Significant X-ray variability and flaring has been observed from Sgr A* but is poorly understood from a theoretical standpoint. We perform general relativistic magnetohydrodynamic simulations that take into account a population of non-thermal electrons with energy distributions and injection rates that are motivated by PIC simulations of magnetic reconnection. We explore the effects of including these non-thermal electrons on the predicted broadband variability of Sgr A* and find that X-ray variability is a generic result of localizing non-thermal electrons to highly magnetized regions, where particles are likely to be accelerated via magnetic reconnection. The proximity of these high-field regions to the event horizon forms a natural connection between IR and X-ray variability and accounts for the rapid timescales associated with the X-ray flares. The qualitative nature of this variability is consistent with observations, producing X-ray flares that are always coincident with IR flares, but not vice versa, i.e., there are a number of IR flares without X-ray counterparts.

The data harvest from the Voyagers' (V1 and V2) Ultraviolet Spectrometers (UVS) covers encounters with the outer planets, measurements of the heliosphere sky-background, and stellar spectrophotometry. Because their period of operation overlaps with many ultraviolet missions, the calibration of V1. and V2 UVS with other spectrometers is invaluable. Here we revisit the UVS calibration to assess the intriguing sensitivity enhancements of 243% (V1) and 156% (V2) proposed recently. Using the Ly alpha airglow from Saturn, observed in situ by both Voyagers, and remotely by International Ultraviolet Explorer (IUE), we match the Voyager values to IUE, taking into account the shape of the Saturn Ly alpha line observed with the Goddard High Resolution Spectrograph on board the Hubble Space Telescope. For all known ranges of the interplanetary hydrogen density, we show that the V1 and V2 UVS sensitivities cannot be enhanced by the amounts thus far proposed. The same diagnostic holds for distinct channels covering the diffuse He I 58.4 nm emission. Our prescription is to keep the original calibration of the Voyager UVS with a maximum uncertainty of 30%, making both instruments some of the most stable EUV/FUV spectrographs in the history of space exploration. In that frame, we reassess the excess Ly alpha emission detected by Voyager UVS deep in the heliosphere, to show its consistency with a heliospheric but not galactic origin. Our finding confirms results obtained nearly two decades ago-namely, the UVS discovery of the distortion of the heliosphere and the corresponding obliquity of the local interstellar magnetic field (similar to 40 degrees from upwind) in the solar system neighborhood-without requiring any revision of the Voyager UVS calibration.

The Monte Carlo (MC) method is an effective approach to simulate the radiative transfer in an inhomogeneous cloudy atmosphere. It is based on the direct physical simulation of the extinction processes that solar and thermal photons incur when traveling through the atmosphere. A detailed description of the MC method is presented in the second chapter. A new three-dimensional Monte Carlo radiative transfer model, based on a pre-existing model (Trautmann et al. [1999]), has been developed. Some outstanding characteristics of this model are discussed in chapter 3. Several simulations of reflectances, transmittances, absorptances and horizontal flux densities have been performed, the results of which have been compared with worldwide accepted codes (chapter 4). The two cases selected for the radiative transfer computations were taken from the Intercomparison of 3D Radiative Codes (I3RC) project: an ARM (Atmospheric Radiation Measurements) reconstructed cloud and a 3D marine boundary layer cloud / Die Monte Carlo (MC) Methode ist ein effektives Verfahren, um den Strahlungstransport in einer inhomogenen bewölkten Atmosphäre zu simulieren. Es begründet sich auf der direkten Simulation der Extinktionsprozesse eines solaren oder thermischen Photons auf seinem Weg durch die Atmosphäre. Eine detallierte Beschreibung der MC Methode erfolgt in Kapitel 2. In Kapitel 3 wird ein neues dreidimensionales MC-Strahlungstranportmodell vorgestellt, das, aufbauend auf einem schon bestehenden Modell (Trautmann et al. [1999]), entwickelt wurde. Mehrere Simulationen von Reflektanzen, Transmittanzen, Absorptanzen und Strahlungsflussdichten für zwei Fälle des \"Intercomparison of 3D radiative Codes\" projektes, nämlich eine ARM rekonstruierte Wolke und eine 3D marine Grenzschichwolke, wurden durchgeführt, und mit den Ergebnissen anderer weltweit akzeptierten Codes verglichen.

