Detecção automática de regiões planares em nuvens de pontos é um importante passo para muitas aplicações gráficas, de processamento de imagens e de visão computacional. Enquanto a disponibilidade de digitalizadores a laser e a fotografia digital tem nos permitido capturar nuvens de pontos cada vez maiores, técnicas anteriores para detecção de planos são computacionalmente caras, sendo incapazes de alcançar desempenho em tempo real para conjunto de dados contendo dezenas de milhares de pontos, mesmo quando a detecção é feita de um modo não determinístico. Apresentamos uma abordagem determinística para detecção de planos em nuvens de pontos não estruturadas que apresenta complexidade computacional O(n log n) no número de amostras de entrada. Ela é baseada em um método eficiente de votação para a transformada de Hough. Nossa estratégia agrupa conjuntos de pontos aproximadamente coplanares e deposita votos para estes conjuntos em um acumulador esférico, utilizando núcleos Gaussianos trivariados. Uma comparação com as técnicas concorrentes mostra que nossa abordagem é consideravelmente mais rápida e escala significativamente melhor que as técnicas anteriores, sendo a primeira solução prática para detecção determinística de planos em nuvens de pontos grandes e não estruturadas. / Automatic detection of planar regions in point clouds is an important step for many graphics, image processing, and computer vision applications. While laser scanners and digital photography have allowed us to capture increasingly larger datasets, previous techniques are computationally expensive, being unable to achieve real-time performance for datasets containing tens of thousands of points, even when detection is performed in a non-deterministic way. We present a deterministic technique for plane detection in unorganized point clouds whose cost is O(n log n) in the number of input samples. It is based on an efficient Hough-transform voting scheme and works by clustering approximately co-planar points and by casting votes for these clusters on a spherical accumulator using a trivariate Gaussian kernel. A comparison with competing techniques shows that our approach is considerably faster and scales significantly better than previous ones, being the first practical solution for deterministic plane detection in large unorganized point clouds.

Neste projeto de pesquisa planeja-se obter indícios da correlação entre o conteúdo metálico estelar deduzido para a Via Láctea e os indíces metálicos obtidos para as Nuvens de Magalhães. O ponto de apoio para tal comparação é que cada encontro deixará importantes traços na eficiência de formação dos membros do tripleto. À medida que os encontros dependem da história dinâmica, suas ``impressões digitais\'\' deixadas nos conteúdos estelares colocam importantes limitações na história dinâmica e vice-versa. Para tanto, são utilizados os dados espectroscópicos já obtidos com o telescópio Magellan Clay, para uma amostra contendo 28 estrelas ricas em carbono encontradas nas Nuvens de Magalhães. A quantidade de carbono bem como a existência ou não de binaridade nas estrelas da amostra são indícios das possíveis fontes da sobreabundância do carbono. A caracterização da amostra é essencial para o estudo da relação entre duas populações estelares aparentemente distintas: a de estrelas de carbono e a de estrelas pobres em metais enriquecidas em carbono. Para tanto, são utilizados catálogos de espectros assim como critérios fotométricos. A descoberta de alguma relação entre as duas populações pode ajudar a esclarecer o problema dos processos de enriquecimento de carbono, notado nas atmosferas estelares. A futura determinação dos parâmetros físicos das estrelas que compõem a amostra pode revelar detalhes sobre a história de formação estelar dos objetos das Nuvens e, portanto, vincular a evolução dessas duas galáxias-satélites à história da Via Láctea. / This project searches for signs of correlation between metallic stellar content, available for the Milky Way, and the metallicities indices obtained for the Magellanic Clouds. This comparison is supported by the traces on the formation of these three galaxies, that should have been left by each triplet members encounter. Since these crosses depend on the dynamical history, their fingerprints left by stellar content can estabilish constraints to the Galactic dynamical history. Spectroscopic data for a sample of carbon stars, obtained on the Magellanic Telescope, are used in this work. The amount of carbon, as the existence or not of binary stars in this sample, indicates possible sources of this element. The determination of sample properties is essential for studying the constraints between two stellar populations that are apparently distincts: carbon stars and carbon enhanced metal-poor stars. To do this, spectral catalogues and photometric criteria are used. Finding the correlations between both populations will bring some light to the unknown carbon enrichment processes that occured at the stellar atmospheres. Variability, emission lines, and binarity are studied for the sample. Stellar parameters are discussed, although there is no method applicable to this sample. By estimating the physical parameters of the stars presented in this sample and by analysing their carbon abundances, one can provide hints of the star formation history of objects in the Magellanic Clouds and therefore constraint the evolution of these satellite-galaxies to the Milky Way.

