This dissertation deals with spectral measurements of solar radiation in the visible and near infrared wavelength range. The data were collected during a field campaign on the Cape Verde Islands in January / February 2008 within the DFG research group SAMUM 2 (Saharan Mineral Dust Experiment). During this campaign airborne measurements of upward radiances and irradiances were performed over aerosol layers. Since the Cape Verde Islands are in the advection area of air masses from the Sahara region northeast of the islands and from regions with bush fires from the southeast, the sampled aerosol mainly consists of mineral dust, biomass burning smoke or a mixture of both. These radiation measurements and airborne lidar measurements of aerosol extinction coefficients were used to calculate the dust radiative forcing at the top of atmosphere with an one-dimensional radiative transfer model. This required the spectral surface albedo and aerosol optical properties, determined by model retrievals. The dependence of the calculated dust radiative forcing on the aerosol optical thickness was used to distinguish between aerosol distributions with mineral dust only or mixed with biomass burning smoke. This mainly model-based method was compared with another mainly measurement-based method, which requires the net radiation at the flight altitude and its dependence on the aerosol optical thickness to distinguish between the different aerosol distributions. The mainly model-based method shows no differences between the calculated radiative forcings of aerosols mainly consisting of mineral dust and those mixed with biomass burning smoke due to high uncertainties. In contrast to the mainly model-based method, the mainly measurement-based method shows clear differences between aerosols with and without biomass burning smoke. Thus the mainly measurement-based method is the preferred method, because it omits the retrieval of the aerosol optical properties, which leads to high uncertainties, in contrast to the mainly model-based method.

This study is focused on the identification of the sources and processes affecting levels and composition of PM10 and PM2.5 from air quality monitoring networks in Eastern Spain. This is a multidisciplinary study. Time series of TSP and PM10 recorded from 1996 to 2000 are interpreted to assess the role of local, regional and distant sources in PM levels in this region. To this is end, the influence of the meteorology on PM levels recorded at nineteen rural, urban and industrial monitoring stations is studied by means of synoptic charts, back-trajectories and local meteorological variables. The impact of African dust transport on PM levels is investigated by means of simulations in the SKIRON forecast system and by TOMS satellite observations. Moreover, a chemical characterisation of PM10, PM2.5 and TSP size segregated fractions and a daily PM10 and PM2.5 source apportionment by two receptor modelling techniques are performed. For this purpose, 24-hour samples of PM10 were collected for 16, 12 and 5 months at rural, urban and industrial sites, respectively. PM2.5 was sampled at the urban site simultaneously with PM10. The results showed that day-to-day variations in PM levels are highly influenced by the meteorology. From March to October, PM levels at rural, urban and industrial sites vary as a function of the concatenation of Atlantic air mass advections (Atlantic episodes with low PM levels) and regional circulations (Regional events with high PM and O3 levels) which induce the PM transport from urban/industrial to rural sites and the ageing of polluted air masses in the Western Mediterranean. From November to February low PM levels are recorded at rural sites, and variations in PM levels at urban/industrial sites are governed by the successive occurrence of Atlantic episodes and Local urban/industrial pollution events (with high PM, NOx and CO levels). The African dust events mainly occur from January to October and induce very high PM levels in all sites. As a result of the frequency and intensity of the Regional episodes, PM levels at rural sites undergo a seasonal evolution with a summer maximum. The difference between PM levels at urban and rural sites presents an autumn-winter maximum owing to the occurrence of intensive Local urban pollution events. The meteorological context in which the aforementioned episodes occur is discussed, with special emphasis on the synoptic scenarios giving rise to the African dust outbreaks in the different seasons. Levels of PM components at the different study sites were compared. The seasonal evolution and the grain size distribution of these components was studied and the form of occurrence was determined. The chemical characterisation shows that high levels of natural mineral dust components (e.g. Al, Fe, Mg, Ti, Sr, Ca) are simultaneously recorded in PM10 and PM2.5 at all sites during African episodes. Other interesting findings are: 1) a marked seasonal evolution of nitrate levels and grain size distribution due to the occurrence of ammonium-nitrate in autumn-winter, 2) an excess of Na with respect to the Na/Cl marine ratio in summer owing to reactions of acids with sea salt, and 3) high background levels of ammonium-sulphate in summer. Moreover, the levels of elemental, organic and mineral carbon were determined. At the rural site, the PM10 annual mean reaches 22µg/m3, the main contributions being secondary particles from industrial emissions (27% of PM10), vehicle