Nuvens desempenham papel fundamental no balanço radiativo terrestre, e o conhecimento de suas propriedades micro e macrofísicas é importante para o estudo do clima global. O desenvolvimento de nuvens esta ligado à dinâmica da atmosfera, fluxos de energia térmica e radiativa à superfície, e também depende crucialmente do tamanho de seus hidrometeoros. Uma ferramenta importante para o estudo de hidrometeoros em nuvens de grande escala e o sensoriamento remoto por satélite, que representa uma alternativa para se estudar propriedades de nuvens em grandes escalas espaciais. No entanto, atualmente não há estudos para subsidiar análises de propriedades microfísicas de nuvens a partir de plataformas geoestacionárias. Neste trabalho foram comparados dois códigos de transferência radiativa amplamente utilizados pela comunidade científica, SBDART e libRadtran, em simulações considerando variações de propriedades atmosféricas, de superfície, macro e microfísicas de nuvens. Em seguida estudou-se a transferência radiativa em nuvens de água e gelo determinando o impacto simulado de diversas geometrias de iluminação e observação nas radiâncias medidas por sensoriamento remoto. Finalmente, foram desenvolvidas tabelas de referência para a determinação de tamanho de hidrometeoros em nuvens, para fases líquida e solida, a partir de radiâncias medidas por um sensor em satélite geoestacionário. As comparações entre SBDART e libRadtran mostram resultados sistematicamente subestimados pelo libRadtran. Em um céu sem nuvens a diferença entre as refletâncias calculadas em ambos os modelos é inferior a 4% quando consideradas variações em albedo de superfície e coluna de ozônio. Em um céu com nuvens a diferença entre os resultados dos modelos para comprimento de onda de 630 nm e intervalo visível (590 a 660 nm) pode chegar ate 18% para variações de profundidade óptica entre 0 e 20. Para profundidade óptica maior que 20 essas diferenças variam entre 4 e 9%, além de refletâncias constantes para o intervalo entre 3850 e 4000 nm e para comprimento de onda de 3900 nm. A divergência de resultados obtidos pelos dois modelos é atribuída a diferenças estruturais entre os códigos, uma vez que atualmente apenas o libRadtran apresenta atualizações periódicas pela comunidade científica. Assim, este modelo foi utilizado para a construção da tabela de referência simulando medidas de radiâncias em plataforma geoestacionária. A tabela construída compreende 86 valores de raios efetivos de hidrometeoros, variando entre 2 e 59m, com fases termodinâmicas de água e gelo, 16 valores de geometria de observação solar, 3 valores de âgulos de observação, 2 valores de azimute relativo, e condições fixas de profundidade óptica de nuvem de 50, conteúdo integrado de ozônio de 255 DU e 60mm de vapor d água. Essas condições foram escolhidas como representativas para a região Amazônica. Em um teste de aplicação direta da tabela de referência a medidas de radiância realizadas com satélite geoestacionário foram obtidos raios efetivos entre 2 e 30m para gotículas de água e até 24m para cristais de gelo. Os resultados obtidos neste trabalho poderão ser aplicados futuramente a medidas obtidas por plataformas geoestacionárias em estudos de tamanhos de hidrometeoros, tornando possível a análise de sua evolução temporal. / Clouds play a fundamental role in the terrestrial radiative balance, and knowledge of its micro and macrophysical properties is important for the study of global climate. Cloud development is linked to the dynamics of the atmosphere, thermal and radiative energy flows to the surface, and also depends crucially on the size of its hydrometeors. An important tool for the study of large-scale cloud hydrometeors is satellite remote sensing, which represents an alternative to study cloud properties at large spatial scales. However, there are currently no studies to support analyzes of microphysical properties of clouds from geostationary platforms. In this work, two radiative transfer codes widely used by the scientific community, SBDART and libRadtran, were compared in simulations considering variations of atmospheric, surface, macro and microphysical