Comparamos duas nuvens moleculares, Rosette (RMC) e Monoceros R2 (Mon R2), localizadas no Complexo de Monoceros com o objetivo de estudar suas condições físicas relacionadas às primeiras fases da formação estelar. Tratam-se de regiões interessantes por apresentarem características que podem ser confrontadas com a hipótese de formação estelar provocada pela passagem de nuvens de altas velocidades atravessando o plano Galáctico (HVCs). Avaliamos as propriedades dessas nuvens por meio de mapas de vários traçadores da formação estelar com base em diferentes bandas espectrais visando estudar a estrutura de densidade das nuvens, bem como os objetos estelares jovens, em particular as fontes masers de H2O que apresentam características típicas de protoestrelas massivas. Nossa análise permitiu verificar algumas semelhanças entre RMC e Mon R2, mas também nos revelou diferenças interessantes. De uma forma geral há concordância entre AV, CO e emissão de poeira em 100 microns; RMC possui muitos clumps, entretanto poucos aglomerados e nebulosidades exceto uma única região HII principal (NGC2244) enquanto Mon R2 apresenta poucos clumps, vários aglomerados jovens e pequenas nebulosidades; em RMC há mais estrelas massivas, distribuídas uniformemente; Mon R2 tem poucas estrelas B, distribuídas em estruturas filamentárias com maiores índices de AV, do que em RMC; as fontes emissão maser apresentam cores IRAS compatíveis com candidatas a protoestrelas massivas, mas não parecem estar associadas a fontes de raios-X, sugerindo que masers estão relacionados à fase protoestelar, ao passo que fontes-X representam fase Pré-Sequência Principal. Concluímos que a distribuição de objetos e a estrutura das nuvens estão de acordo com as simulações dos modelos de HVCs. Porém, nossos resultados também são compatíveis com modelos alternativos, que simulam a dinâmica da Galáxia, para explicar o cenário de formação estelar no Complexo de Monoceros. / We compare two molecular clouds of the Monoceros Complex in order to study their physical conditions related to the early stages of star formation. The selected clouds, Rosette (RMC) and Monoceros R2 (Mon R2), are interesting regions due to their characteristics that may be confronted with the hypothesis of star formation triggered by high velocity clouds (HVCs) crossing through the Galactic plane. We evaluate the properties of these clouds using maps obtained on the basis of dierent spectral bands to trace the density of the clouds and the young stellar objects, in particular H2O masers that show typical features of massive protostars. This analysis allowed us to verify some similarities between RMC and Mon R2, but also revealed interesting dierences. In a general way there is an agreement between Av, CO and dust emission at 100 microns; RMC has many clumps, a few clusters and a single main nebulosity that is an HII region around NGC2244, while Mon R2 has a few clumps, several young clusters and small nebulosities. In RMC there is a large number of massive stars, uniformly distributed, while Mon R2 has a few B stars, distributed in lamentary structures with levels of Av higher than in RMC; maser sources have IRAS colors compatible with massive protostars candidates, but do not seem to be associated with X-ray sources, suggesting that masers are more related to the protostellar phase, while X-ray sources are related to pre main sequence phase. We conclude that the distribution of objects and the structure of the clouds are in accordance with the simulations of HVC models. However, our results are also compatible with alternative models of the Galaxy dynamics that explain the scenario of star formation in the Monoceros Complex.

