Il est maintenant bien établi que les cirrus ont un impact important sur le climat. Cependant, l'estimation de cet effet est difficile car notre connaissance des propriétés microphysiques de ce type de nuage est encore incertaine. L'objectif de cette thèse est donc d’améliorer notre compréhension de la microphysique complexe du cirrus composé principalement de cristaux de glace de forme irrégulière et d'estimer ainsi un contenu en glace (ice water content, IWC) plus précis. Pour cela, nous avons développé un algorithme permettant de restituer le profil vertical d'IWC du cirrus. La méthodologie considère une synergie entre les mesures d'un lidar et celles d'un radiomètre infrarouge thermique (IRT) effectuées depuis le sol, via une méthode d'estimation optimale. Ce travail s’est déroulé en trois étapes: (1) Le contenu en glace intégré verticalement (ice water path, IWP) est estimé à partir des mesures passives IRT. (2) L'information sur la distribution verticale d'IWC à l'intérieur du nuage est obtenue avec les mesures actives du lidar. Cette restitution dépend fortement du rapport entre la rétrodiffusion et l'extinction des cristaux de glace obtenu avec un modèle microphysique. La fonction de phase du modèle utilisée pour définir ce rapport ne prend pas en compte le pic de rétrodiffusion. Nous montrons que cette hypothèse aboutit à des résultats non réalistes par rapport aux mesures IRT. (3) Par conséquent, les deux types d’informations sont combinées en synergie pour estimer, lors de la restitution des profils verticaux d'IWC, un facteur correctif permettant de rendre compte de ce pic de rétrodiffusion. Finalement, les résultats et les hypothèses associées sont discutés. / There is a broad consensus that cirrus clouds strongly influence the climate of the Earth. However, their net radiative effect is still poorly quantified nowadays due to an insufficient knowledge of their microphysical properties. This thesis aims to improve our understanding of the complex microphysics of this cloud type mainly composed of irregularly shaped ice crystals and thereby improve estimates of the ice water content (IWC). For this purpose, we developed an algorithm to retrieve vertical profiles of the IWC of cirrus clouds. The methodology combines the measurements of a ground-based lidar and a thermal infrared (TIR) radiometer in a common optimal estimation framework. It follows three steps: (1) An algorithm to retrieve the vertically integrated amount of ice (ice water path, IWP) from the passive TIR measurements is established. (2) The information about the vertical distribution of the IWC inside the cloud is obtained from the active lidar measurements. These retrievals strongly depend on the backscatter-to-extinction ratio of the ice crystals which is obtained from a bulk ice microphysical model. The scattering phase function of this model used to define the backscatter-to-extinction ratio assumes a flat ending without backscattering peak. We show that this assumption is unrealistic since it results in the retrieval of IWC profiles which are inconsistent with the TIR measurements. (3) Consequently, both types of measurements are combined in a synergistic algorithm allowing to estimate together with the IWC profiles a correction factor for the phase function in backscattering direction. Finally, the retrieval results and associated hypotheses are discussed.

Les nuages sont les principaux modulateurs du bilan radiatif terrestre et ils jouent un rôle prépondérant sur le climat de la Terre. Depuis plusieurs années, il est admis que la représentation des propriétés nuageuses est l'une des principales sources d’incertitude dans les modèles de prévision du climat et du temps météorologique. Dans ce travail de thèse, nous nous sommes concentrés sur les propriétés macrophysiques des atmosphères nuageuses et plus particulièrement sur leur structure verticale (caractère monocouche/multicouche, altitude et extension verticale des couches nuageuses). Pour cela, nous nous sommes basés sur les mesures multiangulaires effectuées dans la bande A de l’oxygène par le radiomètre POLDER embarqué sur la plateforme satellitaire PARASOL. Dans un premier temps, nous avons cherché à caractériser puis à distinguer de manière statistique les situations nuageuses monocouches et multicouches en construisant un arbre de décision. Ensuite, nous avons construit des paramétrisations permettant d'estimer les pressions de sommet et de milieu des nuages monocouches, ainsi que leur extension verticale. Enfin, des analyses statistiques réalisées sur cinq ans de données, complétées par des cas d'étude variés, ont permis de vérifier la validité de ces nouveaux produits géophysiques et d'en déterminer leurs limites. Cette étude a été rendue possible par la richesse des informations colocalisées de POLDER avec celles de deux instruments de télédétection active de la constellation de satellites A-Train, qui fournissent des informations précises, mais avec une faible couverture spatiale, sur le profil vertical de l'atmosphère. / Clouds are key modulators of the Earth's radiative balance and they play a major role in the climate of Earth. It has been recognized for several years that the representation of cloud properties is one of the main sources of uncertainty in climate and meteorological prediction models. In this thesis, we focused on the macrophysical properties of cloudy atmospheres and particularly on their vertical structure (single /multi-layered character, altitude and vertical extension of cloud layers). For this purpose, we relied on the multiangular measurements in the oxygen A band made by the radiometer POLDER on board PARASOL satellite platform. As a first step, we intended to characterize and to statistically distinguish monolayer and multilayer cloudy situations by designing a decision tree. Then, we built parameterizations in order to estimate top and middle pressures of monolayers clouds as well as their vertical extension. Finally, statistical analyses performed on five years of data, completed by various case studies, allowed us to check the validity of these new geophysical products and to determine their limits. This study was made possible by the richness of POLDER information collocated with two active sensors of the satellite constellation A- Train, which provide accurate information on the vertical profile of the atmosphere, but with a low spatial coverage.

Microphysical and dynamical interactions between aerosols and clouds are associated with some of the largest uncertainties in projections of future climate. Many possible aerosol effects on clouds have been suggested, but large uncertainties remain. In order to improve model projections of future climate, it is essential that we improve our quantitative understanding of anthropogenic aerosol effects. Several studies investigating interactions between satellite-observed aerosol and cloud properties have been published in recent years. However, the observed relationships are not necessarily due to aerosol effects on clouds. They may be due to cloud and precipitation effects on aerosol, meteorological covariation, observational data errors or methodological errors. An analysis of methodological errors arising through climatological spatial gradients is performed. For region sizes larger than 4°×4°, commonly used in the literature, spurious spatial variations in retrieved cloud and aerosol properties are found to introduce widespread significant errors to calculations of aerosol-cloud relationships. Small scale analysis prior to error-weighted aggregation to larger region sizes is recommended. Appropriate ways of quantifying relationships between aerosol optical depth (τ) and cloud properties are considered, and results are presented for three satellite datasets. There is much disagreement in observed relationships between τ and liquid cloud droplet number concentration and between τ and liquid cloud droplet effective radius, particularly over land. However, all three satellite datasets are in agreement about strong positive relationships between τ and cloud top height and between τ and cloud fraction (f_c). Using reanalysis τ data, which are less affected by retrieval artifacts, it is suggested that a large part of the observed f_c-τ signal may be due to cloud contamination of τ. General circulation model simulations further demonstrate that positive f_c-τ relationships may primarily arise due to covariation with relative humidity, and that negative f_c-τ relationships may arise due to scavenging of aerosol by precipitation. A new method of investigating the contribution of meteorological covariation to the observed relationships is introduced. Extratropical cyclone storm-centric composites of retrieved aerosol and cloud properties are investigated. A storm-centric description of the synoptics is found to be capable of explaining spurious f_c-τ relationships, although the spurious relationships explained are considerably smaller than observed relationships.