This thesis has addressed several of the fundamental challenges in correlating the structure thermal transport properties of complex nanoporous polycrystalline zeolite materials. Two types of zeolite materials, MFI and LTA, were employed in order to investigate the effects of temperature and both the framework and non-framework cations on the thermal conductivity of zeolite. The thermal conductivity values of both materials were measured using a well intergrown zeolite film 3-omega method. The thermal transport mechanisms in these materials were investigated by separately analyzing the contributions of different phonon scattering processes. This thesis represents our progress towards a robust framework for understanding and predicting thermal transport properties of zeolite materials and complex crystals in general. Furthermore, the important roles of boundary and defect scattering, as illustrated in this thesis, also imply that the thermal conductivity of zeolite materials can be tuned by exploiting not only the composition but also the pore structure. In addition, a non-equilibrium molecular dynamics simulation with external force was developed and employed to predict the thermal conductivity of materials. It has shown that this method can accurately predict the thermal conductivity of simple materials, such as argon and quartz; however, it failed to predict the thermal conductivity of complex materials, such as zeolite. This thesis presents possible factors that can explain the phenomena and future recommendations to elucidate this issue.

Le cycle de condensation/sublimation du CO2 contrôle le climat martien et forme en hiver des dépôts saisonniers glacés, contaminés en glace d’eau et en poussière. L’objectif de cette thèse est de comprendre, caractériser, et si possible quantifier les différents processus microphysiques des dépôts saisonniers durant ce cycle de condensation/sublimation à l’aide de la simulation, en laboratoire, de ces processus à l’intérieur de la cellule expérimentale CarboN-IR. La condensation de glace de CO2 polycristalline sur un régolithe minéral, la modification du régime de condensation de la glace de CO2 dans la nuit polaire en présence de gaz non condensables, la stratification de glace d’eau sur la glace de CO2 lors de la sublimation des dépôts saisonniers et la remontée d’albédo de la glace polycristalline de CO2 lors de sa sublimation au printemps sont les phénomènes ayant été reproduits avec succès dans la cellule expérimentale. La condensation de glace polycristalline de CO2 sur Mars a également été observée durant sa formation, en automne, aux latitudes plus basses que la nuit polaire à l’aide des données du spectromètre OMEGA, de la sonde Mars Express. L’objet le plus lointain du système solaire à avoir été exploré, Pluton, partage de nombreux points communs avec Mars, notamment la couverture d’une partie de sa surface par des glaces (de CH4, N2 et CO pour Pluton) en équilibre avec son atmosphère. L’évolution de ces glaces peut être considérée comme analogue en termes de processus de surface les affectant : stratification des glaces en fonction de leur volatilité lors de leur sublimation où condensation à la surface, formation de glace sous forme polycristalline,... . Les données d’imagerie hyperspectrale (LEISA) fournies par la sonde New Horizons lors de son survol de Pluton en juillet 2015 ont permis la cartographie précise des espèces chimiques présentes à sa surface, ainsi que de leur état physique, étapes préliminaires à toute interprétation géologique. / The condensation/sublimation cycle of CO2 controls the martian climate and, in winter, forms frozen seasonal deposits, contaminated with water ice and dust. This thesis’ objective is to unders- tand, characterize, and if possible, to quantify the different microphysical processes occuring in the seasonal deposits during this condensation/sublimation cycle, with the insight of experimental simulation of these processes inside the CarboN-IR experimental cell. The CO2 ice condensation in a polycristalline form on a mineral regolith, the modification of the CO2 ice condensation regime in the martian polar night due to the presence of non condensable gases, the stratification of water ice onto CO2 ice during the sublimation of the seasonal deposits and the increase of the albedo of CO2 slab ice during its springtime sublimation are all phenomena that have been reproduced successfully inside the experiemental cell. The CO2 slab ice condensation on the Martian surface has been observed during its formation, in autumn, at lower latitudes than polar night limit with hyperspectral imaging from the OMEGA spectrometer onboard of the Mars Express probe. Yet the farthest object explored in the solar system, Pluto, shares many similarities with Mars, parti- cularly a partial icy cover of its surface (of CH4, N2 and CO ices for Pluto) in equilibrium with its atmosphere. The evolution of these ices can be considered similar in terms of the surface pro- cesses affecting them : stratification of ices in function of their volatility during their sublimation or condensation at the surface, formation of slab ice ... The hyperspectral imagery data (LEISA), provided by the New Horizons probe during its Pluto’s flyby in july 2015, allowed to determine the accurate cartography of chemical species at the surface, along with their physical state - the preliminary steps of any geological interpretation.