Les bandes interstellaires diffuses (DIBs) représentent un mystère centenaire : aucune des centaines de bandes n'a pu être identifiée avec certitude avec un porteur spécifique, à l'exception très probable du cation buckminsterfullerène C60+. Il est obligatoire d'identifier et de quantifier la quantité de grosses molécules carbonées interstellaires qui sont très probablement responsables des DIBs : Les porteurs de DIB représentent probablement le plus grand réservoir de matière organique dans le Milieu interstellaire (MIS) et constituent un élément important de la chaîne des processus qui régissent le cycle interstellaire/stellaire. Jusqu'à récemment, la plupart des études liées aux DIBs avaient pour objectif unique l'identification des porteurs et, pour ce faire, se concentraient sur un nombre limité d'étoiles chaudes, distantes et rougies. Mon travail de recherche en thèse marque un tournant dans les méthodes et les objectifs associés aux DIBs, une évolution permise et motivée par le nombre croissant de relevés stellaires avec des instruments de plus en plus puissants à haute résolution spectrale. En effet, il est aujourd'hui possible de recueillir des quantités massives de données, tant du point de vue du nombre d'étoiles cibles que du point de vue du nombre de DIBs observées simultanément. Cela a ouvert la voie à de nouveaux types d'études, à des objectifs plus ambitieux et, surtout, à de nouvelles comparaisons potentielles avec les données de laboratoire. Ma thèse présente un grand nombre d'extractions de DIBs et quatre de ces nouvelles analyses :- Méthodes d'extraction et recherche de nouvelles DIBs.- Lien avec les propriétés physiques des nuages.- Constitution de bases de données pour la cartographie du MIS.- Tomographie des structures individuelles.- Des extractions de profils de DIBs basées sur des lignes de visée soigneusement sélectionnées, des études de leurs sous-structures et de leur variabilité spatiale et des contraintes subséquentes sur leurs porteurs moléculaires potentiels. / Diffuse interstellar bands (DIBs) represent a century-old mystery: none of the hundreds of bands could be identified with certainty with a specific carrier, at the very likely exception of the buckminsterfullerene cation C60+. Identifying and quantifying the amount of the large interstellar carbonaceous molecules that are very likely responsible for the DIBs is mandatory: DIB carriers likely represent the largest amount of organic matter in the Universe and are an important piece of the chain of processes that govern the interstellar/stellar cycle.Up to recently, most of the DIB studies have had as a unique goal the identification of the carriers, and to do so have been focusing on a limited number of hot, distant and reddened stars, using increasingly powerful instruments. This thesis marks a turning point in the methods and goals associated with the DIBs, an evolution allowed and motivated by the increasing number of stellar surveys with high multiplex instruments. As a matter of fact, it is possible today to gather massive amounts of data, both from the point of view of the number of target stars and from the point of view of the number of DIBs simultaneously observed. This has opened the way to new types of studies, more ambitious goals, and, importantly, new potential comparisons with laboratory data. This thesis presents a large number of DIB extractions and four of these novel analyses :- Methods of extraction and search for new DIBs.- Statistical studies of the link between DIB strengths and the physical properties of their hosting clouds.- Tomographic studies of the carriers on large and small spatial scales.- Line profile extractions based on carefully selected sightlines, studies of their substructures and spatial variability and subsequent constraints on their potential molecular carriers.

