Atmospheric trace components play a critical role in the earth–atmosphere system through their interaction and perturbation to global atmospheric chemistry. They perturb the climate through scattering and absorbing of solar radiation (direct effects), thereby impacting on the heat energy balance of the atmosphere, and alter cloud microphysical properties affecting cloud formation, cloud lifetime and precipitation formation (indirect effects). These trace components can also have adverse effects on human health, visibility and air quality (AQ) composition, including various feedback processes on the state of the atmosphere. As well as their direct and indirect effects, aerosols are important for cloud formation. They serve as cloud condensation and ice nuclei (CCN and IN) during cloud droplet and ice crystal formations. Although many connections between clouds and aerosol effects have been established in cloud physics and climate modelling, aerosol–cloud interaction (ACI) is still one of the areas of large uncertainties in modern climate and weather projections. Different models have been developed placing much emphasis on ACIs, to have robust and more consistent description processes within the meteorological and chemical variables to account for ACIs and feedback processes. Because pollutant distributions are controlled by a specific meteorology that promotes residence times and vertical mixing in the atmosphere, reliable chemical composition measurements are required to understand the changes occurring in the earth–atmosphere system. Also, because atmospheric pollution is a combination of both natural and man-made (anthropogenic) sources, to direct controlled and/or mitigation procedures efficiently, contributions of different sources need to be considered. Occasionally these are explored from a particular region or global environment, depending on a specific area of interest. A fully coupled online meteorology–chemistry model framework (WRF-Chem) is used to investigate atmospheric ACIs over southern Africa –a region characterized by a strong and intense seasonal biomass burning (BB) cycle. The large transport of aerosol plumes originating from the seasonal burning from agriculture, land-use management and various activities give rise to a unique situation warranting special scrutiny. Simulations are conducted for the 2008 dry season BB episode, implementing a chemical dataset from various emission sources (anthropogenic, BB, biogenic, dust and sea salt) with the meteorological conditions. A base line (CNTRL) simulation was conducted with all emission sources from 26 August to 10 September 2008. To probe the contribution of BB on the regional pollution and influence on ACIs, a sensitivity (TEST) simulation was conducted without BB emissions and compared to the base line. The impact of natural and anthropogenic aerosol particles is studied and quantified for the two simulations, focusing on aerosol concentration and cloud responses under different model resolutions. A statistical analysis of pollutant concentration of major regulated species and cloud variables is conducted and the percentage difference used to assess the contribution due to BB emissions. Results confirm the high variability of spatial and temporal patterns of chemical species, with the greatest discrepancies occurring in the tropical forests whereas the subtropics show more urban/industrial related emissions. Whilst CO and O3 show statistically significant increases over a number of cities/towns, the trend and spatial variability is much less uniform with NO2 and PM in most urban and populous cities. Statistical analysis of major chemical pollutants was mainly influenced by BB emissions. O3, NOx, CO and PM increase by 24%, 76%, 51%, 46% and 41% over the main source regions, whereas in the less affected regions concentrations increased by 5%, 5%, 5%, 3% and 2% when BB emissions are included. This study sheds new light on the response of cloud processes to changing aerosol concentrations and different model resolutions. In the parameterised case (dx = 20 km), clouds become more cellular, correlated with high supersaturations, whereas in the resolved case (dx = 4 km), they become more faint with relatively lower supersaturations. Aerosol effects on cloud properties were further studied and statistical analysis conducted on CCN, cloud droplet number concentration (CDNC), supersaturation and aerosol optical depth (AOD) at two different grid spacings. Most clouds occur to the west of the domain coincident with increase in aerosol concentration and AOD, while single scattering albedo (SSA) decreases. A considerable cloud ‘burn-off’ occurs in tropical west Africa, where aerosols can also be lofted up to 500-hPa level when BB emissions are included in the simulation. Due to BB, absorbing aerosol increased by 76% and 23% over tropical west and subtropical southeast, while tropical east shows no change. The study shows that tropical central Africa is characterized by an increased build-up in biomass burning aerosols (BBAs), forming a regional haze with high AOD; this becomes stronger near active burning areas with a significant proportion occurring to the west. AOD enhancement increases up to 38%, 31% and 11% in the west, east and south respectively. Although CDNC increased in areas with high aerosol concentration, supersaturation decreases (in the small domains) since increase in aerosol number concentration decreases maximum supersaturation Smax. Changes in absorbed radiation increased by +56 Wm-2, +23 Wm-2 and +14 Wm-2 in the west, east and southeast. To further evaluate the model sensitivity and its skill, an analysis was conducted by comparing the model performance with measurement data. Simulated AOD, surface concentrations of CO and O3, ozonesondes and liquid water path (LWP) were compared with measured data from MODIS satellite, SAFARI2000 field study and Cape Point WMO. The model shows a good skill in capturing and reproducing the trends as that measured. However, a severe lack of measurement data over southern Africa makes it more difficult to effectively evaluate WRF-Chem over southern Africa. There is a need for increased availability of measurements to adequately compare with models. This study is one of the first WRF-Chem studies conducted over southern Africa to simulate the weather and pollution interaction. The novelty of the present study is the combined analysis of ACI sensitivity to aerosol loading and cloud response in a regime-based approach. The study concludes with a brief discusssion of future directions for work on AQ and modelling interactions between pollution and weather over southern Africa.