The parameterization of radiative transfer is a part of numerical weather prediction and general circulation models that is both essential and computationally very expensive, and is therefore subject to never­ending compromises between accuracy and computational cost. The present thesis offers an improvement to the existing broadband radiation scheme by revising its critical components - gaseous transmissions, cloud optical properties, and calculation of internal longwave exchanges. The accuracy of the full­spectrum broadband approach is thus raised to the level required for the short range numerical weather forecast. The intermittent update of broadband gaseous transmissions is introduced as a new component, reducing computational cost while preserving the full cloud­radiation interaction. The scalability of longwave computations is ensured by adopting the net exchanged rate decomposition with bracketing, improved by an intermittently applied self­learning algorithm determining the interpolation weights. It has been demonstrated that under conditions of operational weather forecasting, this developed scheme is fully competitive with the mainstream approach, due to the improved error balance between the...

In this thesis, I study numerical and semi-analytical models of self-gravitating protostellar discs, with the aim of furthering our understanding of the role of disc-self gravity in planet formation. At the time of writing, the ALMA era of observational astronomy is upon us. Therefore, I place my research into this context with synthetic images of both numerical and semi-analytical models. I begin with an examination into the apparent lack of convergence, with increasing resolution, of the fragmentation boundary in Smoothed Particle Hydrodynamics (SPH) simulations of a protostellar disc. I run a suite of SPH with different numerical implementations, and find that even very similar implementations can fundamentally change the final answer. I analyse a suite of SPH simulations that fragment to form gravitationally bound objects, with the motivation of informing future population synthesis model development. I find that fragment-fragment and fragment-disc interaction dominates the orbital evolution of the system even at very early times, and any attempt to produce a population of objects from the gravitational instability process must include these interactions. Before a disc fragments, it will go through a self-gravitating phase. If the disc cools globally on a timescale such that it is balanced by heating due to gravitational stresses, the disc will be in a state of quasi-equilibrium. So long as the disc mass is sufficiently low, and spirals are sufficiently tightly wound, then angular momentum transport can be described by the local approximation, for which there is an analytical description. Using this analytical description, I develop an existing 1D model into 3D, and examine a wide range of parameter space for which disc self-gravity produces significant non-axisymmetry. Using radiative transfer calculations coupled with synthetic observations, I determine that there is a very narrow range of parameter space in which a disc will have sufficiently large gravitational stresses so as to produce detectable spirals, but the stresses not be so large as to cause the disc to fragment. By developing a simple analytical prescription for dust, I show that this region of parameter space can be broadened considerably. However, it requires grains that are large enough to become trapped by pressure maxima in the disc, so I conclude that if self-gravitating spiral arms are detected in the continuum, it is likely that at least some grain growth has taken place.

Simulações acopladas do escoamento reativo e dos processos de transferência de calor para o estudo de chamas são problemas dispendiosos computacionalmente. A transferência de calor por radiação em processos de combustão, devido às elevadas temperaturas, é o processo de troca energética dominante. Ainda, o comportamento altamente irregular do coeficiente de absorção com o comprimento de onda se constitui em uma dificuldade adicional na modelagem da transferência radiativa em meios participantes. Para contornar essa dificuldade modelos espectrais foram desenvolvidos com o objetivo de simular o comportamento de um gás real. Dentre esses modelos destacam-se o gás cinza (GG: Gray Gas), o mais simples, que negligencia o comportamento espectral do coeficiente de absorção, e o modelo da soma-ponderada-de-gases-cinza (WSGG: Weighted-Sum-of-Gray-Gases) onde a integração sobre todo o espectro é substituída por um número finito de gases cinza. Com o avanço de ferramentas computacionais, principalmente códigos CFD (Computational Fluid Dynamics), abordagens computacionais se tornaram atrativas frente ou em complemento às abordagens experimentais. Desse modo, o presente trabalho tem por objetivo a aplicação dos modelos WSGG e GG com novas correlações na simulação detalhada de chamas difusivas laminares de metano diluído com