons que les incertitudes générées par cette technique définissent mieux la gamme des "véritables" compensations potentielles par rapport aux incertitudes plus libérales proposées dans d’autres études, pourtant ils se révèlent assez volumineux. Après vérification de l'efficacité du code en mesurant à nouveau 3 ensembles de données paléosismiques, nous avons l’appliqué à une faille de décrochement qui était historiquement capable d'un séisme de chute de contrainte importante (MW ~ 8,2 en 1855), la faille de Wairarapa. Nous avons identifié et analysé un total d'environ 700 marqueurs géomorphiques déplacés le long d'une zone de données LiDAR de 70 km, ce qui en fait l'un des ensembles de données paléosismiques les plus vastes et les plus denses. Les décalages latéraux mesurés vont de quelques mètres à environ 800 m, mais la majorité d'entre eux sont inférieurs à 80 m, ce qui permet d'examiner les plus récents glissements de faille latéraux. Les décalages verticaux varient entre 0 et ~ 30 m et suggèrent des rapports de glissement vertical / latéral généralement compris entre 10 et 20%. Nous avons effectué les analyses statistiques de la collection dense de décalages mesurés séparément le long des principaux segments successifs qui constituent l'étendue de la faille étudiée. Dans la plupart des segments, cette analyse a révélé la présence de 6 à 7 amas décalés dans la plage allant de 0 à 80 m, suggérant la rupture de la faille de Wairarapa lors de 6 à 7 grands séismes précédents. Les plus grandes glissades que nous déduisons pour ces tremblements de terre passés sont importantes, la plupart dans la plage 7-15 m. Chaque glissement sismique semble varier le long de la faille et généralement plus grand dans sa partie sud. La faille de Wairarapa a ainsi provoqué à plusieurs reprises d'importants séismes dus à la chute de contraintes au cours de la période préhistorique, ce qui souligne le risque sismique élevé qu'elle pose dans le sud de la Nouvelle-Zélande. Par conséquent, l'utilisation de notre nouveau code 3D_Fault_Offsets avec des données topographiques à haute résolution telles que LIDAR peut permettre de mieux évaluer le comportement futur des failles sismogènes. / The main scope of this PhD thesis is to utilize very high-resolution remote sensing LiDAR data to extract some of the tectono-geomorphic traces imprinted in the morphology from large prehistoric earthquakes. The information that is accessed in these traces is the cumulative slip history of successive large paleoearthquakes along a given fault. The slip history allows the determination of the number of events and the largest slips produced by those respective events. The knowledge of the largest slips produced by historic and prehistoric large earthquakes will enable some inference into the potential magnitude of future events. Characterizing the distribution of surface slip provides important insights into fault mechanics, controls on rupture propagation, and repeatability of rupture at certain points along the fault. However, properly characterizing and measuring the slip distribution from tectonically-displaced geomorphic landforms comes with considerable uncertainties mostly resulting from erosion and depositional processes. These uncertainties could lead to both underestimation and overestimation of the slip, and to conflicting results from different surveys of the same fault. Therefore, we have developed a new MATLAB-based technique, 3D_Fault_Offsets, to mathematically, and hence automatically, characterize the 3D geometry of offset geomorphic markers (defined by 9 geometric features either side of the fault), and then calculate the lateral and vertical components of slip. We believe that the uncertainties obtained from this technique better define the range of potential ‘true’ offsets compared to more liberal uncertainties offered in other studies, yet they reveal to be fairly large. Upon verification of the code efficacy by successfully re-measuring 3 paleoseismic datasets, we applied it to a strike-slip fault in New Zealand that was historically capable of a large stress drop earthquake (MW~8.2 in 1855), the Wairarapa fault. We identified and analyzed a total of ~700 displaced geomorphic markers along a 70-km stretch of LiDAR data, making this one of the largest and densest paleoseismic datasets. Measured lateral offsets range from a few meters to about 800 m, but the majority are lower than 80 m, providing the means to examine the most recent lateral fault slips. The vertical offsets range between 0 and ~30 m, and suggest vertical to lateral slip ratios commonly in the range 10-20%. We conducted the statistical analyses of the dense collection of measured offsets separately along the successive major segments that form the investigated fault stretch. In most segments, this analysis revealed 6-7 offset clusters in the range 0-80 m, suggesting the Wairarapa fault ruptured in 6-7 previous large earthquakes. The largest slips we infer for these past earthquakes are large, most in the range 7-15 m. Each earthquake slip seems to vary along the fault length, and be generally greater in its southern part. The Wairarapa fault has thus repeatedly produced large stress drop earthquakes in prehistoric time, which emphasizes the elevated seismic hazard it poses in Southern New Zealand. Therefore, the use of our new code 3D_Fault_Offsets with high resolution topographic data such as LIDAR can lead to better assessments of future behavior of seismogenic faults.