exhausts (14-23%), natural mineral dust (23%) and sea spray (5-9%). At the urban kerbside station, the PM10 annual mean reaches 49µg/m3, the main contributions being vehicle exhausts (35-45% of PM10), secondary particles from industrial emissions (24-31%), natural + road dust (25%) and sea spray (4-6%). At the urban kerbside station, the PM2.5 annual mean reaches 34µg/m3, the main contributions being vehicle exhausts (41-53% of PM2.5), secondary particles from industrial emissions (29-35%) and mineral dust (9-11%). These results have important implications for the implementation of the PM10 EU standards. The natural load in ambient PM10 levels in Eastern Spain accounts for 30-40% of the 2010 EU annual PM10 limit value (20µg/m3). In this region it will not be easy to meet this limit value given that the annual PM10 levels are in the range 17-20µg/m3 at rural, 30-45µg/m3 at urban and 45-60µg/m3 at industrial sites. The high background levels of PM10 in this region are favoured by the specific orographic and meteorological context of the Mediterranean and by the high load of mineral dust caused by the soil re-suspension and the frequent occurrence of African dust events. At rural sites, 2-5 exceedances of the EU daily PM10 limit value (50µg/m3) are recorded every year during African dust outbreaks. On average, 15 African induced and 40-80 non-African induced exceedances of the EU daily PM10 limit value are recorded every year at the urban kerbside stations. The parameter selected for PM monitoring is a key factor. Most of the PM species resulting from combustion and vehicle exhaust emissions have a fine size distribution (<2.5µm), with exception of nitrate in summer, whereas sea spray and mineral dust present a coarse size distribution. The interference of African dust in the PM monitoring is significantly reduced when PM2.5 instead of PM10 is monitored. Natural mineral dust concentrations during African episodes are in the ranges 20-30µg/m3 in PM10 and 10-15µg/m3 in PM2.5. However, PM2.5 is not a suitable parameter for PM monitoring in all environments. The selection of PM10 or PM2.5 should be conditioned by the type of anthropogenic activity. At urban sites, PM2.5 contains mainly vehicle exhaust products, whereas the road traffic dust principally occurs in the 2.5-10µm fraction. PM2.5 is not a suitable parameter for monitoring some industrial activities (e.g. ceramics, cement production or mining) with primary PM emissions in the 2.5-10µm range.

Im Rahmen der vorliegenden Arbeit wurden die Eigenschaften von atmosphärischen Aerosolpartikeln im tropischen Regenwald des Amazonasgebietes bestimmt. Dazu wurden die Daten einer fast einjährigen Lidarmesskampagne ausgewertet und diskutiert. Die Messungen wurden mit einem automatischen Mehrwellenlängen-Polarisations-Raman-Lidar im zentralen Amazonasbecken nahe Manaus, Brasilien, im Zeitraum von Januar bis November 2008 durchgeführt. Somit konnten erstmalig optische und mikrophysikalische Aerosoleigenschaften im Amazonasgebiet während der Regenzeit (ca. Dezember-Mai) und Trockenzeit (ca. Juni-November) höhenaufgelöst charakterisiert werden. Einleitend werden die meteorologischen Bedingungen im Amazonasgebiet erläutert und eine Literaturübersicht über Aerosolforschung in dieser Region gegeben. Das Messgerät sowie verschiedene Kalibrier- und Korrekturschemen, die zur Datenauswertung notwendig sind, werden vorgestellt. Auch Vergleiche mit anderen Messgeräten werden diskutiert. Diese zeigen, dass die aus den Lidarmessungen abgeleiteten Parameter von hoher Qualität sind. Anhand von Fallstudien werden mit Hilfe von Rückwärtstrajektorien und Satellitenmessungen typische Aerosolbedingungen am Messstandort diskutiert. Um die generellen Unterschiede zwischen Regen- und Trockenzeit zu quantifizieren, wird eine statistische Auswertung aller analysierten Lidarmessungen präsentiert. Die Analyse der Lidardaten zeigt, dass während der Regenzeit im Amazonasgebiet in ca. der Hälfte aller Fälle sehr saubere Bedingungen mit einer Aerosol Optischen Dicke (AOD) von weniger als 0.05 (bei 532 nm) vorherrschen können. Allerdings wurde in ca. 30% aller analysierten Fälle im Zeitraum von Januar bis Mai auch afrikanisches Aerosol, vornehmlich Saharastaub und Biomasseverbrennungsaerosol (BBA), am Messstandort detektiert. Dabei dominierte meist BBA die Aerosolpopulation, wie die Depolarisationsmessungen zeigten. In der Trockenzeit ist die Atmosphäre im Amazonasbecken hauptsächlich mit BBA aus Südamerika belastet. Daher ist die AOD im Durchschnitt um einen Faktor drei größ er als in der Regenzeit. BBA wurde zu dieser Jahreszeit regelmäßig bis zu einer Höhe von 4-6 km detektiert. Basierend auf den vorgestellten Langzeitmessungen werden erstmalig die optischen Eigenschaften von südamerikanischem BBA statistisch analysiert und diskutiert. / Continuous lidar measurements were performed in the Amazon rain forest for almost one year in 2008. The results of the automated multiwavelength-Raman-polarization lidar observations were presented in this dissertation. These measurements are the first long-term observations of the vertical aerosol structure ever made in the Amazon Basin. The advanced lidar observations were conducted 60 km north of Manaus in the central northern part of Amazonia. The area is widely covered with pristine rain forest. A HYSPLIT backward-trajectory analysis showed that the observations were representative on a regional scale for the central northern part of the Amazon rain forest. The general