La qualité de l’air est un enjeu environnemental majeur de notre société. Cette thèse se focalise sur l’observation depuis l’espace du polluant gazeux majeur, l’ozone troposphérique. Les instruments satellitaires de nouvelle génération permettent de le mesurer journalièrement aux échelles régionales à globale. Cependant, les mesures d’un seul domaine spectral n’amène pas d’information suffisante pour quantifier les concentrations de l’ozone dans la couche limite atmosphérique, où se situe l’air que nous respirons. Le couplage des mesures de multiples domaines spectraux offre un fort potentiel pour mieux caractériser la pollution à l’ozone. La première approche multispectrale à 2 bandes combinant les mesures de l’infrarouge (IR) de IASI et l’ultraviolet (UV) de GOME-2 a été développée au laboratoire LISA, que j’ai intégré dans le cadre de ma thèse. L’objectif de mon travail de thèse est de franchir un pas supplémentaire dans la quantification de la pollution à l’ozone en incluant la bande visible (VIS) de Chappuis, afin de développer un couplage multispectral inédit à trois bandes spectrales UV+VIS+IR. Premièrement, j’ai mis au point une approche pour extraire l’information sur l’ozone des spectres VIS de GOME-2, en sélectionnant les longueurs d’onde affectées majoritairement par l’absorption à l’ozone et rejetant celles avec des absorptions d’autres gaz, ainsi que déterminant les variables d’ajustement conjoint de l’approche (profil vertical d’ozone, albédo de surface, shifts spectraux, etc) et les bases de données nécessaires pour simuler les spectres VIS (albédo de surface, spectroscopie).Ensuite, j’ai mis en œuvre la méthode multispectrale à trois bandes UV+VIS+IR, premièrement sur des observations simulées défilantes de IASI et GOME-2 et géostationnaires de MAGEAQ afin d’estimer l’amélioration en terme de sensibilité et précision, par rapport aux méthodes existantes. Dernièrement, j’ai appliqué l’approche multispectrale à 3 bandes à des observations satellitaires réelles de IASI et GOME-2. Des comparaisons avec d’autres données montrent des meilleures performances des nouvelles observations UV+VIS+IR pour caractériser des évènements de pollution à l’ozone, par rapport à l’approche à 2 bandes / Air quality is a major environmental issue for our society. This PhD thesis focuses on space borne observation of the major gaseous pollutant. Tropospheric ozone. New generation satellite instruments allow measuring it daily at regional to global scales. However, single spectral domain measurements do not provide enough information for quantifying ozone concentrations at the atmospheric boundary layer, where the air we breathe is located. Synergism of measurements from multiple spectral domains offers a great potential for better characterizing ozone pollution. The first 2-band multispectral approach combining infrared (IR) measurements of IASI and the ultraviolet (UV) of GOME-2 was developed at LISA laboratory, where I conducted my thesis. The goal of my thesis is to go a step further in the quantification of ozone pollution by including the visible (VIS) Chappuis band, in order to develop an unprecedented 3-band multispectral synergism UV+VIS+IR. First, I developed a approach for extracting ozone information from GOME-2 VIS spectra, by selecting the wavelengths mostly affected by ozone absorption and rejecting those with absorptions from other gases, as well as determining the joint adjustment variables of the approach (vertical ozone profile, surface albedo, spectral shifts, etc) and the databases needed to simulate the spectra VIS (surface albedo, spectroscopy).Then, I implemented the multispectral method with three bands UV+VIS+IR, first on simulated low-orbit and geostationary observations from respectively IASI and GOME-2 and MAGEAQ to estimate the improvement in terms of sensitivity and precision, as compared to existing methods. Last, I applied the 3-band multispectral approach to real satellite observations of IASI and GOME-2. Comparisons with other data shows enhanced performances of the new UV+VIS+IR observations to characterize ozone pollution events, with respect to the 2-band approach

This thesis deals with the global aspects of plasmaconvection in the magnetosphere as measured by the low-altitudepolar orbiting Astrid-2 and FAST satellites. The major focus ison the electric field measurements, but they are alsocomplemented by magnetic field, ion and electron particle data,which is fundamental for the understanding of theelectrodynamics of the high-latitude auroral ovals and polarcap, which are the regions analysed here. The essential subjectof this thesis is the so-called magnetic reconnection processthat drives plasma convection in the Earth's magnetosphere. Itis shown that the ionospheric convection, being intimatelycoupled to the magnetospheric convection, responds in about15-25 min depending on geomagnetic activity after the arrivalof the solar wind at the magnetopause. It also responds on alonger time scale, around 55-75 min, which is interpreted asthe unloading of solar wind energy previously stored in thelarge-scale current system of the magnetotail. These resultshave been found previously using ionospheric parameters such asthe auroral electrojet AL index. What is new is that these sameresults are reproduced by using a discrete set of cross-polarpotential measurements. Using an extensive set of electric andmagnetic field data combined with particle precipitation datafrom the FAST satellite, it is shown that the reconnectionprocess can also be applied to explain features of sunwardplasma convection in the polar cap with a likely antiparallelmerging site in the lobe magnetopause region. The lobereconnection is found to depend strongly on IMF Byand to coexist with dayside subsolar merging.Finally, a comparison is performed between the Weimer electricfield model and Astrid-2 electric field data. Empiricalelectric field models are important in understanding thecomplete convection pattern at any one time, something, whichcannot be provided by measurements from single satellites. <b>Keywords:</b>Satellite measurements, electric fields,magnetosphere, magneticreconnection, plasma convection, lobecell convection, empirical electric field models.