dióxido de carbono e nitrogênio com o código CFD comercial ANSYS/Fluent. Foram desenvolvidas rotinas de usuário (UDF: User-Defined Functions) para o acoplamento dos modelos espectrais ao código CFD. A verificação das rotinas de usuário foi realizada comparando os resultados obtidos via simulação Fluent com dados obtidos pelo modelo WSGG com um código FORTRAN próprio desenvolvido pelo grupo de pesquisa do Laboratório de Radiação Térmica (LRT/UFRGS) para o problema unidimensional de superfícies negras e infinitas preenchidas por um meio não-isotérmico e não-homogêneo. Os erros encontrados para o fluxo de calor radiativo nas superfícies e para o termo fonte radiativo ao longo do meio foram da ordem de 1% indicando o funcionamento correto das rotinas UDF acopladas ao Fluent. Por fim, as rotinas foram aplicadas na simulação numérica para chamas de potência constante com diluição dos reagentes e os dados obtidos com a solução numérica foram comparados com dados experimentais para a fração radiante e fluxo de calor radiativo. Os desvios médios encontrados para o fluxo de calor radiativo ficaram em torno de 10% para todas as chamas, excetuando as chamas com diluição de CO2 de 30%, 40% e 50%, em volume, para as quais os desvios médios ficaram em torno de 15%. O termo fonte para as chamas apontou para a predominância da emissão do meio em relação à absorção. Todas as chamas estudadas se encontram no regime opticamente fino (optically thin) para o qual, segundo apontam estudos da literatura, a escolha do modelo espectral possui impacto pequeno em resultados globais da chama como a temperatura e a concentração das espécies na mistura. Nesse aspecto os resultados encontrados concordaram com a previsão da literatura, entretanto para a transferência radiativa, o modelo GG se mostrou sensivelmente menos preciso em comparação ao modelo WSGG, principalmente para a fração radiante e para o fluxo radiativo na região da pluma aquecida, indicando a dependência do modelo espectral adotado. / Coupled simulations of the reactive flow with the heat transfer processes for flame studying are computationally demanding problems. The radiative transfer in combustion processes is the main heat transfer mechanism due to the high temperatures involved. However, the highly irregular behavior o f the absorption coefficient with the wavenumber composes in an additional difficulty on modeling the radiative transfer in participating media. In order to overcome this issue, spectral models were developed with the objective of simulate the behavior of real gases. Some of the most known models are the gray gas (GG) for which the spectral behavior of the radiative properties of the medium is neglected and the weighted-sum-of-gray-gases (WSGG) for which the integration over the entire spectrum is replaced by a summation over a finite number of gray gases with constant absorption coefficients. With the development of computational tools, mainly Computational Fluid Dynamics (CFD) codes, numerical approaches became attractive instead or in complement of experimental set ups. In this way, the present work aims to couple the WSGG and the GG models with new correlations in a detailed simulation of diffusive laminar flames of methane diluted with carbon dioxide and nitrogen with the commercial CFD code ANSYS/Fluent. User-defined functions (UDF) were developed to the coupling of the spectral models. The verification was carried out through the WSGG model by comparing the Fluent solution with a solution obtained with a FORTRAN code developed by the Thermal Radiation Laboratory (LRT/UFRGS) research group for the one-dimensional system of black surfaces filled with a non-homogeneous and non-isothermal medium. The deviations for the radiative heat flux for the walls and the radiative heat source along the domain were of 1% or less, indicating the correct coupling between the UDF routines and the CFD code. Finally, the UDF were applied in the solution of constant power flames with fuel diluted with carbon dioxide and nitrogen. The obtained data was then compared with experimental measurements for the radiant fraction and the radiative heat flux along the flame axis. The average deviations found were in order of 10% for all flames, except for the flames with 30%, 40% and 50% of CO2 dilution, in volume, for which the deviatioms found were in order of 15%. The radiative heat source was plotted and indicated for the medium emission predominance in comparison with the medium absorption. All flames studied were optically thin flames for which, studies pointed, the spectral model have minor impact over global results as flame temperature and mixture concentration. For this aspect the results found showed agreement with the literature studies predictions, however the GG model showed itself less accurate in comparison with the WSGG model for the radiant fraction and the radiative heat flux computations. So the spectral models have influence on the radiative transfer even if its effect on flame structure can be negligible.

