On propose traditionnellement que les panaches mantelliques jouent un rôle important dans l'amincissement de la lithosphère. Des données sismologiques sous Hawaïi et Cape Verde suggèrent une limite lithosphère-asthénosphère (LAB) jusqu'à 50 km plus superficielle qu'autour. Des modèles numériques ont montré, en effet, qu'une convection à petite échelle (SSC, pour small-scale convection) dans la couche à faible viscosité formée à la base de la lithosphère par l'accumulation de la matière des panaches peut être un mécanisme efficace d'érosion du manteau lithosphérique. Cependant, ces modèles montrent que, si la plaque se déplace, l'érosion thermo-mécanique de la lithosphère ne dépasse pas 30 km. Afin de mieux étudier les interactions panache/lithosphère, et d'ainsi caractériser les paramètres contrôlant cette érosion, nous avons effectué des simulations numériques en 2D qui utilisent un modèle pétro-thermomécanique basé sur des approches en différences finies associées à des marqueurs actifs. Nous avons focalisé sur : (1) la dynamique de la SSC dans la couche à faible viscosité formée par étalement du panache à la base de la lithosphère et (2) l'effet de la fusion partielle sur cette dynamique. La plaque lithosphérique et le manteau sous-jacent sont caractérisés par une composition péridotitique homogène à viscosité newtonienne dépendante de la température et de la pression. Une vitesse constante, comprise entre 5 et 12,5 cm/an, est imposée au sommet de la plaque. Les panaches sont créés en imposant une anomalie thermique de 150 à 350 K en base du modèle (700 km de profondeur). La fusion partielle est calculée à partir d'un paramétrization des solidus et liquidus pour la fusion anhydre des péridotites. Nous modélisons la déplétion de la péridotite et son effet sur la fusion partielle en supposant que le degré de fusion ne peut qu'augmenter au cours du temps. Le liquide est accumulé jusqu'à un seuil et la masse fondue en excès est extraite instantanément. La rhéologie de la péridotite partiellement fondue est déterminée utilisant une constitutive relation basée sur un modèle de contiguïté, qui permet de prendre en compte les effets de la distribution de matière liquide à l'échelle de grain. La densité varie en fonction du degré de fusion partielle et de la déplétion du résidu solide. Nous analysons la cinématique du panache lors de son interaction avec une plaque mobile, la dynamique de la convection à petite-échelle (SSC) et le rajeunissement thermique de la lithosphère qui en résulte. Le temps de démarrage et la vigueur de la SSC et, par conséquent, le nouvel état d'équilibre thermique de la lithosphère à l'aplomb du panache dépendent du nombre de Rayleigh (Ra) dans la couche instable à la base de la lithosphère, qui est contrôlé par l'anomalie de température et la rhéologie dans cette couche. Pour des panaches chauds et vigoureux, le démarrage de la SSC ne dépend pas de la vitesse de la plaque. Pour des panaches plus faibles, le temps de démarrage diminue avec l'augmentation de la vitesse de la plaque. Ce comportement est expliqué par une différence dans la structure thermique de la lithosphère, due à des échanges diffusifs à la base lithosphère plus efficaces pour des panaches lents. La diminution de la viscosité associée à la présence de magma et la diminution de la densité du résidu solide accélèrent le démarrage et accroissent la vigueur de la SSC, entraînant une érosion plus efficace et plus proche du point d'impact de panache sous la lithosphère. / Mantle plumes are traditionally proposed to play an important role in thinning the lithosphere. Seismic images beneath Hawaii and Cape Verde, for instance, show a lithosphere-asthenosphere boundary (LAB) up to 50 km shallower than the surroundings. However, previous numerical modeling of plume-lithosphere interaction implies that unless the plate is stationary the thermo-mechanical erosion of the lithosphere does not exceed 30 km. We used 2D petrological-thermo-mechanical numerical models based on a finite-difference method on a staggered grid and marker in cell method to further study the plume-lithosphere interaction. We focused on: (1) analyzing the dynamics of the small-scale convection (SSC) in the plume wake as a function of the plume vigor and plate velocity and (2) quantifying the effect of partial melting on this SSC. A homogeneous peridotite composition with a Newtonian temperature- and pressure-dependent viscosity is used to simulate both the plate and the convective mantle. A constant velocity, ranging from 5 to 12.5 cm/yr, is imposed at the top of the plate. Plumes are created by imposing a thermal anomaly of 150 to 350 K on a 50 km wide domain at the base of the model (700 km