weather conditions in this region are characterized by a wet (December-June) and a dry season (July-November). During the dry season, a high fire activity occurs in Amazonia, which heavily influences the atmospheric conditions. With the lidar instrument, vertical profiles of the particle backscatter coefficient at 355, 532, and 1064 nm, of the particle extinction coefficient at 355 and 532 nm, and of the particle linear depolarization ratio at 355 nm can be determined. The results from the long-term lidar observations performed in Brazil contain a lot of new information about the aerosol conditions in the central northern Amazon Basin and corroborate certain findings from former aerosol measurements in Amazonia. It was shown for the first time that advection of Saharan dust together with biomass burning aerosol (BBA) from Africa occurred regularly throughout the wet season. In about one third (32%) of all lidar observations during the wet season, African aerosol was dominating the optical aerosol properties in Amazonia. The analysis of the vertical aerosol structure during such events revealed that the African aerosol arriving in the central northern Amazon Basin was usually trapped in the lowermost 3-3.5 km of the troposphere. To quantify the amount of Saharan dust and African smoke transported towards the lidar site, the dust contribution to the measured optical aerosol properties was separated by means of the measured particle depolarization ratio. This study led to the result that in about one half of the cases with African aerosol advection, smoke particles contributed to more than 50% to the total Aerosol Optical Depth (AOD). The smoke transport from Africa towards Amazonia occurred predominantly between January and April when the fire activity in Central Africa was highest. BBA is thus a major constituent of the aerosol plumes that are regularly transported from Africa towards Amazonia. This is a key finding of the presented study. During clean conditions, an AOD (532 nm) of less than 0.05 was observed and the aerosol was trapped in the lowermost 2 km of the troposphere. However, the analysis of the long-term data set revealed that these clean atmospheric conditions occurred in only 48% of all wet-season cases. One example for such background conditions was intensively discussed and it was shown that a major meso-scale rain event occurred in the Amazon region at the same time. This precipitation event was possibly partly responsible for the very low aerosol load. Two case studies from the dry season were presented for which BBA dominated the optical properties. In the first case, a comparable high aerosol load (AOD of 0.41) prevailed while in the second one, a medium aerosol load (AOD of 0.15) was observed. Aged BBA advected from regions south of the lidar site were identified to be the dominant aerosol species for both cases. However, very different geometrical, optical and microphysical properties of BBA (e.g., vertical layering, lidar ratio, Ångström exponent, effective radius, SSA) were observed on both days. In the first case, aerosol was present up to about 4.5 km. Extinction-related Ångström exponent s of about 1 and lidar ratios between 70 and 90 sr were found at different heights for the smoke aerosol. The BBA was highly absorbing (SSA of 0.81) at heights of the highest RH (85%), whereas above under dry conditions (RH=50%) only moderate absorption (SSA of 0.93) was detected. In the second case, smoke was detected up to 4.5 km, and Ångström exponent of about 2 and lidar ratios of 45-55 sr were measured in the aerosol layers. The BBA was only moderately absorbing indicated by SSA values between 0.92 and 0.94. The reason for the differences in the smoke properties could be the shorter travel time to the lidar site (<24 h), different aging processes (e.g., cloud/rain processing), or different burning conditions. In both cases, no depolarizing effects of the BBA could be observed. The strong contrast between the aerosol conditions in the dry season and the wet season were confirmed by the statistical analysis of all lidar observations in 2008. Due to the high BBA concentration in the atmosphere, the mean AOD of the dry season was found to be a factor of 3 higher than the mean AOD of the wet season (0.26 compared to 0.08 at 532 nm). Maximum AOD values were less than 0.55 (at 532 nm) and hence show that the lidar location was not in the direct vicinity of fire events. In only 7% of all cases in the dry season 2008, an AOD below 0.1 was observed. Also the maximum extinction and backscatter coefficient values in the dry season 2008 were 2-3 times higher than during the wet season of this year. The vertical aerosol distributions differ also significantly between the two seasons. In the wet season, the aerosol was mostly trapped in the lowermost 2.5 km, while in the dry season aerosol typically reached up to 4.5 km. Aerosol was occasionally detected up to 6.14 km in the dry season. The majority of the aerosol (95% of the AOD), however, was found to be on average below 2.3 km in the wet season and below 3 km in the dry season. During the wet season, lofted aerosol layers and multiple aerosol stratification was less frequent than in the dry season. The extent of BBA plumes during the dry season showed no correlation to the ML top height. Virtually uniform smoke haze layers were observed up to the AL top. Thus, pyro-convection and/or cloud-related mixing seem to be the major processes for the vertical distribution of BBA.

