Atmospheric concentrations of surface ozone (O3) are strongly affected by deposition to the biosphere. Deposition processes are very sensitive to turbulence, temperature, relative humidity and soil moisture deficit and are expected to respond to global climate change, with implications for both air quality (e.g. human health) and ecosystem services (e.g. crop yields). In this PhD study, the global chemistry aerosol model UKCA (United Kingdom Chemistry Aerosol model) dry deposition scheme was thoroughly investigated. Some errors in the existing implementation of the current UKCA stomatal resistance and in-canopy aerodynamic resistance terms for O3 and NOw (NO2, PAN, PPAN, MPAN) were identified and corrected (WES scheme). These model corrections led to a decrease of the total annual dry deposition of -150 Tg(O3) yr-1 (-13%) which brings UKCA more in line with multi-model inter-comparison estimates. This was associated with a large increase of surface O3 concentration over land in the Northern Hemisphere (NH) with values up to 12 ppb (+50%) higher on annual average. Many studies have shown that O3 stomatal uptake by vegetation, which is the pathway leading to damage, accounts for 40-60% of total deposition on average. The remaining non-stomatal deposition flux is to external foliar surfaces, and soil. A more mechanistic non-stomatal dry deposition approach along with a scheme to simulate the effect of moisture on foliar surfaces on the stomatal transport (ZHG scheme) was introduced in UKCA to study the relative contributions of O3 flux occurring to stomatal and non-stomatal pathways at the global scale, and to explore the sensitivity of simulated surface O3 and O3 deposition flux. The ZHG scheme, led to significant changes in the O3 dry deposition velocity (Vd) (+40% in the North Hemisphere over boreal forests and -30% over tropical regions on annual average). The results of this study show that the ZHG scheme significantly changes the partitioning between stomatal and non-stomatal O3 flux. The non-stomatal fraction increased throughout the year and considerably during the cooler season and in spring (with maxima values by up to 60% for C3 grass and by up to 70% for needle leaf trees). The performance of both UKCA dry deposition schemes were compared with measurements, focussing on the diurnal and seasonal variations of the dry deposition velocity terms and the partitioning of O3 fluxes between stomatal and non-stomatal sinks. Overall, both UKCA dry deposition schemes capture the diurnal variations of Vd reasonably well. However, this study highlighted difficulties in comparing large grid (~280 x 390 km at mid-latitudes) averaged modelled values with site and vegetation specific characteristics of the measured exchange processes (~1 km2) and the driving meteorological variables. These differences in scale are a large source of uncertainty in the comparison of measured and modelled O3 Vd. Off-line simulation tests conducted on the non-stomatal deposition component with the ZHG scheme demonstrated the importance of modelling some key environmental and meteorological factors accurately (e.g. relative humidity, friction velocity, leaf area index). This was found to be crucial in order to improve O3 Vd model performance as well as improving the representation of specific vegetation properties. A comparison of the modelled global surface O3 concentration against observations both in the NH and SH revealed that the model performs well in the NH using both schemes, capturing the observed surface O3 cycle and the absolute values. The ZHG scheme led to a reduction of the annual bias (up to -13.5% on average) in the NH monitoring sites considered for this study. This is associated with a decrease in O3 deposition simulated with ZHG (as much as of -20% on annual average). By contrast, the seasonal cycle and absolute values of the observed surface O3 are not well reproduced by the model across the SH monitoring sites used in this study and a larger bias was found using the ZHG scheme (60% on average) compared to WES scheme (47% on average), as a consequence of an increase in O3 deposition (as much as of +20% on annual average) calculated with ZHG. A future climate integration for the 2090s using RCP 8.5 scenario was used to investigate the response of UKCA modelled O3 to climate change. The effect of climate change (by altering only the GHG concentrations predicted with RCP 8.5) on the dry deposition sink of O3 was addressed contrasting the two non-stomatal deposition parameterizations, and ignoring the changes in land-use and anthropogenic emissions. The study showed that O3 Vd over land declines from 2000 to 2100, and most strongly over vegetated areas (up to -24% over S. America, -17% over N. America and -10% over Europe). Climate change led to an increase of surface O3 concentration over land (by up to 20%). Whilst the two schemes behave similarly, and an increase in turbulence has been identified as the main driver, the decrease in land Vd is generally stronger in ZHG. This effect is more important over N. America and Eurasia where ZHG exhibits larger differences in deposition compared to WES as a result of changing climate. The increase in surface O3 over Arctic and Antarctic regions shows the effect that changes in O3 deposition might have on the long-range transport of O3. Finally, the influence of climate change on the partitioning of the O3 deposition flux was examined. This analysis revealed that more O3 is predicted to deposit through stomatal pathways with ZHG over N. America, C. Europe and E. Asia (up to +30%) compared to WES as a result of changing climate. Given that ZHG scheme captures the influence of meteorology and changing climate on surface O3 better than WES, it was concluded that modelled surface O3 using ZHG scheme showed a larger sensitivity to a changing climate than WES. These results imply potentially important effects of climate change on tropospheric O3, degrading air quality through the later decades of this century.