Measurements by OSIRIS, an infrared imaging system that measures emission from excited-state molecular oxygen, are used to retrieve mesospheric ozone through a photochemical model. The design of the model and the spectral data used in the calculation of pho- tochemical production coe±cients are presented. The model has been run in a variety of modes to provide comparisons with measurements of excited-state molec- ular oxygen; it has been found that the model results are in excellent agreement with measurements. The model is used in conjunction with a retrieval scheme to estimate the con- centrations of mesospheric ozone from measurements made on a satellite platform. An analysis of the sensitivity of retrieved ozone to key model parameters is done, and it is shown that uncertainty in several kinetic reaction rates can signi¯cantly change the retrieved results. Comparisons are made for the ozone retrieved in this work and that from several other instruments that make similar measurements. The retrieved concentrations are shown to be in excellent agreement with results from other instruments. The present work provides a climatological database of mesospheric ozone and will provide useful comparisons with concurrent measurements of mesospheric ozone. The new data obtained in this work are in good agreement with expected results, and it is shown that the unprecedented high-spatial resolution of OSIRIS reveals interesting features that should be further investigated.

Measurements by OSIRIS, an infrared imaging system that measures emission from excited-state molecular oxygen, are used to retrieve mesospheric ozone through a photochemical model. The design of the model and the spectral data used in the calculation of pho- tochemical production coe±cients are presented. The model has been run in a variety of modes to provide comparisons with measurements of excited-state molec- ular oxygen; it has been found that the model results are in excellent agreement with measurements. The model is used in conjunction with a retrieval scheme to estimate the con- centrations of mesospheric ozone from measurements made on a satellite platform. An analysis of the sensitivity of retrieved ozone to key model parameters is done, and it is shown that uncertainty in several kinetic reaction rates can signi¯cantly change the retrieved results. Comparisons are made for the ozone retrieved in this work and that from several other instruments that make similar measurements. The retrieved concentrations are shown to be in excellent agreement with results from other instruments. The present work provides a climatological database of mesospheric ozone and will provide useful comparisons with concurrent measurements of mesospheric ozone. The new data obtained in this work are in good agreement with expected results, and it is shown that the unprecedented high-spatial resolution of OSIRIS reveals interesting features that should be further investigated.

<p>This thesis deals with the global aspects of plasmaconvection in the magnetosphere as measured by the low-altitudepolar orbiting Astrid-2 and FAST satellites. The major focus ison the electric field measurements, but they are alsocomplemented by magnetic field, ion and electron particle data,which is fundamental for the understanding of theelectrodynamics of the high-latitude auroral ovals and polarcap, which are the regions analysed here. The essential subjectof this thesis is the so-called magnetic reconnection processthat drives plasma convection in the Earth's magnetosphere. Itis shown that the ionospheric convection, being intimatelycoupled to the magnetospheric convection, responds in about15-25 min depending on geomagnetic activity after the arrivalof the solar wind at the magnetopause. It also responds on alonger time scale, around 55-75 min, which is interpreted asthe unloading of solar wind energy previously stored in thelarge-scale current system of the magnetotail. These resultshave been found previously using ionospheric parameters such asthe auroral electrojet AL index. What is new is that these sameresults are reproduced by using a discrete set of cross-polarpotential measurements. Using an extensive set of electric andmagnetic field data combined with particle precipitation datafrom the FAST satellite, it is shown that the reconnectionprocess can also be applied to explain features of sunwardplasma convection in the polar cap with a likely antiparallelmerging site in the lobe magnetopause region. The lobereconnection is found to depend strongly on IMF B_yand to coexist with dayside subsolar merging.Finally, a comparison is performed between the Weimer electricfield model and Astrid-2 electric field data. Empiricalelectric field models are important in understanding thecomplete convection pattern at any one time, something, whichcannot be provided by measurements from single satellites.</p><p><b>Keywords:</b>Satellite measurements, electric fields,magnetosphere, magneticreconnection, plasma convection, lobecell convection, empirical electric field models.</p>