Les nuages sont une composante majeure du système Terre-océan-atmosphère car ils couvrent en moyenne 40% de la surface du globe. Ils contribuent à refroidir la troposphère car ils réfléchissent une part importante du rayonnement solaire (effet d’albédo), mais ils contribuent également à la réchauffer car ils réémettent une partie du rayonnement infrarouge terrestre (effet de serre naturel). La détermination des effets radiatifs des nuages a été identifiée par le GIEC comme l’une des sources principales d’incertitude sur la prévision du climat. Dans ce contexte, la question se pose quant au rôle de la variabilité décennale de ces propriétés nuageuses. Plus particulièrement, on cherche à identifier une possible évolution ou variation des propriétés des nuages et à comprendre l'impact de cette variabilité sur celle du climat régional (température au sol), et inversement : ici, la région d'étude est l'Europe. On sait en effet qu'au 1er ordre le climat régional européen est contrôlé par la circulation atmosphérique de grande échelle (Cattiaux et al., ou Cassou et al., 2005) mais celle-ci ne suffit pas à expliquer certaines anomalies ou extrêmes de température. Des anomalies de propriétés nuageuses sont donc une piste importante à étudier pour expliquer ce type d’événement extrême. Durant cette thèse, nous proposons donc d'appréhender cette question à partir d'observations spatiales et de simulations. Les observations seront celles de l'Aqua-Train : (1) l'instrument MODIS à bord du satellite Aqua permet, à l'aide d'algorithmes développés à la Nasa, de caractériser le forçage radiatif des nuages depuis 9 ans (i.e. leur capacité à refroidir ou à réchauffer); (2) le lidar du satellite CALIPSO couplé au radar du satellite CloudSat nous renseigne sur la structure verticale de ces mêmes couches nuageuses, leurs propriétés précipitantes et microphysiques depuis 7 ans. Les simulations utilisées pour compléter ces observations sont déjà existantes : elles ont été réalisées avec le modèle régional WRF et couvrent l'ensemble de l'Europe sur une période suffisamment longue pour pouvoir travailler sur la variabilité interannuelle à décennale des nuages. L’objet de la thèse est d'analyser à l'échelle régionale les relations entre les anomalies nuageuses et les anomalies de température. On cherchera à comprendre si les propriétés des nuages sont perturbées pour un régime synoptique donné et si ces perturbations peuvent expliquer certaines anomalies de température via leur effet radiatif direct ou plus indirect dans le cas des nuages convectifs. La démarche suivante sera appliquée : (1) Les observations et les simulations seront analysées conjointement pour mieux caractériser les propriétés nuageuses et surtout leur variabilité spatiale et interannuelle. (2) Ce travail de caractérisation des propriétés nuageuses devra se faire pour chaque régime synoptique identifié. Les saisons d'hiver et d'été seront caractérisées grâce aux régimes de l'Oscillation Nord Atlantique (NAO) (en collaboration avec le LSCE), tandis qu'un travail sur les régimes des saisons intermédiaires sera nécessaire. (3) Nous chercherons ensuite à comprendre comment ces propriétés typiques d’un régime sont perturbées. A régime fixé, nous tenterons de relier des anomalies de couverture nuageuse à des anomalies de température, et les périodes identifiées seront étudiées en détail pour comprendre quels sont les mécanismes qui permettent de passer d’une anomalie de l’un à une anomalie de l’autre : en d’autres termes, il s’agira d’estimer si le forçage radiatif de ces « anomalies nuageuses » peut conduire à l’anomalie de température détectée. Les comparaisons entre simulations et observations seront utiles pour analyser ces liens (existent dans les 2, ou uniquement dans l'un et pourquoi). Le travail de thèse pourra alors consister à réaliser de nouvelles simulations afin de mieux comprendre ces relations. / We characterize the seasonal and inter-annual variabilities of the clouds fraction profiles in both observations and simulation since they are critical to better assess the impact of clouds on climate variability. The spaceborne lidar onboard CALIPSO, providing cloud vertical profiles since 2006, is used together with a 23-year WRF simulation at 20 km resolution. A lidar simulator helps to compare consistently model with observations. The bias in observations due to the satellite under-sampling is first estimated. Then we examine the vertical variability of both occurrence and properties of clouds. It results that observations indicate a similar occurrence of low and high clouds over continent, and more high than low clouds over the sea except in summer. The simulation shows an overestimate (underestimate) of high (low) clouds comparing to observations, especially in summer. However the seasonal variability of the cloud vertical profiles is well captured by WRF. Concerning inter-annual variability, observations show that in winter, it is twice more important for high clouds than for low clouds, which is well simulated. In summer, the observed inter-annual variability is vertically more homogeneous while the model still simulates more variability for high clouds than for low clouds. The good behavior of the simulation in winter allows us to use the 23 years of simulation and 8 years of observations to estimate the time period required to characterize the natural variability of the cloud fraction profile in winter, i.e the time period required to detect significant anomalies and trends.

Graph convolutional networks (GCNs) showed promising results in learning from point cloud data. Applications of GCNs include point cloud classification, point cloud segmentation, point cloud upsampling, and more. Recently, the introduction of Deep Graph Convolutional Networks (DeepGCNs) allowed GCNs to go deeper, and thus resulted in better graph learning while avoiding the vanishing gradient problem in GCNs. By adapting impactful methods from convolutional neural networks (CNNs) such as residual connections, dense connections, and dilated convolutions, DeepGCNs allowed GCNs to learn better from non-Euclidean data. In addition, deep learning methods proved very effective in the task of point cloud upsampling. Unlike traditional optimization-based methods, deep learning-based methods to point cloud upsampling does not rely on priors nor hand-crafted features to learn how to upsample point clouds. In this thesis, I discuss the impact and show the performance results of DeepGCNs in the task of point cloud part segmentation on PartNet dataset. I also illustrate the significance of using GCNs as upsampling modules in the task of point cloud upsampling by introducing two novel upsampling modules: Multi-branch GCN and Clone GCN. I show quantitatively and qualitatively the performance results of our novel and versatile upsampling modules when evaluated on a new proposed standardized dataset: PU600, which is the largest and most diverse point cloud upsampling dataset currently in the literature.