The atomic physics relevant to gaseous nebulae is critically examined using modelling software with particular emphasis on radio recombination lines (RRLs). The theoretical spectral line intensities can be deduced if we know the population structure of the bound electrons in the gas under non-thermal equilibrium conditions. The population structure of hydrogen is solved for various environments using a capture-collision-cascade model that incorporates an ambient radiation eld. The validity of assuming Case B (Baker & Menzel, 1938) for nebulae is investigated. It is known that Case B is appropriate for levels with small principal quantum numbers (n < 40), but this assumption is re-examined for high levels which are relevant to RRLs. E ects of an ambient radiation eld on the population structure is examined and processes that are stimulated by a radiation eld are included in the model. This is done as a preliminary investigation to extend the model to a photoionization code. / Physics / (M. Sc. (Astronomy)

Esta tese tem por objetivo propor uma metodologia para recuperação de perfis verticais de temperatura na atmosfera com nuvens a partir de medidas de radiância feitas por satélite, usando redes neurais artificiais. Perfis verticais de temperatura são importantes condições iniciais para modelos de previsão de tempo, e são usualmente obtidos a partir de medidas de radiâncias feitas por satélites na faixa do infravermelho. No entanto, quando estas medidas são feitas na presença de nuvens, não é possível, com as técnicas atuais, efetuar a recuperação deste perfil. É uma perda significativa de informação, pois, em média, 20% dos pixels das imagens acusam presença de nuvens. Nesta tese, este problema é resolvido como um problema inverso em dois passos: o primeiro passo consiste na determinação da radiância que atinge a base da nuvem a partir da radiância medida pelos satélites; o segundo passo consiste na determinação do perfil vertical de temperaturas a partir da informação de radiância fornecida pelo primeiro passo. São apresentadas reconstruções do perfil de temperatura para quatro casos testes. Os resultados obtidos mostram que a metodologia adotada produz resultados satisfatórios e tem grande potencial de uso, permitindo incorporar informações sobre uma região mais ampla do globo e, consequentemente, melhorar os modelos de previsão do tempo. / This thesis presents a methodology for retrieving vertical temperature profiles in the atmosphere with clouds from radiance measurements made by satellite, using artificial neural networks. Vertical temperature profiles are important initial conditions for numerical weather prediction models, and are usually obtained from measurements of radiance using infrared channels. Though, when these measurements are performed in the atmosphere with clouds, it is not possible to retrieve the temperature profile with current techniques. It is a significant loss of information, since on average 20% of the pixels of the images have clouds. In this thesis, this problem is solved as a two-step inverse problem: the first step is an inverse problem of boundary condition estimation, where the radiance reaching the cloud basis is determined from radiance measured by satellite; the second step consists in determining the vertical temperature profile from the boundary condition estimated in the first step. Reconstructions of temperature profile are presented for four test cases. The results show that the proposed methodology produces satisfactory results and has great potential for use, allowing to incorporate information from a wider area of the planet and thus to improve numerical weather prediction models.

La Tomographie Optique Diffuse (TOD) est une nouvelle technique d'imagerie médicale permettant de reconstruire les propriétés optiques des tissus biologiques dans le but de détecter des tumeurs cancéreuses. Il s’agit, toutefois, d’un problème inverse mal-posé et sous-déterminé. Le travail de cette thèse s’articule autour de la résolution de ce problème en utilisant l’équation du transfert radiatif comme modèle de propagation de la lumière (modèle direct). L’analyse de sensibilité a montré que le facteur d’anisotropie g de la fonction de phase de Henyey-Greenstein est le paramètre le plus influant sur la sortie du modèle direct suivi du coefficient de diffusion µs puis du coefficient d’absorption µa. Dans un premier temps, un algorithme de Gauss-Newton a été implémenté en utilisant les fonctions de sensibilités. Toutefois, ce dernier ne permet d’estimer qu'un nombre très limité de paramètres optiques (supposés constants en espace). Dans un second temps, un algorithme de Quasi-Newton a été développé pour reconstruire les distributions spatiales des propriétés optiques. Le gradient de la fonction objectif a été calculé efficacement par la méthode adjointe à travers le formalisme de Lagrange avec une approche Multi-fréquences. Les reconstructions sont obtenues à partir des données simulées en surface. Le facteur g est reconstruit comme un nouvel agent de contraste en TOD. Le problème de diaphonie entre µs g a été donc mis en évidence dans cette thèse. Notre algorithme a permis de reconstruire en 2D et 3D une ou plusieurs inclusions tumorales présentant différentes formes. La qualité des images reconstruites a été examinée en fonction du nombre de fréquences, de la diaphonie, du niveau de contraste (Inclusion/Fond), du niveau de bruit et de la position des