An accurate and generalized global retrieval algorithm from satellite observations is a prerequisite to understand the radiative effect of atmospheric aerosols on the climate system. Current operational aerosol retrieval algorithms are limited by the inversion schemes and suffering from the non-uniqueness problem. In order to solve these issues, a new algorithm is developed for the retrieval of non-spherical dust aerosol over land using multi-angular radiance and polarized measurements of the POLDER (POLarization and Directionality of the Earth’s Reflectances) and wide spectral high-resolution measurements of the MODIS (MODerate resolution Imaging Spectro-radiometer). As the first step to account for the non-sphericity of irregularly shaped dust aerosols in the light scattering problem, the spheroidal model is introduced. To solve the basic electromagnetic wave scattering problem by a single spheroid, we developed an algorithm, by transforming the transcendental infinite-continued-fraction-formeigen equation into a symmetric tri-diagonal linear system, for the calculation of the spheroidal angle function, radial functions of the first and second kind, as well as the corresponding first order derivatives. A database is developed subsequently to calculate the bulk scattering properties of dust aerosols for each channel of the satellite instruments. For the purpose of simulation of satellite observations, a code is developed to solve the VRTE (Vector Radiative Transfer Equation) for the coupled atmosphere-surface system using the adding-doubling technique. An alternative fast algorithm, where all the solid angle integrals are converted to summations on an icosahedral grid, is also proposed to speed-up the code. To make the model applicable to various land and ocean surfaces, a surface BRDF (Bidirectional Reflectance Distribution Function) library is embedded into the code. Considering the complimentary features of the MODIS and the POLDER, the collocated measurements of these two satellites are used in the retrieval process. To reduce the time spent on the simulation of dust aerosol scattering properties, a single-scattering property database of tri-axial ellipsoid is incorporated. In addition, atmospheric molecule correction is considered using the LBLRTM (Line-By-Line Ra- diative Transfer Model). The Levenberg-Marquardt method was employed to retrieve all the interested dust aerosol parameters and surface parameters simultaneously. As an example, dust aerosol properties retrieved over the Sahara Desert are presented.

This dissertation deals with spectral measurements of solar radiation in the visible and near infrared wavelength range. The data were collected during a field campaign on the Cape Verde Islands in January / February 2008 within the DFG research group SAMUM 2 (Saharan Mineral Dust Experiment). During this campaign airborne measurements of upward radiances and irradiances were performed over aerosol layers. Since the Cape Verde Islands are in the advection area of air masses from the Sahara region northeast of the islands and from regions with bush fires from the southeast, the sampled aerosol mainly consists of mineral dust, biomass burning smoke or a mixture of both. These radiation measurements and airborne lidar measurements of aerosol extinction coefficients were used to calculate the dust radiative forcing at the top of atmosphere with an one-dimensional radiative transfer model. This required the spectral surface albedo and aerosol optical properties, determined by model retrievals. The dependence of the calculated dust radiative forcing on the aerosol optical thickness was used to distinguish between aerosol distributions with mineral dust only or mixed with biomass burning smoke. This mainly model-based method was compared with another mainly measurement-based method, which requires the net radiation at the flight altitude and its dependence on the aerosol optical thickness to distinguish between the different aerosol distributions. The mainly model-based method shows no differences between the calculated radiative forcings of aerosols mainly consisting of mineral dust and those mixed with biomass burning smoke due to high uncertainties. In contrast to the mainly model-based method, the mainly measurement-based method shows clear differences between aerosols with and without biomass burning smoke. Thus the mainly measurement-based method is the preferred method, because it omits the retrieval of the aerosol optical properties, which leads to high uncertainties, in contrast to the mainly model-based method.

The objective of this study is to assess the uncultured bacterial diversity in the snowpack of the Mont Blanc (MtBl) glacier containing Saharan dust deposited during four dust events during the period 2006 – 2009 by means of molecular phylogenetics. The final goal is to discover the bacteria that could be involved in the establishment of snow microbiota. Bacterial diversity was evaluated using rybotyping and subsequent sequencing of partial (V3-V5) and full-length 16S rRNA genes. For comparison purpose we also studied following samples: “clean” MtBl snow containing no Saharan dust; Saharan sand collected in Tunisia; Saharan dust collected in Grenoble (200 m a.s.l.) and recovered later on MtBl (4250 m a.s.l.). In order to verify possible microbial activity in situ, both rDNA and rRNA approaches were implemented for the “clean” snow sample. To evaluate the survival/colonization abilities of bacterial phylotypes recovered in snow samples with Saharan dust, we analyzed their closest strain physiology as well as sources of environmental clones using a threshold of ≥98% sequence similarity. For the result interpretation, we also used data on dust elemental composition and dust