It is increasingly recognized that the atmosphere composition of snow covered regions – especially polar regions – is noticeably affected by air-snow interactions. Indeed, the snowpack is a multiphase reactor, but physico-chemical processes which take place inside are still poorly understood. A detailled understanding of snow-atmosphere interactions is essential for understanding and modeling properly the composition and reactivity of the atmosphere above snow covered regions. Reconstructions of past trends in atmospheric composition using ice cores also require to understand snowpack processes that affected the composition of interstitial air and burried snow after its deposition.Nitrate (NO3-) present in the snowpack plays an important role as it photochemically produces nitrogen oxides (NOx=NO+NO2), which affect the oxidative capacity of the atmosphere through ozone production.This thesis thus aimed at studying physico-chemical processes which take place inside the snowpack and modify nitrate concentration.In a first part, a reaction mechanism to reproduce nitrate photochemistry in snow were developed, based on previous studies. The main hypothesis was that chemical reactions take place in a quasi-liquid layer located on the surface of snow cristals. However, the properties of this ice-air interface are poorly known, and it appeared that this approach had too many uncertainties to be continued.Then, a thorough discussion were carried out to assess current attempts in snow chemistry modeling, and to propose another approach which could prevail given current knowledge on this topic.In a second part, physico-chemical exchange processes between air and snow were studied and modeled. This concerned adsorption, solid phase diffusion and co-condensation. Among the results that arise, it appeared that current parameterizations of nitrate surface coverage are unable to reproduce measured concentrations, in the studied case of Dome C surface snow, and further reveal sizeable overestimations. On the contrary, simultaneous modeling of solid phase diffusion and co-condensation allows a qualitatively good reproduction of measurements, which cover more than a year, thus including both austral summer and winter with their specific features.This study reveals the importance of exchange processes for snow chemistry modeling, and give basis for future work on this topic.