The surface expression of upper crustal deformation varies widely based on geologic settings. Normal faults within an intra-arc basin, strike-slip faulting within a wide shear zone, and creeping fault behavior all manifest differently and require a variety of techniques for analysis. In this dissertation I studied three different actively deforming regions across a variety of geologic settings. First, I explored the drivers of extension within the La Pine graben in the Oregon Cascades. I mapped >20 new Quaternary faults and conducted paleoseismic trenching, where I found evidence for a mid-late Holocene earthquake on the Twin Lakes maar fault. I suggest that tectonics and not volcanism is responsible for the most recent deformation in the region based on fault geometries and earthquake timings, although more research is needed to tease out finer temporal and genetic relationships between tectonics and volcanism regionally. Second, I investigated the rupture pattern and earthquake history of the Calico fault system in the Eastern California Shear Zone. We mapped ~18 km of continuous rupture, with a mean offset of 2.3 m based on 39 field measurements. We also found evidence for two earthquakes, 0.5 - 1.7 ka and 5.5 - 6.6 ka through paleoseismic trenching. We develop a number of different multifault rupture scenarios using our rupture mapping and rupture scaling relationships to conduct Coulomb stress change modeling for the most recent earthquake on the Calico fault system. We find that the most recent event places regions adjacent to the fault in a stress shadow and may have both delayed the historic Landers and Hector Mine ruptures and prevented triggering of the Calico fault system during those events. Last, I studied the spatial distribution of the southern transition zone of the creeping section of the San Andreas fault at Parkfield, CA to determine if it shifted in response to the M6 2004 Parkfield earthquake. I used an Iterative Closest Point algorithm to find the displacement between two lidar datasets acquired 13 years apart. I compared creep rates measured before the 2004 earthquake to creep rates calculated from my lidar displacement results and found that there is not a discernible change in the overall pattern or distribution of creep as a response to the 2004 earthquake. Peaks within the lidar displacement results indicate complexity in the geometry of fault locking. / Doctor of Philosophy / Fault behavior varies widely across different regions, depending on the type of fault and local geology. In this dissertation I examine three regions with different mechanisms controlling deformation within them. First, I study the relationship between volcanic and tectonic induced faulting in the La Pine graben in the Oregon Cascades. While volcanoes and tectonics can both produce faults within a region, the surface expression of those faults changes depending on the underlying driver. I map > 20 new faults in the La Pine graben. I also conduct paleoseismic trenching on one of the newly identified faults, the Twin Lakes maar fault, and find that its most recent rupture occurred < 7.6 ka. I conclude that tectonism is the dominant driver of faulting within the La Pine graben based on the fault geometries and timing between identified regional earthquakes and volcanism. Second, I explore recent rupture on the Calico fault system in the Eastern California Shear Zone, which is a wide region across eastern California where deformation is distributed among many faults. Faulting in this region is complex, with some earthquakes occurring on multiple connected faults. I conducted a paleoseismic survey to determine the timing of the most recent earthquake(s) on the Calico fault system. This trenching effort found evidence for 1-2 earthquakes, the most recent occurring 0.5 – 1.7 ka. I use the rupture mapping and earthquake timing to develop a number of various rupture scenarios. I use these scenarios as inputs for computer modeling to explore the regional stress changes from these events and find that they reduce the overall stress in the area, elongating the amount of time between regional earthquakes. Last, I examine how creeping fault behavior on the San Andreas fault near Parkfield, CA changes as a response to an earthquake. Creeping behavior is where the two sides of a fault are continuously moving past one another. I examine the spatial distribution of where the San Andreas fault transitions from creeping to locked behavior by differencing two high-resolution lidar topographic datasets taken after the M6 2004 Parkfield earthquake. I compare my displacement results to pre-2004 datasets and conclude that the transition zone did not appreciably change as a result of the earthquake.