Les mécanismes de formation des nuages de glace arctiques durant la nuit polaire sont encore mal définis en raison de l’absence d’observations et de l’éloignement de cette région. Pourtant, leur influence sur les conditions météorologiques et sur le climat dans l’hémisphère nord est d’une importance primordiale ; et les connaissances sur la modification de leurs propriétés, liées à des processus d’interaction aérosol-nuage, doivent être améliorées. Les fortes concentrations d’aérosols en Arctique durant la nuit polaire sont associées au transport des aérosols anthropiques des latitudes moyennes jusqu’au pôle Nord. Les observations et les modèles montrent que cela peut conduire à un transport important de particules d’aérosol acidifiées. Les mesures en laboratoire et in situ montrent qu’à basse température (< -30°C), le revêtement d’acide sur les noyaux glaçogènes (IN) peut réduire leurs propriétés de nucléation de la glace. Par conséquent, leur concentration est réduite dans ces régions entraînant une plus faible concentration de plus gros cristaux de glace en raison d’une diminution de la compétition pour une humidité disponible similaire. De nombreuses mesures de terrain et par télédétection par satellite (CloudSat et CALIPSO) révèlent l’existence de deux types de nuages de glace (TIC) en Arctique durant la nuit polaire. Les nuages de glace de type 1 (TIC-1) ne sont visibles que par le lidar tandis que les nuages de glace de type 2 (TIC-2) sont perçus à la fois par le lidar et le radar. Les TIC-2 sont divisés en TIC-2A et TIC-2B. Les TIC-2A sont recouverts d’une fine couche de petits cristaux de glace non-précipitant (invisible par le radar) (TIC-1), tandis que les TIC-2B ne le sont pas. Ils sont caractérisés par une faible concentration de gros cristaux de glace. On suppose que la microstructure des TIC-2B est liée à l’acidification des aérosols. Pour vérifier cette hypothèse, des études de cas et des approches statistiques ont été combinées afin d’analyser le transport des aérosols et les propriétés des nuages de glace en Arctique. La première partie de la thèse enquête sur les propriétés microphysiques des TIC-1/2A et TIC-2B, en analysant des mesures aéroportées et satellitaires de cas spécifiques observés durant la campagne de mesures ISDAC (Alaska, Avril 2008). Pour la première fois, les microstructures des TIC-1/2A et TIC-2B en Arctique sont comparées en utilisant les observations in-situ des nuages. (...) La deuxième partie de la thèse enquête sur l’origine des masses d’air formant deux cas spécifiques de TICs ISDAC : TIC-1/2A (1 Avril 2008) et TIC-2B (15 Avril 2008), en utilisant des outils de trajectoire et des données satellitaires. / Arctic ice cloud formation during winter is poorly understood mainly due to the lack of observations and the remoteness of this region. Yet, their influence on Northern Hemisphere weather and climate is of paramount importance, and the modification of their properties, linked to aerosol-cloud interaction processes, needs to be better understood.Large concentration of aerosols in the Arctic during winter is associated to long-range transport of anthropogenic aerosols from the mid-latitudes to the Arctic. Observations and models show that this may lead to a significant transport of acidified aerosol particles. Laboratory and in-situ measurements show that at cold temperature (< -30°C), acidic coating may reduce the ice nucleating properties of ice nuclei (IN). Therefore, the IN concentration is reduced in these regions, resulting to a smaller concentration of larger ice crystals and because of the reduced competition for the same available moisture.Extensive measurements from ground-based sites and satellite remote sensing (CloudSat and CALIPSO) reveal the existence of two Types of Ice Clouds (TICs) in the Arctic during the polar night and early spring. The first Type of Ice Clouds (TIC-1) are visible only by the lidar while the second Type of Ice Clouds (TIC-2) are detected by both the lidar and radar. TIC-2 are divided into TIC-2A and TIC-2B. TIC-2A are topped by a cover of non-precipitating very small (radar-unseen) ice crystals (TIC-1), while TIC-2B are not. They are characterized by a low concentration of large ice crystals. It is hypothesized that the observed low concentration of large ice crystals, leading to precipitation (e.g. cloud regime TIC-2B), is linked to the acidification of aerosols. To check this, we are combining case studies and statistical approaches to analyse aerosol transport and cloud properties in the Arctic.The first part of the thesis investigate the microphysical properties of TIC-1/2A and TIC-2B, analysing airborne in-situ and satellite measurements of specific cases observed during the ISDAC campaign (Alaska, April 2008). For the first time, Arctic TIC-1/2A and TIC-2B microstructures are compared using in-situ cloud observations. Results show that the differences between them are confined in the upper part of the clouds where ice nucleation occurs. TIC-2B were characterized by fewer (< 10 L-1) and larger (> 110 μm) ice crystals, a larger ice supersaturation (> 15 %) and a fewer ice nuclei (IN) concentration (< 2 order of magnitude) when compared to TIC-1/2A. Ice crystal growth in TIC-2B clouds seems explosive whereas it seems more gradual in TIC-1/2A. It is hypothesized that these differences are linked to the number concentration and the chemical composition of aerosols. The second part of the thesis investigate the origin of air masses forming two specific cases ; TIC-1/2A (1 April 2008) and TIC-2B (15 April 2008), using trajectory tools and satellite data.(...)