depth); the plate right above the thermal anomaly is 40 Myr old. Partial melting is modeled using the batch-melting solidus and liquidus in anhydrous conditions. We model the progressive depletion of peridotite and its effect on partial melting by assuming that the melting degree only strictly increases through time. Melt is accumulated until a porosity threshold is reached and the excess melt is instantaneously extracted. The rheology of the partially molten peridotite is determined using a viscous constitutive relationship based on a contiguity model, which enables to take into account the effects of grain-scale melt distribution. The density varies as a function of the melt fraction and of the depletion of the residue. We analyze the kinematics of the plume as it impacts a moving plate, the dynamics of time-dependent small-scale convection (SSC) instabilities developing in the low-viscosity layer formed by spreading of hot plume material at the lithosphere base, and the resulting thermal rejuvenation of the lithosphere. The onset time and the vigor of SSC and, hence, the new equilibrium thermal state of the lithosphere atop the plume wake depends on the Rayleigh number (Ra) in the unstable layer at the base of the lithosphere, which is controlled by the temperature anomaly and rheology in the plume-fed layer. For vigorous, hot plumes, SSC onset times do not depend on plate velocity. For more sluggish plumes, SSC onset times decrease with increasing plate velocity. This behavior is explained by differences in the thermal structure of the lithosphere, due to variations in the spreading behavior of the plume material at the lithosphere base. Reduction of the viscosity in partial molten domains and decrease in density of the depleted residuum accelerate and enhance the vigor of small-scale convection in the plume-fed low-viscosity layer at the lithosphere base. It also reduces SSC onset times, leading to more effective erosion closer to the plume-lithosphere impact.

Pour mieux comprendre le processus géodynamique qui a permis la destruction du craton Nord Chinois (NCC), le rôle des fluides mantelliques a été examiné. Pour cela, les distributions des teneurs en eau et des compositions isotopiques du Lithium dans le manteau lithosphérique NCC ont été déterminées à partir des xénolites de péridotite entrainés par les basaltes mésozoïques et cénozoïques. Une variation temporelle des teneurs en eau est observée. Le manteau lithosphérique cénozoïque est appauvri en eau, sans doute suite à l'amincissement crustal et au réchauffement du manteau résiduel par un flux ascendant asthénosphérique. Le manteau lithosphérique mésozoïque montre des teneurs en eau intermédiaire entre les teneurs élevées du Crétacé et les teneurs basses cénozoïques, indiquant une déshydratation du manteau commençant dès le début de sa destruction. Cette déshydratation, facilitée par la destruction du manteau lithosphérique profond, permet de renforcer la rigidité de la lithosphère et lui permet de résister à la convection mantellique. Les distributions élémentaire et isotopique du Li montrent une grande hétérogénéité, aux échelles intra et inter-cristallines. Par simulation numérique, nous démontrons que deux enrichissements successifs ont affecté le manteau, un enrichissement limité (<5ppm) avec une signature pauvre en 7Li ([delta]7Li ~ -20 [pour mille]), suivi d'un enrichissement important (> 100 ppm) avec une signature riche ([delta]7Li ~ +20 [pour mille]), précédent de peu l'exhumation des xénolites. La formation des liquides métasomatiques responsables de ces enrichissements nécessite une distribution hétérogène dans le manteau NCC d'éléments recyclés lors de la subduction à l'est du NCC / In order to investigate the geodynamic cause for destruction of the North China Craton (NCC), the role of mantle fluids is examined. The aim of the PHD work is to clarify H2O contents and lithium isotopic compositions of the NCC lithospheric mantle by studying peridotite xenoliths hosted by Mesozoic-Cenozoic basalts across eastern NCC. A temporal variation of H2O content has been revealed, and it has deep implications for processes of craton destruction. The Cenozoic lithospheric mantle was featured by low H2O content, interpreted to be the relict mantle that survived the lithospheric thinning and has been dewatered by reheating from upwelling asthenospheric flow. The late-Mesozoic lithospheric mantle showed relatively high H2O content, a hydrous status intermediate between the Cretaceous hydration and the Cenozoic dryness, indicating the dehydration of the NCC mantle with time during NCC destruction. The dehydration, facilitated by thinning of weak mantle pieces at bottom, is one way by which