inclusions tumorales / Diffuse Optical Tomography (DOT) is a new medical imaging technique used to reconstruct the optical properties of biological tissues in order to detect cancerous tumors. However, this is an ill-posed and under-determined inverse problem. The work of this thesis deals with the resolution of this problem using the radiative transfer equation as a forward model of light propagation. The sensitivity analysis showed that the anisotropy factor g of the Henyey-Greenstein phase function is the most sensitive parameter of the forward model followed by the scattering coefficient µs and then the absorption coefficient µa. In a first step, a Gauss-Newton algorithm was implemented using the sensitivity functions. However, this algorithm allows to estimate a very limited number of the optical parameters (assumed to be constant in space). In a second step, a Quasi-Newton algorithm was developed to reconstruct the spatial distributions of the optical properties. The gradient of the objective function was efficiently computed by the adjoint method through the Lagrangian formalism with a Multi-frequency approach. The reconstructed images were obtained from simulated boundary data. The g factor was reconstructed as a new optical contrast agent in DOT and the crosstalk problem between this factor and µs has been studied. The results showed that the algorithm is efficient to reconstruct in 2D and 3D one or several tumor inclusions having different shapes. The quality of the reconstructed images was examined according to several parameters: the number of frequencies, the crosstalk, the contrast and the noise levels

Les disques de débris orbitant dans l'environnement des étoiles proches constituent un indicateur très important des propriétés des systèmes planétaires extra-solaires. Depuis l'espace et au sol, les moyens observationnels actuels permettent de déterminer dans divers domaines de longueurs d'ondes les propriétés spatiales de ces disques et celles des grains de poussières circumstellaires. Cette thèse aborde le sujet de la modélisation des disques de débris, à partir de données fournies par de multiples instruments, en premier lieu les télescopes spatiaux Hubble et Herschel, et les interféromètres infrarouges du VLTI, CHARA, et KIN. Mes premiers projets ont pris place dans le cadre de deux programmes-clés de l'Observatoire Spatial Herschel dédiés à l'étude des disques circumstellaires autour des étoiles proches. Au sein du projet GASPS, j'ai obtenu des observations spectro-photométriques de HD 181327, une jeune étoile (12+8-4 millions d'années, Ma) de type solaire entourée d'un anneau de débris massif de 90 unités astronomique (UA) de rayon vu aussi en lumière diffusée par le télescope spatial Hubble. La bonne détermination de la géométrie de l'anneau permet de se concentrer sur la modélisation de la distribution spectrale d'énergie, afin de mieux caractériser les propriétés des poussières. J'ai utilisé le code de transfert radiatif GRaTer et démontré que le système héberge une population de planétésimaux glacés, qui pourrait représenter une source d'eau et de volatils susceptible d'être libérée sur des planètes telluriques encore en formation. Je discute quelques résultats additionnels obtenus avec Herschel à propos de disques de débris jeunes, notamment HD 32297, et d'analogues faibles de la Ceinture de Kuiper. Les disques exo-zodiacaux (exozodis), analogues du nuage Zodiacal du Système Solaire, représentent une contrepartie chaude (ou tiède) aux disques de débris, résidant proche de la zone habitable (moins de quelques unités astronomiques) et encore mal connue. Ils sont révélés par leur émission proche et moyen infrarouge et peuvent être étudiés avec la précision et la résolution requises grace à l'interférométrie optique. Dans le cas de l'étoile Beta Pictoris (12+8-4 Ma), dont le disque est vu par la tranche, une fraction significative du disque externe diffuse de la lumière vers le champ de vue des interféromètres ; une composante interne chaude doit tout de même être invoquée pour justifier de l'excès mesuré dans l'infrarouge proche. En m'appuyant sur l'exemple de l'étoile Véga (440±40 Ma), je présente la méthodologie employée et démontre que les exozodis chauds se caractérisent par une abondance de poussières sub-microniques, près de la distance de sublimation de l'étoile. D'un point de vue théorique, le mécanisme de production de ces petits grains non-liés est encore incompris. J'aborde plus en détails le cas du disque exozodiacal à deux composantes (chaude et tiède) de Fomalhaut (440±40 Ma). Je développe une nouvelle méthode de calcul des distances de sublimation et recense les processus variés qui peuvent affecter un grain de poussière afin de fournir un cadre pour l'interprétation : l'exozodi chaud à ~0.1 - 0.2 UA serait la signature indirecte d'une ceinture d'astéroïdes située à 2 UA à l'activité dynamique particulièrement intense. Finalement, je dresse un bilan des propriétés des disques de débris et de ce qu'ils peuvent nous apprendre quand on les compare au Système Solaire, et propose de futures directions de recherche pour explorer davantage les systèmes planétaires et leur dynamique. / Debris disks orbiting in the environment of nearby stars are a very important indicator of extrasolar planetary systems properties. From space and from the ground, current observational facilities enable a multi-wavelength determination of the disks structures and