particles size distribution. As a result 8 clone libraries (including rRNA-based one) were constructed using V3-V5 16S rRNA gene sequences for 5 snow samples (4 with Saharan dust and one “clean”), sample of Saharan dust collected in Grenoble and Saharan sand sample. Furthermore, 4 clone libraries were generated using full-length 16S rRNA gene amplicons obtained from 4 of the above snow samples (three with Saharan dust and one ‘clean'). Species content and dominant phylotypes and their assigning to major divisions varied significantly in alpine snow on a Mont Blanc glacier associated with four depositions of Saharan dust over a 3-year. Dominant phylotypes revealed are belonged to Actinobacteria, Proteobactreia, Firmicutes, Deinococcus-Thermus, Bacteroidetes and Cyanobacteria. Such variability was detected by both partial and full-length 16S rRNA gene sequencing and seems to be caused more by conditions of dust transport than bacterial load from the original dust source. Also the preservation period of dust in snowpack could affect the species composition. Thirteen icy phylotypes as candidates into snow microbiota establishing were recognized in snow containing Saharan dust and only two in “clean” snow sample. Of them, both dominant and minor phylotypes of Cyanobacteria, Proteobacteria, Actinobacteria и Firmicutes were revealed. Data on the closest strain physiology of recognized icy phylotypes suggests that representatives of genera Massilia (Betaproteobacteria), Tumebacillus (Firmicutes), Phormidium and Stigonema (both Cyanobacteria) are most relevant findings in terms of propagation in snow. By analyzing 16S rRNA from the “clean” snow containing no Saharan dust and comparing the data with those obtained for 16S rDNA library, it has been shown that Stigonema-like cyanobacterium identified could be propagating in snow at subzero temperature. Among all identified phylotypes, 10% were categorized as HA-phylotypes based on their con-specificity (≥98% similarity) with normal (non-pathogenic) human microbiome representatives. Furthermore, 11% out of all phylotypes showed less than 90% similarity with known taxa, thus, presenting novel taxa. Sequencing of both partial (V3-V5) and full-length 16S rRNA genes permitted to describe microbial diversity more fully and get more detailed picture. / The objective of this study is to assess the uncultured bacterial diversity in the snowpack of the Mont Blanc (MtBl) glacier containing Saharan dust deposited during four dust events during the period 2006 – 2009 by means of molecular phylogenetics. The final goal is to discover the bacteria that could be involved in the establishment of snow microbiota. Bacterial diversity was evaluated using rybotyping and subsequent sequencing of partial (V3-V5) and full-length 16S rRNA genes. For comparison purpose we also studied following samples: “clean” MtBl snow containing no Saharan dust; Saharan sand collected in Tunisia; Saharan dust collected in Grenoble (200 m a.s.l.) and recovered later on MtBl (4250 m a.s.l.). In order to verify possible microbial activity in situ, both rDNA and rRNA approaches were implemented for the “clean” snow sample. To evaluate the survival/colonization abilities of bacterial phylotypes recovered in snow samples with Saharan dust, we analyzed their closest strain physiology as well as sources of environmental clones using a threshold of ≥98% sequence similarity. For the result interpretation, we also used data on dust elemental composition and dust particles size distribution. As a result 8 clone libraries (including rRNA-based one) were constructed using V3-V5 16S rRNA gene sequences for 5 snow samples (4 with Saharan dust and one “clean”), sample of Saharan dust collected in Grenoble and Saharan sand sample. Furthermore, 4 clone libraries were generated using full-length 16S rRNA gene amplicons obtained from 4 of the above snow samples (three with Saharan dust and one ‘clean'). Species content and dominant phylotypes and their assigning to major divisions varied significantly in alpine snow on a Mont Blanc glacier associated with four depositions of Saharan dust over a 3-year. Dominant phylotypes revealed are belonged to Actinobacteria, Proteobactreia, Firmicutes, Deinococcus-Thermus, Bacteroidetes and Cyanobacteria. Such variability was detected by both partial and full-length 16S rRNA gene sequencing and seems to be caused more by conditions of dust transport than bacterial load from the original dust source. Also the preservation period of dust in snowpack could affect the species composition. Thirteen icy phylotypes as candidates into snow microbiota establishing were recognized in snow containing Saharan dust and only two in “clean” snow sample. Of them, both dominant and minor phylotypes of Cyanobacteria, Proteobacteria, Actinobacteria и Firmicutes were revealed. Data on the closest strain physiology of recognized icy phylotypes suggests that representatives of genera Massilia (Betaproteobacteria), Tumebacillus (Firmicutes), Phormidium and Stigonema (both Cyanobacteria) are most relevant findings in terms of propagation in snow. By analyzing 16S rRNA from the “clean” snow containing no Saharan dust and comparing the data with those obtained for 16S rDNA library, it has been shown that Stigonema-like cyanobacterium identified could be propagating in snow at subzero temperature. Among all identified phylotypes, 10% were categorized as HA-phylotypes based on their con-specificity (≥98% similarity) with normal (non-pathogenic) human microbiome representatives. Furthermore, 11% out of all phylotypes showed less than 90% similarity with known taxa, thus, presenting novel taxa. Sequencing of both partial (V3-V5) and full-length 16S rRNA genes permitted to describe