properties of clouds. Next, radiative transference was studied in water and ice clouds, determining the simulated impact of various lighting and observation geometries on radiances measured by remote sensing. Finally, look-up tables were developed for the determination of the size of hydrometeors in clouds, for liquid and solid phases, from radiances measured by a geostationary satellite sensor. The comparisons between SBDART and libRadtran show results systematically underestimated by the libRadtran. In a cloudless sky the difference between the reflectances calculated in both models is less than 4% when considering variations in surface albedo and ozone. In a clouded sky the difference between the model results for wavelengths of 630 nm and the visible range (590 to 660 nm) can reach up to 18% for optical depth variations between 0 and 20. For optical depths greater than 20, these differences range from 4 to 9%, in addition to constant reflections for the range of 3850 to 4000 nm and for a wavelength of 3900 nm. The divergence of results obtained by the two models is attributed to structural differences between the codes, since currently only the libRadtran presents periodic updates by the scientific community. Thus, this model was used for the construction of the reference table simulating measurements of radiances in geostationary platform. The constructed table comprises 86 values of effective radii of hydrometeors, ranging from 2 to 59m, with thermodynamic phases of water and ice, 14 values of geometry of solar observation, 3 values of angles of observation, 2 relative azimuth values, and fixed cloud optical depth conditions of 50, integrated ozone content of 255 DU and 60mm of water vapor. These conditions were chosen as representative for the Amazon region. In a test of direct application of the look-up table to measurements of radiance obtained with geostationary satelite we obtained effective radius up to between 2 and 30 for water droplets and up to 24 for ice crystals. The results obtained in this work can be applied in the future to measurements obtained by geostationary platforms in studies of sizes of hydrometeors, making possible the analysis of their temporal evolution.

La nouvelle fenêtre spectroscopique dans le sub-millimétrique, ouverte par l’avènement del’observatoire spatial Herschel, a permis la détection d’espèces azotées simples, les hydruresd’azote NH, NH2 et NH3, dans les enveloppes froides de proto-étoiles. Ces enveloppes sontconstituées de gaz dense et froid caractéristique des conditions physico-chimiques des nuagesmoléculaires. L’observation d’hydrures d’azote dans de tels environnements a donc permis d’apporterde nouvelles contraintes sur la chimie interstellaire de ces nuages, et nous a donné enparticulier l’occasion de ré-explorer la chimie de l’azote.L’objectif de mon travail de thèse a été d’analyser en détail cette chimie interstellaire etprincipalement la formation en phase gazeuse d’espèces polyatomiques simples : les hydruresd’azote. Dans des conditions de gaz dense et froid (n = 104 cm−3, T = 10 K), la chimie de cesderniers est initiée par une chimie lente (la conversion de N en N2 par réactions neutre-neutre),contrairement à celles de ses analogues oxygénés et carbonés. Nous nous sommes particulièrementintéressés à cette étape de la chimie de l’azote, au vu des récents travaux théoriqueset expérimentaux menés par plusieurs équipes d’experts (Bordeaux, Besançon). De plus, lesrécents travaux concernant la conversion ortho-para de l’hydrogène moléculaire et les nouveauxcalculs de rapports de branchement de spins nucléaires pour les voies de production des hydruresd’azote dans leurs configurations ortho et para, menés à l’IPAG, nous ont permis d’entreprendrele calcul auto-cohérent des différentes symétries de spin des hydrures d’azote et de l’hydrogènemoléculaire. Nous avons ainsi pu développer un nouveau réseau chimique de l’azote, bénéficiantdes taux cinétiques les plus à jour pour les réactions critiques impliquées dans la chimie deshydrures d’azote.Ce nouveau réseau est utilisé pour modéliser l’évolution temporelle des abondances desespèces azotées