the lithosphere strengthens itself to survive in the convecting mantle. Extreme Li and isotopic disequilibria were observed intra- and inter-mineral in the peridotites. With numerical simulations, we demonstrate two superimposed Li enrichment events occurring at the mantle: a limited Li enrichment (< 5 ppm) and large delta7Li depletion (-20~-10[per 1000]) of the mantle domain, followed by a recent and transient infiltration of high Li and delta7Li (up to +20 [per 1000]) melts/fluids. The anomalous Li isotopic compositions of mantle metasomatic agents call upon the same of their mantle sources, and we assume recycled components, both Li isotopically heavy and light, in the mantle beneath the eastern NCC

Nós usamos tomografia sísmica de tempo de percurso para estudar o manto superior sob as regiões Sudeste e Centro-Oeste do Brasil. Este método baseia-se na inversão de resíduos relativos de tempo para as ondas P e S (VanDecar, 1991), que foram obtidos para mais de 80 estações em uma área de 20 x 20 graus. Mais de 11000 e 8000 resíduos de tempo foram obtidos para as ondas P (P direta e PKPdf) e S (S direta, ScS, SKS e SKKS), respectivamente, utilizando correlação cruzada de forma de ondas para até 12 estações operando simultaneamente. Para avaliar a robustez dos resultados com respeito aos dados, nós utilizamos o método estatístico de re-amostragem Jackknife, o qual inerentemente leva em conta a altamente variável cobertura dos raios e os erros das medidas, e pode fornecer limites de confiança para as anomalias. Inversões regionais foram realizadas para estudar a influência da parametrização nas anomalias sísmicas. Nossos resultados mostraram boa correlação das anomalias sísmicas com as principais estruturas tectônicas e revelaram novas anomalias que ainda não haviam sido observadas nos trabalhos anteriores. Anomalias de alta velocidade na porção oeste do Cráton do São Francisco apóiam a hipótese de que este cráton foi parte de uma placa Neoproterozóica maior. Anomalias de baixa velocidade sob a Província Tocantins (principalmente nas faixas móveis entre os Crátons Amazônico e do São Francisco) foram interpretadas como causadas por afinamento litosférico, consistente com a boa correlação entre a sismicidade intraplaca e as anomalias de baixa velocidade nesta região (Assumpção et al., 2004b). A melhora na resolução da anomalia de alta velocidade sob a Bacia do Paraná (~200 km) permitiu uma discussão sobre a geometria do núcleo cratônico desta Bacia. A subducção da Placa de Nazca foi observada como uma anomalia de alta velocidade sob a Bacia do Paraná (profundidades entre 700 e 1200 km). Nestas profundidades, uma grande anomalia de baixa velocidade aparece próxima da Placa de Nazca. Testes sintéticos mostraram que esta anomalia é um artefato da inversão gerado pela presença da Placa de Nazca. / We used travel time seismic tomography to study the upper mantle beneath SE and Central Brazil. This method is based on the inversion of P- and S-wave relative travel time residuals (VanDecar, 1991) obtained from more than 80 stations in an area of 20 x 20 degrees. More than 11000 P and PKP residuals, and more than 8000 S, ScS, SKS, and SKKS residuals were obtained from waveform cross-correlations for up to 12 simultaneous stations. To evaluate the robustness of the tomographic results with respect to the data, we use the Jackknife re-sampling method, which inherently take into account the highly variable ray coverage and measurement errors, and can provide confidence limits for the anomalies. Regional inversions were carried out to study influence of the parameterization on the seismic anomalies. Our results show correlations of seismic anomalies with the main tectonic structures and reveal new anomalies not yet observed in previous works. High velocity anomalies in the western portion of the São Francisco Craton support the hypothesis that this craton was part of a major Neoproterozoic plate. Low velocity anomalies beneath the Tocantins Province (mainly fold belts between the Amazon and São Francisco cratons) are interpreted as due to lithospheric thinning, consistent with a good correlation between intraplate seismicity and low velocity anomalies in this region (Assumpção et al., 2004b). The resolution improvement of the high velocity anomalies beneath the Parana Basin (at ~200 km) allows a discussion about the geometry of the cratonic nucleous of this Basin. The slab of the Nazca Plate is observed as a high velocity anomaly beneath the Parana basin (at 700-1200 km depth). At these depths, large low velocity anomaly appears next to the slab. Synthetic tests show that these anomalies are an artifact of the inversion generated by the presence of the slab.