Cette thèse est consacrée à l’étude de l’évolution des propriétés physico-chimiques d’un panache d’aérosols désertiques nord africains au cours de son transport vers l’Europe du Nord. Cet épisode de mai 2008 est considéré comme l’évènement de transport d’aérosols désertiques le plus important de ces cinq dernières années observé sur le continent européen. La propagation de ce panache a été étudiée en combinant des moyens d’observation (sols, aéroportées, spatiales) au modèle de recherche Méso-NH. L’analyse des processus de dépôt a révélée que les quantités éliminées de l’atmosphère par dépôt sec et humide représentent respectivement 7 et 40 % de la quantité émise. Les valeurs d’épaisseur optique mesurées à 440 nm sur l’Europe variaient de 0.1 à 0.8, avec un maximum proche de 1 au-dessus des Pays-Bas. Au-dessus de cette région, les aérosols désertiques sont principalement situés entre 2.5 et 5.2 km d’altitude, mais aussi proche de la surface. La concentration en aérosols désertiques au sein de ces deux couches a été estimée respectivement à 350 et 450 ��g.m-3. Néanmoins, les mesures réalisées ne présentaient pas les caractéristiques optiques habituellement associées à la présence d’aérosols désertiques. En particulier, la présence d’une importante dépendance spectrale a été observée le 30 mai à proximité de Cabauw. L’exploitation des simulations ont permis de montrer que cela résultait d’une efficacité de lessivage plus importante du mode grossier. La présence d’aérosols désertiques coïncide avec une augmentation de la concentration des noyaux de condensation (CCN). Nos résultats suggèrent que le mélange entre le panache et les aérosols de pollution d’origine anthropique a conduit à une augmentation des capacités hygroscopiques des aérosols désertiques. Ce travail de recherche confirme ainsi que les conditions physico-chimiques de l'atmosphère régissent le cycle de vie des aérosols. / This thesis focuses on the evolution of the dust physical and chemical properties through a case of long-range transport of Saharan dust over Northern Europe. This episode of May 2008 is considered as the strongest event of Saharan dust transport to Europe observed since these last five years. This spread of dust is investigated by combining observations (ground-based, airborne, satellite) and the meso-scale model Méso-NH.The analysis of the removal processes reveals that the amounts lost by dry and wet deposition represent 7 and 40 % the total dust emitted respectively. The observed aerosol optical thickness ranged from 0.1 to 0.8 at the wavelength of 440 nm, with a maximum value close to 1 is found over the Netherlands. Over that site, the dust layer was mainly located between 2.5 and 5.2 km, moreover dust was also present at 0.5 km. The concentration of dust inside these two layers is estimated to 350 and 450 ��g.m-3 respectively. Nevertheless, the usual optical characteristics of Saharan dust were not observed. In particular, the scattering coefficient measurements revealed a strong spectral dependence observed during the 30th May, close to Cabauw. The analysis of the numerical experiements revealed that this was due to high precipitation scavenging efficiency for the coarse mode. The presence of Saharan dust coincides with an increase of the cloud condensation nuclei (CCN) concentration. Our results suggest that the mixing processes between the Saharan dust and anthropogenic particles have led to an increase of the Saharan dust hygroscopic properties. Thus, this thesis confirms that physical and chemical conditions of the atmosphere govern the life cycle of dust.

La mer Méditerranée est une des zones océaniques les plus oligotrophes de la planète pour laquelle la fixation biologique d’azote a été proposée comme jouant un rôle important dans les flux de carbone et d’azote. Comme les organismes planctoniques potentiellement responsables de cette activité n’y ont jamais été étudiés, les objectifs de cette thèse étaient de caractériser la diversité (abondance et richesse spécifique), la distribution spatio-temporelle, et les facteurs de contrôle des diazotrophes, en s’axant particulièrement sur les cyanobactéries unicellulaires (UCYN). Pour cela, des hybridations in situ (TSAFISH)sur différentes fractions de taille en combinaison avec des comptages microscopiques et des analyses phylogénétiques sur les gènes 16S ADNr et nifH ont été réalisées, à la fois lors d’un cycle saisonnier à la station côtière SOMLIT de Marseille, et lors du transect Méditerranéen BOUM. Alors que les cyanobactéries filamenteuses Richelia intra cellularis et Trichodesmium sp. ont été détectées seulement ponctuellement dans le temps et dans l’espace, en très faible concentration (0.02 filament.ml-1), la communauté des cyanobactéries diazotrophes était dominée à 99,9% par du picoplancton hybridé avec la sonde Nitro821, spécifique aux UCYN. Ces cellules, de petite (0.7-1.5 μm) ou de grande (2.5-3.2 μm)taille, ont été retrouvées en faible concentration (1-6 cellule.ml-1) dans toute la Méditerranée et tout au long de l’année, à l’exception de l’été 2006 lorsque la concentration des petites cellules a atteint 5300 cellule.ml-1 au cours d’un évènement exceptionnel de pollution atmosphérique urbaine. Des efflorescences similaire sont été observées suite à des enrichissements en poussières sahariennes à une station côtière en Corse et au large au centre de la Méditerranée, simultanément avec des augmentations de l’activité de fixation d’azote.L’affiliation des petites cellules aux UCYN-A a été confirmée par phylogénie dans les eaux du large de la Méditerranée occidentale. Les librairies de clones nifH de la fraction de taille picoplanctonique étaient dominées par des séquences d’α-protéobactéries appartenant à un nouveau groupe marin de Bradyrhizobium. D’autres groupes de rhizobia et des