dans des conditions de gaz dense et froid ( 103 < n < 106 cm−3, T ≤ 50 K).Les résultats à l’état stationnaire sont comparés aux observations de NH, NH2 et NH3, dans lesenveloppes froides de proto-étoiles de faible masse, en étudiant l’influence des abondances totalesen phase gazeuse du carbone, de l’oxygène et du soufre. Nos modèles chimiques reproduisent lesabondances des hydrures d’azote observés et leurs rapports pour un rapport C/O élementaire, enphase gazeuse, de ∼ 0.8 et à condition que l’abondance totale de soufre soit déplétée d’au moinsun facteur 2. Les rapports ortho/para prédits par nos modèles, pour NH2 et NH3, respectivement∼ 2.3 et ∼ 0.7, sont compatibles avec les observations de ces derniers dans des nuages diffusfroids. Les abondances des hydrures d’azote, dans des conditions de nuages sombres, sont donccohérentes avec une synthèse purement en phase gazeuse. De plus, nos résultats soulignent lefait que NH provient d’une voie de formation différente de celle de NH2 et NH3. NH vient de larecombinaison dissociative de N2H+ alors que la formation de NH2 et NH3 est principalementdue à la recombinaison dissociative de l’ion ammonium (NH+4 ), lui même molécule fille deN+. Ainsi, NH2 et NH3 procèdent de l’échange de charge dissociatif N2 + He+, tandis que NHprovient de la réaction N2 + H+3 . / The new spectroscopic window opened by the advent of the Herschel Space Observatory,has enabled the detection of simple nitrogen species, the nitrogen hydrides NH, NH2, and NH3,in the cold envelope of protostars. These envelopes are made of dense cold gas characteristicof the physico-chemical conditions of molecular clouds. The observation of nitrogen hydrides insuch environments has brought new constraints on the interstellar chemistry of these kind ofclouds, and gives, in particular, the opportunity to revisit the chemistry of nitrogen.The aim of my thesis was to comprehensively analyse the interstellar chemistry of nitrogen,focussing on the gas-phase formation of the simplest polyatomic species, namely nitrogen hydrides.Under dense, cold gas conditions (n = 104 cm−3, T = 10 K), the chemistry of theselatter is initiated by a slow chemistry (the conversion from N to N2 with neutral-neutral reactions),in contrast to their carbonated and oxygenated analogues. We have investigated andrevisited this specific part of the nitrogen chemistry in the light of recent theoretical and experimentalwork carried out by several expert teams (Bordeaux, Besançon). In addition, recentwork about the ortho-para conversion of molecular hydrogen and new calculations of nuclearspin branching ratios for the production pathways of nitrogen hydrides in their ortho and paraconfigurations conducted at IPAG, enabled us to treat self-consistently the different spin symmetriesof the nitrogen hydrides together with the ortho and para forms of molecular hydrogen.We were able to develop a new network of chemical nitrogen in which the kinetic rates of criticalreactions involved in the nitrogen chemistry have been updated.This new network is used to model the time evolution of the nitrogen species abundancesin dense cold gas conditions (T ≤ 50 K, 103 < n < 106 cm−3). The steady-state resultsare compared to observations of NH, NH2 and NH3 towards a sample of low-mass protostars,with a special emphasis on the influence of the overall amounts of gaseous carbon, oxygen, andsulphur. Our chemical models reproduced the nitrogen hydrides abundances and their ratios fora gas-phase elemental C/O ratio of ∼ 0.8, provided that the total amount of sulphur is depletedby at least a factor of two. Our predicted ortho-to-para ratios for NH2 and NH3, ∼ 2.3 and∼ 0.7 respectively, are in good agreement with the observations towards cold diffuse clouds.Then, in dark gas conditions, the nitrogen hydride abundances are consistent with a pure gasphasesynthesis. Moreover, our results are based on the fact that NH is coming from a differentpathway than NH2 and NH3. NH is the daughter molecule of N2H+, deriving from the reactionN2+H+3 , while NH2 and NH3 proceed from NH+4 , itself daughter molecule of N+, resulting fromthe dissociative charge exchange N2 + He+.