Riedel shear zones are geometric fault patterns commonly associated with strike-slip fault systems. The progressive evolution of natural Riedel shear zones within the Navajo Sandstone of southern Utah is interpreted from the spatial evolution of small-scale, incipient Proto-Riedel Zones (PRZs) to better-developed Riedel shear zones using field mapping and three-dimensional digital modeling. PRZs nucleate as a tabular zone of localized shearing marked by en èchelon deformation bands, each of which is no more than a few mm wide and tens of cm long, and oriented at 55° - 85° to the trend of the zone. With increasing strain, deformation bands and sedimentary markers are sheared ductily through granular flow and assume a sigmoidal form. The temporal and spatial evolution of bands comprising a Riedel shear zone suggests that PRZs nucleate as transitional-compactional deformation bands under localized, supra-lithostatic fluid pressure. Subsequent bands develop under modified regional stresses as conjugate shear fractures within the strain- hardened axis of the PRZ. These antithetic driven systems are not compatible with traditional synthetic driven models of Riedel shear zones. Unlike most synthetic driven examples, these antithetic driven systems are not controlled by preexisting "basement" structures, thus their geometries reflect a primary propagation or secondary passive deformation mechanism.

Ce manuscrit concerne l'analyse tectonique et géomorphologique du système de failles de Longriba (LFS), localisé à l'Est du plateau tibétain à environ 200 km au Nord-ouest de la chaîne des Longmen Shan. Le LFS est constitué de deux zones de failles décrochantes dextres, parallèles et d'orientation N55°E : la faille de Longriqu, au Nord, et la faille de Maoergai, au Sud. Le rôle géodynamique de ce système est primordial puisqu'il accommode 5 ± 1 mm/an de la composante décrochante induite par la convergence oblique du bloc Aba, elle-même liée à la collision Inde-Asie. De plus, le LFS partitionne la déformation de la marge Est tibétaine; les structures des Longmen Shan étant essentiellement chevauchantes. Cependant l'histoire long-terme du LFS est mal contrainte. L'objectif de cette thèse est donc d'apporter des informations spatio-temporelles sur l'activité du système à l'échelle du Cénozoïque. Pour cela trois études ont été réalisées. La première a permis de mieux contraindre le comportement sismogénique des deux zones de failles à l'aide d'images satellites de basse (90m) et très haute résolution (50cm). L'analyse des déplacements cumulés le long de la faille de Maoergai a permis de proposer que celle-ci était active vers ~15Ma. La seconde étude a mis en évidence un contrôle du système sur la répartition des taux de dénudation au travers de la marge Est tibétaine à l'échelle de l'Holocène. Enfin, les données de thermochronologie basse température suggèrent que la faille de Maoergai a accommodé un mouvement vertical vers ~10Ma. Ce mouvement est probablement lié au rebond isostatique de la marge, découplé du mouvement décrochant qu'elle accommode à la même période. / This manuscript concerns the tectonic and geomorphic analysis of the Longriba fault system (LFS), located in the eastern Tibetan plateau at about 200 km north-west of the Longmen Shan. The LFS consists of two dextral strike-slip fault zones, parallel and N55 °-trending: the Longriqu fault to the north and the Maoergai fault the south. The geodynamic role of the system is essential as accommodates 5 ± 1 mm / year of the slip component induced by the oblique convergence of the Aba block, itself linked to the India-Asia collision. In addition, the LFS partitions the deformation of the east Tibetan margin; the structures of the Longmen Shan are mainly thrust faults. However the long-term history of LFS is poorly constrained. The objective of this thesis is to bring spatial and temporal information on system activity throughout the Cenozoic. To do so, three studies were performed. The first led to better constrain the seismogenic behavior of the two fault zones with low resolution (90m) and very high resolution satellite images (50cm). The analysis of cumulative displacements along the Maoergai fault allowed to propose that it was active at ~ 15 Ma. The second study showed that the system controls the distribution of the denudation rates over the EastTibetan margin throughout the Holocene. Finally, the low-temperature thermochronology data suggest that, in particular, the Maoergai fault has accommodated a vertical movement at ~ 10 Ma. This movement is probably related to the isostatic rebound of the margin, decoupled from the strike-slip movement it accommodates at the same period.