γ-protéobactéries ont été détectés ponctuellement.Alors que l’absence de Trichodesmium sp. au large pourrait être liée aux faibles concentrations en phosphate et en fer, les facteurs inhibant le développement des UCYN-B et –C restent inconnus. Lesdiazotrophes picoplanctoniques (Bradyrhizobium, rhizobia, UCYN-A) pourraient avoir développé des stratégies spécifiques, telles que des associations avec des eucaryotes ou des particules inertes, et/ou la capacité de photosynthèse, pour acquérir le carbone nécessaire au soutien du processus de la diazotrophie / The Mediterranean Sea is one of the most oligotrophic marine areas on earth where nitrogenfixation has been formally believed to play an important role in carbon and nitrogen fluxes. Although thisview is under debate, the diazotrophs responsible for this activity have still not been investigated. The aimsof this PhD were to characterise the diversity (abundance and species richness) and the spatio-temporaldistribution of diazotrophs, as well as factors controlling their development, with a particular focus onunicellular cyanobacteria. A combination of microscopic counts with size fractionated in situ hybridization(TSA-FISH), and 16S rDNA and nifH phylogenies were done, either over a year and a half seasonal cycle atthe coastal SOMLIT station off Marseilles, and across the entire Mediterranean Sea during the BOUMtransect. Low concentrations of diazotrophic cyanobacteria were detected and this community wasdominated at 99.9% by picoplankton hybridized with Nitro821 probe, specific for unicellular diazotrophiccyanobacteria (UCYN). Among filamentous cyanobacteria, only 0.02 filament ml-1 of Richeliaintracellularis and Trichodesmium sp. were detected sporadically in time and space. Small (0.7-1.5 μm) andlarge (2.5-3.2 μm) Nitro821-targeted cells were recovered in low concentrations (1-6 cell ml-1) across theentire Mediterranean Sea and all the year long, except over a month period in summer 2006 whenconcentrations of small cells reached 5300 cell ml-1, during an exceptionally high urban pollution event.Similar blooms of small and large cells were reported after Saharan dust inputs off Corsica and at open Sea,simultaneously with increases in N2 fixation rates. The affiliation of the small Nitro821-targeted cells toUCYN-A was confirmed by 16S and nifH phylogenies offshore in the western Mediterranean Sea. Rhizobiasequences, including the ones of a new marine group of Bradyrhizobium, were dominating nifH clonelibraries from picoplanktonic size fractions. A few sequences of γ-proteobacteria were also detected incentral Mediterranean Sea. While low phosphate and iron concentrations could explain the absence ofTrichodesmium sp. offshore, the factors that prevent the development of UCYN-B and C remain unknown.It is proposed that the dominating Mediterranean picoplankters (Bradyrhizobium, rhizobia, UCYN-A)probably developed specific strategies, such as associations with protists or particles, and/or photosyntheticactivity, to acquire carbon for sustaining diazotrophy.

Im Rahmen der vorliegenden Arbeit wurden die Eigenschaften von atmosphärischen Aerosolpartikeln im tropischen Regenwald des Amazonasgebietes bestimmt. Dazu wurden die Daten einer fast einjährigen Lidarmesskampagne ausgewertet und diskutiert. Die Messungen wurden mit einem automatischen Mehrwellenlängen-Polarisations-Raman-Lidar im zentralen Amazonasbecken nahe Manaus, Brasilien, im Zeitraum von Januar bis November 2008 durchgeführt. Somit konnten erstmalig optische und mikrophysikalische Aerosoleigenschaften im Amazonasgebiet während der Regenzeit (ca. Dezember-Mai) und Trockenzeit (ca. Juni-November) höhenaufgelöst charakterisiert werden. Einleitend werden die meteorologischen Bedingungen im Amazonasgebiet erläutert und eine Literaturübersicht über Aerosolforschung in dieser Region gegeben. Das Messgerät sowie verschiedene Kalibrier- und Korrekturschemen, die zur Datenauswertung notwendig sind, werden vorgestellt. Auch Vergleiche mit anderen Messgeräten werden diskutiert. Diese zeigen, dass die aus den Lidarmessungen abgeleiteten Parameter von hoher Qualität sind. Anhand von Fallstudien werden mit Hilfe von Rückwärtstrajektorien und Satellitenmessungen typische Aerosolbedingungen am Messstandort diskutiert. Um die generellen Unterschiede zwischen Regen- und Trockenzeit zu quantifizieren, wird eine statistische Auswertung aller analysierten Lidarmessungen präsentiert. Die Analyse der Lidardaten zeigt, dass während der Regenzeit im Amazonasgebiet in ca. der Hälfte aller Fälle sehr saubere Bedingungen mit einer Aerosol Optischen Dicke (AOD) von weniger als 0.05 (bei 532 nm) vorherrschen können. Allerdings wurde in ca. 30% aller analysierten Fälle im Zeitraum von Januar bis Mai auch afrikanisches Aerosol, vornehmlich Saharastaub und Biomasseverbrennungsaerosol (BBA), am Messstandort detektiert. Dabei dominierte meist BBA die Aerosolpopulation, wie die Depolarisationsmessungen zeigten. In der Trockenzeit ist die Atmosphäre im Amazonasbecken hauptsächlich mit BBA aus Südamerika belastet. Daher ist die AOD im Durchschnitt um einen Faktor drei größ er als in der Regenzeit. BBA wurde zu dieser Jahreszeit regelmäßig bis zu einer Höhe von 4-6 km detektiert. Basierend auf den vorgestellten Langzeitmessungen werden erstmalig die optischen Eigenschaften von südamerikanischem BBA statistisch analysiert und diskutiert. / Continuous lidar measurements were performed in the Amazon rain forest for almost one year in 2008. The results of the automated multiwavelength-Raman-polarization lidar observations were presented in this dissertation. These measurements are the first long-term observations of the vertical aerosol structure ever made in the Amazon Basin. The advanced lidar observations were conducted 60 km north