of the dust properties. This thesis addresses the topic of debris disks modelling, based on data from multiple instruments including first of all the Herschel and Hubble space telescopes, and the VLTI, CHARA and KIN infrared interferometers. My first research pro jects took place in the framework of two key programs from the Herschel Space Observatory dedicated to the study of circumstellar disks around nearby stars. As part of the GASPS pro ject, I obtained Herschel far-infrared spectro-photometric observations of HD 181327, a young (12+8−4 Myr) Sun-like star surrounded by a massive, 90 AU-wide debris belt, also imaged in scattered light by the Hubble Space Telescope. Proper determination of the belt geometry allows one to focus on modelling the dust properties. I used the GRaTer radiative transfer code to demonstrate that the system hosts a population of icy planetesimals that may be a source for the future delivery of water and volatiles onto forming terrestrial planets. I discuss additional results obtained with Herschel related to young debris disks, in particular HD 32297, and to faint Kuiper-Belt analogues. Exo-zodiacal disks (exozodis), those analogues to the Solar System Zodiacal cloud, represent a little known hot (or warm) counterpart of debris disks located close the habitable zone (inside of a few AUs). They are revealed by their near- to mid-infrared emission and can be assessed with the required accuracy and resolution with optical interferometers. In the case of the near edge-on star β Pictoris (12+8−4 Myr), I show that a significant fraction of the outer disk scatters light towards the small field-of-view of the interferometers; an inner hot component must nonetheless be invoked to explain the measured near-infrared excess. Based on the example of the star Vega (440 ± 40 Myr), I introduce a methodology to study inner dust disks and I show that hot exozodis are characterized by an abundance of submicron-sized grains, close to the star sublimation distance. From a theoretical point-of-view, the production mechanism for these small, unbound grains is not understood. I go into more details on the case of the Fomalhaut (440 ± 40 Ma) double-component (warm and hot) exozodiacal disk. I develop a new model for the calculation of the dust sublimation distances, and I address the various processes that can affect a dust grain in order to provide a framework for interpreting the exozodi: the hot exozodiacal disk may be the indirect signature of an asteroid belt with a particularly high dynamical activity. Lastly I draw up a summary of the properties of dusty debris disks and of what they can teach us when compared to the Solar System. I propose possible future research directions for further investigations of planetary systems and their dynamics.

Esta tese tem por objetivo propor uma metodologia para recuperação de perfis verticais de temperatura na atmosfera com nuvens a partir de medidas de radiância feitas por satélite, usando redes neurais artificiais. Perfis verticais de temperatura são importantes condições iniciais para modelos de previsão de tempo, e são usualmente obtidos a partir de medidas de radiâncias feitas por satélites na faixa do infravermelho. No entanto, quando estas medidas são feitas na presença de nuvens, não é possível, com as técnicas atuais, efetuar a recuperação deste perfil. É uma perda significativa de informação, pois, em média, 20% dos pixels das imagens acusam presença de nuvens. Nesta tese, este problema é resolvido como um problema inverso em dois passos: o primeiro passo consiste na determinação da radiância que atinge a base da nuvem a partir da radiância medida pelos satélites; o segundo passo consiste na determinação do perfil vertical de temperaturas a partir da informação de radiância fornecida pelo primeiro passo. São apresentadas reconstruções do perfil de temperatura para quatro casos testes. Os resultados obtidos mostram que a metodologia adotada produz resultados satisfatórios e tem grande potencial de uso, permitindo incorporar informações sobre uma região mais ampla do globo e, consequentemente, melhorar os modelos de previsão do tempo. / This thesis presents a methodology for retrieving vertical temperature profiles in the atmosphere with clouds from radiance measurements made by satellite, using artificial neural networks. Vertical temperature profiles are important initial conditions for numerical weather prediction models, and are usually obtained from measurements of radiance using infrared channels. Though, when these measurements are performed in the atmosphere with clouds, it is not possible to retrieve the temperature profile with current techniques. It is a significant loss of information, since on average 20% of the pixels of the images have clouds. In this thesis, this problem is solved as a two-step inverse problem: the first step is an inverse problem of boundary condition estimation, where the radiance reaching the cloud basis is determined from radiance measured by satellite; the second step consists in determining the vertical temperature profile from the boundary condition estimated in the first step. Reconstructions of temperature profile are presented for four test cases. The results show that the proposed methodology produces satisfactory results and has great potential for use, allowing to incorporate information from a wider area of the planet and thus to improve numerical weather prediction models.