Les observations météorologiques depuis les satellites dans le domaine des micro-ondes sont actuellement limitées à 190 GHz. La prochaine génération de satellites météorologiques opérationnels européens (EUMETSAT Polar System-Second Generation-EPS-SG), em- portera un instrument, le Ice Cloud Imager (ICI), avec des fréquences sub-millimétriques jusqu’à 664 GHz, afin d’améliorer la caractérisation globale des nuages de glace. Pour préparer l’exploitation de ces nouvelles données, durant cette thèse, des travaux ont été effectués sur quatre axes complémentaires. Des simulations réalistes de transfert radiatif ont été réalisées de 19 à 700 GHz, pour des scènes météorologiques réelles, couvrant une grande variabilité des nuages en Europe. L’objectif était double : premièrement mieux comprendre la sensibilité des ondes millimétriques et sub-millimétriques à la phase glacée des nuages, deuxièmement créer une base de données robuste pour développer une méthode d’inversion statistique des caractéristiques des nuages de glace. Un code de transfert radiatif atmosphérique (Atmospheric Radiative Transfer Simulator ARTS) a été couplé avec des profils atmosphériques simulés par un modèle méso-échelle de nuage (Weather Research and Forecasting WRF), pour douze scènes européennes aux moyennes latitudes. Les propriétés de diffusion des hydrométéores (glace, neige, graupel, pluie et eau dans le nuage) ont été soigneusement sélectionnées, en particulier pour la phase glace, et la compatibilité avec la microphysique de WRF a été testée : le Discrete-Dipole approximation (DDA) est adopté pour calculer les propriétés diffusantes des particules de neige. Les simulations obtenues ont été systématiquement comparées avec des observations satellitaires coïncidentes d’imageurs et de sondeurs jusqu’à 200 GHz. L’accord entre les simulations et les observations montre la bonne qualité de la base de données, au moins jusqu’à 200 GHz... / The meteorological observations from satellites in the microwave domain are currently limited to 190 GHz. The next generation of European operational Meteorological Satellite (EUMETSAT Polar System-Second Generation-EPS-SG) will carry an instrument, the Ice Cloud Imager (ICI), with frequencies up to 664 GHz, to provide unprecedented measurements in the sub-millimeter spectral range, aiming to improve the characterization of ice clouds over the globe. To prepare this upcoming satellite-borne sub-millimeter imager, during this thesis, scientific efforts have been made on four complementary aspects. Realistic radiative transfer simulations have been performed from 19 to 700 GHz, for real meteorological scenes, covering a large variability of clouds in Europe. The goal was two fold, first to better understand the sensitivity of the microwave to sub-millimeter waves to the cloud frozen phases, and second, to create a robust training database for a statistical cloud parameter retrieval. The Atmospheric Radiative Transfer Simulator (ARTS) is coupled with atmospheric profiles from the Weather Research and Forecasting (WRF) model, for twelve diverse European mid-latitude scenes. The single scattering properties of the hydrometeors (cloud ice, snow, graupel, rain, and cloud water) are carefully selected, especially for the frozen phases, and compatibility with the micro- physics in WRF is tested: the Discrete-Dipole Approximation (DDA) is adopted for snow particles. The resulting simulations have been systematically compared with coincident satellite observations from imagers and sounders up to 200 GHz. The agreement between simulations and observations shows the good quality of the simulated training database, at least up to 200 GHz...

Thesis. 1976. Sc.D.--Massachusetts Institute of Technology. Dept. of Chemical Engineering. / Microfiche copy available in Archives and Science. / Bibliography: leaves 225-237. / Sc.D.

Cool giant and supergiant stars are among the largest and most luminous stars in the Universe and, therefore, dominate the integrated light of their host galaxies. These stars were extensively studied during last few decades, however their relevant properties like photometric variability and mass loss are still poorly constrained. Understanding of these properties is crucial in the context of a broad range of astrophysical questions including chemical enrichment of the Universe, supernova progenitors, and the extragalactic distance scale. Atmospheres of giant and supergiant stars are characterized by complex dynamics due to different interacting processes, such as convection, pulsation, formation of molecules and dust, and the development of mass loss. Current 1D/3D dynamical model atmospeheres are able to simulate these processes and produce a good agreement with the observed spectral features of evolved stars. However, the models lack constraints and need to be confronted to observables. Dynamical processes in stellar atmospheres impact the formation of spectral lines producing their asymmetries and Doppler shifts. Thus, by studying the line-profile variations on spatial and temporal scales it is possible to reconstruct atmospheric motions in evolved stars. As will be shown in this thesis, a tomographic method is an ideal technique for this purpose. The tomographic method is based on construction and cross-correlation of spectral templates (masks) with observed or synthetic stellar spectra in order to recover velocity fields at different optical depths in the stellar atmosphere.The first part of the thesis further improves the original implementation of the tomographic method. This improvement involves the computation of the contribution function in order to correctly determine an optical depth of formation of spectral lines. The tomographic method is, then, fully validated by applying it to a stellar convection simulation of a red supergiant star and correctly recovering its velocity field throughout the atmosphere. The second part of the thesis applies the tomographic method to the red supergiant star μ Cep in order