of Manaus in the central northern part of Amazonia. The area is widely covered with pristine rain forest. A HYSPLIT backward-trajectory analysis showed that the observations were representative on a regional scale for the central northern part of the Amazon rain forest. The general weather conditions in this region are characterized by a wet (December-June) and a dry season (July-November). During the dry season, a high fire activity occurs in Amazonia, which heavily influences the atmospheric conditions. With the lidar instrument, vertical profiles of the particle backscatter coefficient at 355, 532, and 1064 nm, of the particle extinction coefficient at 355 and 532 nm, and of the particle linear depolarization ratio at 355 nm can be determined. The results from the long-term lidar observations performed in Brazil contain a lot of new information about the aerosol conditions in the central northern Amazon Basin and corroborate certain findings from former aerosol measurements in Amazonia. It was shown for the first time that advection of Saharan dust together with biomass burning aerosol (BBA) from Africa occurred regularly throughout the wet season. In about one third (32%) of all lidar observations during the wet season, African aerosol was dominating the optical aerosol properties in Amazonia. The analysis of the vertical aerosol structure during such events revealed that the African aerosol arriving in the central northern Amazon Basin was usually trapped in the lowermost 3-3.5 km of the troposphere. To quantify the amount of Saharan dust and African smoke transported towards the lidar site, the dust contribution to the measured optical aerosol properties was separated by means of the measured particle depolarization ratio. This study led to the result that in about one half of the cases with African aerosol advection, smoke particles contributed to more than 50% to the total Aerosol Optical Depth (AOD). The smoke transport from Africa towards Amazonia occurred predominantly between January and April when the fire activity in Central Africa was highest. BBA is thus a major constituent of the aerosol plumes that are regularly transported from Africa towards Amazonia. This is a key finding of the presented study. During clean conditions, an AOD (532 nm) of less than 0.05 was observed and the aerosol was trapped in the lowermost 2 km of the troposphere. However, the analysis of the long-term data set revealed that these clean atmospheric conditions occurred in only 48% of all wet-season cases. One example for such background conditions was intensively discussed and it was shown that a major meso-scale rain event occurred in the Amazon region at the same time. This precipitation event was possibly partly responsible for the very low aerosol load. Two case studies from the dry season were presented for which BBA dominated the optical properties. In the first case, a comparable high aerosol load (AOD of 0.41) prevailed while in the second one, a medium aerosol load (AOD of 0.15) was observed. Aged BBA advected from regions south of the lidar site were identified to be the dominant aerosol species for both cases. However, very different geometrical, optical and microphysical properties of BBA (e.g., vertical layering, lidar ratio, Ångström exponent, effective radius, SSA) were observed on both days. In the first case, aerosol was present up to about 4.5 km. Extinction-related Ångström exponent s of about 1 and lidar ratios between 70 and 90 sr were found at different heights for the smoke aerosol. The BBA was highly absorbing (SSA of 0.81) at heights of the highest RH (85%), whereas above under dry conditions (RH=50%) only moderate absorption (SSA of 0.93) was detected. In the second case, smoke was detected up to 4.5 km, and Ångström exponent of about 2 and lidar ratios of 45-55 sr were measured in the aerosol layers. The BBA was only moderately absorbing indicated by SSA values between 0.92 and 0.94. The reason for the differences in the smoke properties could be the shorter travel time to the lidar site (<24 h), different aging processes (e.g., cloud/rain processing), or different burning conditions. In both cases, no depolarizing effects of the BBA could be observed. The strong contrast between the aerosol conditions in the dry season and the wet season were confirmed by the statistical analysis of all lidar observations in 2008. Due to the high BBA concentration in the atmosphere, the mean AOD of the dry season was found to be a factor of 3 higher than the mean AOD of the wet season (0.26 compared to 0.08 at 532 nm). Maximum AOD values were less than 0.55 (at 532 nm) and hence show that the lidar location was not in the direct vicinity of fire events. In only 7% of all cases in the dry season 2008, an AOD below 0.1 was observed. Also the maximum extinction and backscatter coefficient values in the dry season 2008 were 2-3 times higher than during the wet season of this year. The vertical aerosol distributions differ also significantly between the two seasons. In the wet season, the aerosol was mostly trapped in the lowermost 2.5 km, while in the dry season aerosol typically reached up to 4.5 km. Aerosol was occasionally detected up to 6.14 km in the dry season. The majority of the aerosol (95% of the AOD), however, was found to be on average below 2.3 km in the wet season and below 3 km in the dry season. During the wet season, lofted aerosol layers and multiple aerosol stratification was less frequent than in the dry season. The extent of BBA plumes during the dry season showed no correlation to the ML top height. Virtually uniform smoke haze layers were observed up to the AL top. Thus, pyro-convection and/or cloud-related mixing seem to be the major processes for the vertical distribution of BBA.