This thesis describes laboratory measurements of the complex refractive index of volcanic ash particles. These measurements are needed to model the radiative impact of volcanic ash, vital for accurate satellite remote sensing. Three experimental methods have been developed, and the results for the complex refractive index and optical properties of a wide range of volcanic ash samples are presented. Measurements were made of the spectral transmission of radiation through suspended volcanic ash particles inside an aerosol cell, using a Fourier transform spectrometer at infrared wavelengths and two diffraction grating spectrometers covering ultraviolet, visible, and near-infrared wavelengths. In addition to the optical measurements, a suite of sampling and sizing instruments were connected downstream of the aerosol cell to measure the particle size distribution. The method was calibrated using two quartz samples. Mass extinction coefficients for nine volcanic ash samples, at 0.3-14 μm, are presented and show considerable variation. These variations are linked to the composition of the samples, measured using X-ray fluorescence (XRF) analysis. The complex refractive index, at 0.3-14 μm, of the two quartz samples and two samples of volcanic ash from the 2010 Eyjafjallajökull eruption were retrieved from the extinction measurements. The forward model used Mie theory and a classical damped harmonic oscillator (CDHO) model to represent the complex refractive index of the samples in terms of a finite set of band parameters, as well as the real refractive index of the sample in the small wavelength limit. Previous studies have shown that there is a redundancy in the retrievals between the band strength parameters and the real refractive index in the small wavelength limit, which can lead to spurious values for the retrieved complex refractive index. This problem was overcome by using an independent measurement of the real refractive index at a visible wavelength, to constrain the model parameter of the real refractive index in the short wavelength limit. Independent measurements of the complex refractive index at visible wavelengths are also important because the extinction produced at these wavelengths is highly sensitive to the particle size distribution, and any uncertainty in the measured size distribution will contribute to significant systematic error in the refractive index retrieved from extinction. The retrieved spectral complex refractive index of Eyjafjallajökull ash was applied using the ORAC retrieval scheme to measurements of the 2010 Eyjafjallajökull eruptionmade by theMODIS instrument aboard NASA's Terra satellite. Significant difference were found in the retrieved plume parameters of optical path, effective radius, and plume altitude, compared to assuming a literature measurement for the refractive index of pumice. For three discrete visible wavelengths (450, 546.7, and 650 nm) an optical microscope was used to make measurements of the complex refractive index of the volcanic ash samples. The long-established Becke line method was used to measure the real refractive index of the samples. For the imaginary refractive index, a new and novelmethod was developed involving measurements of the attenuation of light in individual particles. A strong linear correlation was found between the SiO₂ content of the samples and both their real and imaginary refractive indices at the visible wavelengths investigated. Furthermore, from the XRF compositional analysis of the samples values were calculated for the ratio of non-bridging oxygen atoms per tetrahedral cation (NBO/T), and it was found that NBO/T was an even stronger predictor of real refractive index at visible wavelengths. The optical microscope measurements could only be applied to particles with a radius larger than 10 μm. A new refractometer method was investigated for retrieving the real refractive index of submicron particles from colloidal reflectance measurements close to the critical angle in an internal reflection configuration. A coherent scattering model (CSM) was used to model the coherent reflection from a half-space of monodisperse or polydisperse particles, and a simple extension of the model is presented to properly account for the modified size distribution at the interface in an internal reflection set-up. A rigorous sensitivity analysis was performed to determine how experimental uncertainties propagate into uncertainty associated with the retrieved real refractive index, and the uncertainty due to non-spherical effects was estimated using T-matrix methods. Experimental reflectance data at a wavelength of 635 nm were obtained for spherical monodisperse polystyrene calibration particles, a polydisperse sand sample, and a polydisperse volcanic ash sample. The retrieved values for the real refractive index agreed, within propagated uncertainties, with values measured using other techniques. The method is shown to be a viable technique for measuring the real refractive index of small quantities of submicron particles, and can also retrieve the concentration and size of particles.