to constrain its atmospheric motions and relate them to photometric variability. A phase lag (hysteresis) between the effective temperature and the radial velocity variations is revealed with timescales of a few hundred days, similar to photometric ones. A comparison to a stellar convection simulation of a red supergiant star indicates that hysteresis loops are linked to the stochastic shocks generated and shaped by the underlying large-scale convection and may be responsible for photometric variations in μ Cep. The third part of the thesis applies the tomographic method to spectro-interferometric observations of the Mira-type star S Ori. The uniform-disk angular diameters measured at wavelengths contributing to the tomographic masks increase with decrease of an optical depth probed by the masks. This validates the capability of the tomographic method to probe distinct geometrical depths in the stellar atmosphere. The last part of the thesis applies the tomograhic method to the Mira-type star RY Cep and compares the results to those obtained for μ Cep in this thesis. The comparison reveals differences in their behavior in the temperature-velocity plane pointing to the posibility to differentiate between Mira-type and red supergiant stars from their spectroscopic signatures. / Doctorat en Sciences / info:eu-repo/semantics/nonPublished

Nous cherchons à mettre en évidence l'influence des hétérogénéités de température, de structure ou de composition des atmosphères sur leurs observations. Dans le années à venir, de plus en plus d'appareils vont permettre l'observation par transmission des atmosphères d'exoplanètes. Toutefois, les outils numériques permettant de contraindre ces dernières reposent sur des modèles simples à une dimension. Ils supposent en effet des atmosphères ne possédant qu'une structure verticale (le climat est le même en tout point de la surface, la composition ou la température n'évolue qu'avec l'altitude). Cette approche a le mérite de permettre des calculs rapides et de contraindre les paramètres globaux de l'atmosphère avec des temps raisonnables. Ceci ne serait pas possible en l'état avec une modélisation en 3 dimensions des atmosphères, même si ce serait beaucoup plus réaliste. Ce que nous cherchons à mettre en évidence, ce sont les limites des techniques actuelles d'inversion et donc, de caractérisation des atmosphères qui seront observées. Pour cela, il fallait mettre au point un logiciel capable de résoudre le transfert radiatif au sein d'une atmosphère en 3 dimensions (et non plus 1 seul). Une fois le logiciel terminé, nous avons éprouvé l'algorithme de traitement du signal TauREx en comparant les résultats qu'il proposait à des simulations atmosphériques parfaitement contrôlées. Nous nous sommes tout principalement arrêté sur les biais découlant d'hétérogénéités de température en simulant des atmosphères avec un fort contraste jour/nuit. Ceci nous a permis de caractériser les biais découlant de ces types d'hétérogénéités, de les quantifier et de mettre l'accent sur un biais jusqu'ici très sous-estimé par la communauté, à savoir celui découlant des hétérogénéités le long de la ligne de visée. Nous avons appuyé nos propos et concentré nos efforts sur l'interprétation de l'inversion d'une simulation complexe de l'atmosphère de GJ 1214 b. La reconstitution de la chaine observationnelle : GCM (LMD), Pytmosph3R (LAB) et TauREx (UCL) ouvre les portes d'un vaste panel d'études envisageables, et notamment tout ce qui va concerner l'identification et la caractérisation des biais systématiques qui incomberont les observations à venir. / Transmission spectroscopy provides us with information on the atmospheric properties at the limb, which is often intuitively assumed to be a narrow annulus aound ther planet. Consequently, the few recent studies on the effect of atmospheric horizontal heterogeneities on transmission spectra have used approaches sensitive to variations along the limb only. Here we demonstrate that the region probed in transmission – the limb – actually extends significantly toward the day and night sides of the planet. Consequently we show that thestrong day-night thermal and compositional gradients expected on synchronous exoplanets create sufficient heterogeneities across the limb to result in important systematic effects on the spectrum and bias its interpretation. To quantify these effects, we developed a 3D radiative transfer model able to generate transmission spectra of atmospheres based on 3D atmospheric structures, whether they come from a Global Climate Model or more parametrized models. We first apply this tool to a simulation of the atmosphere of GJ 1214 b toproduce synethic JWST observations and show that producing a spectrum using only atmospheric columns at the terminator results in errors greater than expected noise. This demonstrates the necessity of a real 3D approach to model data for such precise observatories.Second, we investigate how day-night temperature gradients cause a systematic bias in retrieval analysis performed with 1D forward models. For that purpose we synthesize a large set of forward spectra for prototypical HD209458 b and GJ 1214 b type planets varying the temperatures of the day and night sides as well as the width of the transition region. We then perform typical retrievalanalyses and compare the retrieved parameters to the ground truth of the input model. This study reveals systematic biases on the retrieved temperature (found to be higher than the terminator temperature) and absorber abundances. This is due to the fact that the hotter dayside is more extended vertically and screens the nightside—a result of the nonlinear properties of atmospheric transmission.These biases will be difficult to detect as the 1D profiles used in the retrieval procedure are found to provide an excellent match to theobserved spectra based on standard fitting criteria (chi2, posterior distributions). This fact needs to be kept in mind when interpretingcurrent and future data.