L'ozone troposphérique est un gaz à effet de serre et un puissant oxydant qui impacte la santé et la production végétale. Afin d'analyser son impact sur le fonctionnement de la végétation il est nécessaire d'effectuer une partition entre le dépôt stomatique, qui pénètre à l'intérieur des feuilles, et le dépôt non stomatique, qui résulte de la destruction chimique de l'ozone sur les cuticules foliaires, le sol ou dans l'air à l'intérieur du couvert. L'objectif central de ce travail de thèse est d'améliorer la compréhension et la modélisation des processus impliqués dans le dépôt non stomatique d'ozone et plus particulièrement dans le dépôt sur les films d'eau à la surface des feuilles. Le modèle de transfert de masse et d'énergie sol-végétation-atmosphère MuSICA, capable d'estimer les quantités d'eau sur le feuillage à différents niveaux dans le couvert, a permis de modéliser la solubilisation puis la diffusion et réaction de l'ozone dans l'eau sur les feuilles. Cette voie de dépôt a permis d'expliquer les flux d'ozone observés sur blé en conditions humides en ajustant les valeurs du taux de réaction chimique de l'ozone dans le film d'eau de 103 s 1 en début de saison à 105 s 1 pendant la sénescence. Ce taux semble varier entre les espèces végétales (blé, hêtre et pin maritime) mais la tendance avec la sénescence persiste. Ces résultats ont été confortés par un dispositif expérimental original en laboratoire, sur blé, hêtre, pin blanc et chêne pédonculé. Des échanges de composés chimiques entre la feuille et le film d'eau réactifs avec l'ozone pourraient expliquer les ordres de grandeur obtenus et l'augmentation du taux de réaction en période de senescence. / Tropospheric ozone is a greenhouse gas but also a powerful oxidant that impacts vegetation health and productivity. In order to analyse the impact of ozone on vegetation functioning it is necessary to separate stomatal ozone deposition that penetrates inside the leaf interiors from non-stomatal deposition that result from ozone chemical destruction on leaf cuticles, in the soil or in the air inside the canopy. The central objective of this PhD work is to improve our understanding and modelling of the processes involved in non-stomatal ozone deposition and more specifically in ozone deposition on water films on leaf surfaces. For that purpose I used the soil-vegetation-atmosphere transfer model of mass and energy MuSICA. The choice of this model was motivated by its ability to estimate the quantities of water on foliage at different levels inside the canopy. This allowed me to model ozone dissolution, diffusion and reaction in the water film on wet leaves. This new deposition pathway could explain observed ozone fluxes over wheat in wet conditions by adjusting the chemical reaction rate of ozone in the water film from 103 s 1 at the beginning of the growing season, up to 105 s 1 during senescence. This reaction rate seemed to vary between species (wheat, beech, maritime pine) but the trend with the senescence remained. These results were then confirmed by lab experiments using an original setup, on wheat and beech, but also white pine and pedunculate oak. Exchanges of chemical compounds between the leaf and the water film that would react with ozone could explain the order of magnitude of the reaction rate and its increase during senescence.