Nós usamos tomografia sísmica de tempo de percurso para estudar o manto superior sob as regiões Sudeste e Centro-Oeste do Brasil. Este método baseia-se na inversão de resíduos relativos de tempo para as ondas P e S (VanDecar, 1991), que foram obtidos para mais de 80 estações em uma área de 20 x 20 graus. Mais de 11000 e 8000 resíduos de tempo foram obtidos para as ondas P (P direta e PKPdf) e S (S direta, ScS, SKS e SKKS), respectivamente, utilizando correlação cruzada de forma de ondas para até 12 estações operando simultaneamente. Para avaliar a robustez dos resultados com respeito aos dados, nós utilizamos o método estatístico de re-amostragem Jackknife, o qual inerentemente leva em conta a altamente variável cobertura dos raios e os erros das medidas, e pode fornecer limites de confiança para as anomalias. Inversões regionais foram realizadas para estudar a influência da parametrização nas anomalias sísmicas. Nossos resultados mostraram boa correlação das anomalias sísmicas com as principais estruturas tectônicas e revelaram novas anomalias que ainda não haviam sido observadas nos trabalhos anteriores. Anomalias de alta velocidade na porção oeste do Cráton do São Francisco apóiam a hipótese de que este cráton foi parte de uma placa Neoproterozóica maior. Anomalias de baixa velocidade sob a Província Tocantins (principalmente nas faixas móveis entre os Crátons Amazônico e do São Francisco) foram interpretadas como causadas por afinamento litosférico, consistente com a boa correlação entre a sismicidade intraplaca e as anomalias de baixa velocidade nesta região (Assumpção et al., 2004b). A melhora na resolução da anomalia de alta velocidade sob a Bacia do Paraná (~200 km) permitiu uma discussão sobre a geometria do núcleo cratônico desta Bacia. A subducção da Placa de Nazca foi observada como uma anomalia de alta velocidade sob a Bacia do Paraná (profundidades entre 700 e 1200 km). Nestas profundidades, uma grande anomalia de baixa velocidade aparece próxima da Placa de Nazca. Testes sintéticos mostraram que esta anomalia é um artefato da inversão gerado pela presença da Placa de Nazca. / We used travel time seismic tomography to study the upper mantle beneath SE and Central Brazil. This method is based on the inversion of P- and S-wave relative travel time residuals (VanDecar, 1991) obtained from more than 80 stations in an area of 20 x 20 degrees. More than 11000 P and PKP residuals, and more than 8000 S, ScS, SKS, and SKKS residuals were obtained from waveform cross-correlations for up to 12 simultaneous stations. To evaluate the robustness of the tomographic results with respect to the data, we use the Jackknife re-sampling method, which inherently take into account the highly variable ray coverage and measurement errors, and can provide confidence limits for the anomalies. Regional inversions were carried out to study influence of the parameterization on the seismic anomalies. Our results show correlations of seismic anomalies with the main tectonic structures and reveal new anomalies not yet observed in previous works. High velocity anomalies in the western portion of the São Francisco Craton support the hypothesis that this craton was part of a major Neoproterozoic plate. Low velocity anomalies beneath the Tocantins Province (mainly fold belts between the Amazon and São Francisco cratons) are interpreted as due to lithospheric thinning, consistent with a good correlation between intraplate seismicity and low velocity anomalies in this region (Assumpção et al., 2004b). The resolution improvement of the high velocity anomalies beneath the Parana Basin (at ~200 km) allows a discussion about the geometry of the cratonic nucleous of this Basin. The slab of the Nazca Plate is observed as a high velocity anomaly beneath the Parana basin (at 700-1200 km depth). At these depths, large low velocity anomaly appears next to the slab. Synthetic tests show that these anomalies are an artifact of the inversion generated by the presence of the slab.