L’imagerie moléculaire occupe une place majeure dans le domaine de la recherche préclinique. Parmi les modalités existantes, les techniques optiques fondées sur la détection d’un rayonnement visible ou proche infrarouge sont les plus récentes et sont principalement représentées par les méthodes de tomographie optique de luminescence. Ces méthodes permettent une caractérisation 3D d’un milieu biologique par la reconstruction de cartes de concentration ou la localisation de marqueurs luminescents sensibles à des processus biologiques et chimiques se déroulant à l’échelle de la cellule ou de la molécule. La tomographie de luminescence se fonde sur un modèle de propagation de la lumière dans les tissus, un protocole d’acquisition du signal en surface du milieu et une procédure numérique d’inversion de ces mesures afin de reconstruire les paramètres d’intérêts. Ce travail de thèse s’articule donc autour de ces trois axes et apporte un élément de réponse à chacun des problèmes. L’objectif principal de cette étude est d’introduire et de présenter des outils d’évaluation des performances théoriques d’une méthode de tomographie optique. L’un des aboutissements majeurs est la réalisation de reconstructions tomographiques expérimentales à partir d’images acquises par un imageur optique conçu pour l’imagerie planaire 2D et développé par la société Quidd. Dans un premier temps nous abordons la théorie du transport en milieu diffusant afin de poser les concepts et outils sur lesquels vont s’appuyer l’ensemble des travaux. Nous présentons particulièrement deux modèles de propagation différents ainsi que les méthodes de résolution et les difficultés théoriques qui leur sont liées. Dans une deuxième partie nous introduisons les outils statistiques utilisés pour caractériser les systèmes tomographiques et leur résolution potentielle. Nous définissons une procédure et nous l’appliquons à l’étude de quelques situations simples en tomographie de luminescence. La dernière partie de ce travail présente la mise au point d’une procédure d’inversion. Après avoir présenté le cadre théorique dans lequel cette procédure s’inscrit nous la validons à partir de données numériques avant de l’appliquer avec succès à des mesures expérimentales. / Molecular imaging is a major modality in the field of preclinical research. Among the existing methods, techniques based on optical detection of visible or near infrared radiation are the most recent and are mainly represented by luminescence optical tomography techniques. These methods allow for 3D characterization of a biological medium by reconstructing maps of concentration or localisation of luminescent beacons sensitive to biological and chemical processes at the molecular or cellular scale. Luminescence optical tomography is based on a model of light propagation in tissues, a protocol for acquiring surface signal and a numerical inversion procedure used to reconstruct the parameters of interest. This thesis is structured around these three axes and provides an answer to each problem. The main objective of this study is to introduce and present the tools to evaluate the theoretical performances of optical tomography methods. One of its major outcomes is the realisation of experimental tomographic reconstructions from images acquired by an optical imager designed for 2D planar imaging and developed by the company Quidd. In a first step we develop the theory of transport in scattering medium to establish the concept on which our work will rely. We present two different propagation models as well as resolution methods and theoretical difficulties associated with them. In a second part we introduce the statistical tools used to characterise tomographic systems. We define and apply a procedure to simple situations in luminescence optical tomography. The last part of this work presents the development of an inversion procedure. After introducing the theoretical frameworkwe validate the procedure fromnumerical data before successfully applying it to experimentalmeasurements.