We present a method for deriving the radiative effects of absorbing aerosols in cloudy scenes from satellite retrievals only. We use data of 2005–2007 from various passive sensors aboard satellites of the “A-Train” constellation. The study area is restricted to the tropical- and subtropical Atlantic Ocean. To identify the dependence of the local planetary albedo in cloudy scenes on cloud liquid water path and aerosol optical depth (AOD), we perform a multiple linear regression. The OMI UV-Aerosolindex serves as an indicator for absorbing-aerosol presence. In our method, the aerosol influences the local planetary albedo through direct- (scattering and absorption) and indirect (Twomey) aerosol effects. We find an increase of the local planetary albedo (LPA) with increasing AOD of mostly scattering aerosol and a decrease of the LPA with increasing AOD of mostly absorbing aerosol. These results allow us to derive the direct aerosol effect of absorbing aerosols in cloudy scenes, with the effect of cloudy-scene aerosol absorption in the tropical- and subtropical Atlantic contributing (+21.2±11.1)×10−3 Wm−2 to the global top of the atmosphere radiative forcing.

Statistical cloud schemes with prognostic probability distribution functions have become more important in atmospheric modeling, especially since they are in principle scale adaptive and capture cloud physics in more detail. While in theory the schemes have a great potential, their accuracy is still questionable. High-resolution three-dimensional observational data of water vapor and cloud water, which could be used for testing them, are missing. We explore the potential of ground-based remote sensing such as lidar, microwave, and radar to evaluate prognostic distribution moments using the “perfect model approach.” This means that we employ a high-resolution weather model as virtual reality and retrieve full three-dimensional atmospheric quantities and virtual ground-based observations. We then use statistics from the virtual observation to validate the modeled 3-D statistics. Since the data are entirely consistent, any discrepancy occurring is due to the method. Focusing on total water mixing ratio, we find that the mean ratio can be evaluated decently but that it strongly depends on the meteorological conditions as to whether the variance and skewness are reliable. Using some simple schematic description of different synoptic conditions, we show how statistics obtained from point or line measurements can be poor at representing the full three-dimensional distribution of water in the atmosphere. We argue that a careful analysis of measurement data and detailed knowledge of the meteorological situation is necessary to judge whether we can use the data for an evaluation of higher moments of the humidity distribution used by a statistical cloud scheme.

A simple way to diagnose fractional cloud cover in general circulation models is to relate it to the simulated relative humidity, and allowing for fractional cloud cover above a “critical relative humidity” of less than 100%. In the formulation chosen here, this is equivalent to assuming a uniform “top-hat” distribution of subgrid-scale total water content with a variance related to saturation. Critical relative humidity has frequently been treated as a “tunable” constant, yet it is an observable. Here, this parameter, and its spatial distribution, is examined from Atmospheric Infrared Sounder (AIRS) satellite retrievals, and from a combination of relative humidity from the ECMWF Re-Analyses (ERA-Interim) and cloud fraction obtained from CALIPSO lidar satellite data. These observational data are used to evaluate results from different simulations with the ECHAM general circulation model (GCM). In sensitivity studies, a cloud feedback parameter is analyzed from simulations applying the original parameter choice, and applying parameter choices guided by the satellite data. Model sensitivity studies applying parameters adjusted to match the observations show larger positive cloud-climate feedbacks, increasing by up to 30% compared to the standard simulation.

Ship tracks are regarded as the most obvious manifestations of the effect of anthropogenic aerosol particles on clouds (indirect effect). However, it is not yet fully quantified whether there are climatically relevant effects on large scales beyond the narrow ship tracks visible in selected satellite images. A combination of satellite and reanalysis data is used here to analyze regions in which major shipping lanes cut through otherwise pristine marine environments in subtropical and tropical oceans. We expect the region downwind of a shipping lane is affected by the aerosol produced by shipping emissions but not the one upwind. Thus, differences in microphysical and macrophysical cloud properties are analyzed statistically. We investigate microphysical and macrophysical cloud properties as well as the aerosol optical depth and its fine-mode fraction for the years 2005–2007 as provided for by retrievals of the two Moderate Resolution Imaging Spectroradiometer instruments. Water-cloud properties include cloud optical depth, cloud droplet effective radius, cloud top temperature, and cloud top pressure. Large-scale meteorological parameters are taken from ERA-Interim reanalysis data and microwave remote sensing (sea surface temperature). We analyze the regions of interest in a Eulerian and Lagrangian sense, i.e., sampling along shipping lanes and sampling along wind trajectories, respectively. No statistically significant impacts of shipping emissions on large-scale cloud fields could be found in any of the selected regions close to major shipping lanes. In conclusion, the net indirect effects of aerosols from ship emissions are not large enough to be distinguishable from the natural dynamics controlling cloud presence and formation.