Dans l’atmosphère, les particules d’aérosol jouent un rôle clef sur le climat et, par leur action sur la qualité de l’air, ont un impact néfaste sur la santé publique. Ces particules sont composées d’un mélange complexe d’espèces inorganiques et organiques formées à partir d’une grande variété de sources. Si les sources et mécanismes de production des espèces inorganiques sont désormais relativement bien connus, la caractérisation de la fraction organique des aérosols est beaucoup plus complexe : elle est constituée d’aérosols organiques primaires, émis directement dans l’atmosphère, et d’aérosols organiques secondaires (AOS) produits par la conversion gaz-particules de Composés Organiques Volatils (COV). Afin de comprendre les processus de formation des aérosols organiques, des modèles tridimensionnels de chimie-transport sont mis en œuvre. Or, à ce jour, les concentrations en aérosols organiques observées dans l’atmosphère demeurent sous-estimées par ces modèles. L’objectif de cette thèse est d’étudier les différents processus de transport, d’émissions et de transformations chimiques intervenant dans la formation des aérosols organiques à partir du modèle de chimie-transport WRF-Chem (Weather Research and Forecasting – Chemistry). Les sorties du modèle ont été comparées à des mesures effectuées à la station du puy de Dôme au cours de trois situations correspondant à trois saisons (automne, hiver, été) durant lesquelles des masses d’air de diverses origines ont été échantillonnées. Ces mesures documentent les conditions météorologiques, les propriétés des espèces chimiques gazeuses et des particules d’aérosol. En particulier, les mesures fournies par un spectromètre de masse (AMS : Aerosol Mass Spectrometer), fournissent de informations détaillées sur la variabilité temporelle de la composition chimique des aérosols et notamment sur leur concentration en masse. Les comparaisons modèle/mesures ont montré que les variations saisonnières de la composition chimique des aérosols observées au puy de Dôme étaient bien capturées par le modèle. Cependant, il s’est avéré que les concentrations en aérosols organiques étaient fortement sous-estimées par le modèle et plus particulièrement lors de la situation d’été. La confrontation des origines des masses d’air simulées par le modèle WRF-Chem à celles déterminées par le modèle lagrangien HYSPLIT reconnu pour l’étude de la dispersion atmosphérique et l’analyse des variations de la localisation du sommet du puy de Dôme vis-à-vis de la couche limite atmosphérique ont mis en évidence que le transport était correctement reproduit par le modèle. Les mesures de gaz disponibles au puy de Dôme ont mis en évidence une forte sous-estimation des concentrations en COV d’origine anthropique simulées par le modèle. Des tests de sensibilité ont été réalisés sur les émissions de ces espèces pour restituer les niveaux de concentration observés. Les émissions et les rendements en AOS des COV d’origine anthropique implémentés dans la paramétrisation VBS dédiée aux aérosols organiques dans le modèle ont pu être modifiés afin de reproduire les niveaux de concentration en aérosols organiques observés au puy de Dôme. / In the atmosphere, aerosol particles play a key role on both climate change and human health due to their effect on air quality. These particles are made of a complex mixture of organic and inorganic species emitted from several sources. Although the sources and the production mechanisms for inorganic species are now quite well understood, the characterization of the organic fraction is much more difficult to study. Indeed, particulate organic matter comes from primary organic aerosols directly emitted to the atmosphere and secondary organic aerosols (SOA), which are formed from gas-to-particle conversion of Volatile Organic Compounds (VOC). Three-dimensional chemistry-transport models are developed to better understand the organic aerosol formation processes. However, these models underestimate the organic aerosol concentrations. The aim of this thesis is to study the transport, the emissions and the chemical transformations involved in the formation of the organic aerosols using the WRF-Chem chemistry-transport model (Weather Research and Forecasting – Chemistry; Grell .et al., 2005). Model outputs are compared to measurements performed at the puy de Dôme station (France) during three campaigns. These measurements allow for characterizing various air masses and different seasonal behaviours (in autumn, winter and summer). The station hosts many probes for controlling meteorological parameters, gas phase species and aerosol properties. In particular, a mass spectrometer (AMS: Aerosol Mass Spectrometer) provides detailed time evolution of the chemical composition and mass concentration of the particulate matter. The comparisons between model results and observations have shown that seasonal variations of the aerosol chemical composition are captured by the WRF-Chem model. However, the organic aerosols mass concentrations are strongly underestimated and this underestimation is more important for the polluted summer case. The calculated origins of air masses are comparable to the results of the lagrangian model HYSPLIT currently used for atmospheric dispersion. The top of the puy de Dôme is observed to be either in the boundary layer or above depending on the season and these observations are correctly reproduced by the WRF-Chem model. As the anthropogenic VOC concentrations are underestimated by WFR-Chem model, sensitivity tests on the anthropogenic VOC emissions and SOA yields, used in the VBS secondary organic aerosols parameterisation, are done to better reproduce the organic aerosol concentrations observed at the puy de Dôme station.