Évolution de la neige et de l’air sous la surface de la calotte glaciaire de l’Antarctique de l’Est

Ice core gas records provide vital insights into both variations in atmospheric composition over the last 800 ka, and the timing of climatic events. To accurately infer past atmospheric composition, entrapped gases must first be corrected for processes in the firn which modify gas composition prior to bubble formation. Precise understanding of these processes and their site dependence are therefore essential for performing accurate corrections to reduce uncertainties associated with atmospheric reconstructions. While numerous studies have focussed on understanding firn processes, unknowns remain which complicate the interpretation of ice core gas records. The aim of this thesis is to contribute to the field of firn air dynamics and pore closure to reduce uncertainties associated with such reconstructions. Two primary topics are explored to address specific predicaments in ice core sciences. The first concerns the use of δ15N-N2 (δ15N) and δ40Ar records for reconstructing firn column characteristics such as firn column height and firn temperature difference. Despite their usefulness, discrepancies between isotope-based reconstructions and those from physical models are observed during glacial periods East Antarctic sites. Various hypotheses have been proposed to explain this model-data mismatch but there remains no clear consensus on the source of error. In this thesis, we assess our understanding of firn air dynamics, in the context of the δ15N mismatch, using new measurements of δ15N and δ40Ar from year-round firn air sampling at Dome C. We evaluate current understanding of how seasonal thermal effects influence isotope ratios in shallow firn and explore the extent to which seasonal thermal signals will be archived in trapped gases. Consistent with previous findings from South Pole, we observe a discrepancy in magnitude and depth propagation of the winter thermal isotope profiles in the firn which are not captured by the firn air transport model. Our observations indicate a seasonal bias in isotope ratios at Dome C which should be considered for the interpretation of δ15N.Secondly, this thesis examines the use of elemental ratios, specifically, the ratio of O2 to N2 molecules in trapped gases relative to the atmosphere (δO2/N2), for orbital dating. This method is crucial for dating deep, old ice cores, such as the new Beyond EPICA oldest ice core (BEOIC), but an incomplete understanding of the mechanisms driving δO2/N2 variability is a source of chronological uncertainty. It is classically assumed that δO2/N2 variability is driven predominantly by local insolation. However, using a compilation of various Greenlandic and Antarctic records, we show that δO2/N2 is also influenced by local temperature and accumulation rate. This observation motivates further investigation into the physical mechanisms controlling δO2/N2 under different climate conditions with the aim of reducing chronological uncertainties associated with orbital tuning.A final study aims to elucidate the temperature and accumulation rate driven mechanisms of δO2/N2 variability while also revisiting the δ15N model-data mismatch. New measurements of inert gas isotopes, δ15N, and δO2/N2 were obtained from shallow bubble ice and firn air samples over the entire firn column at two contrasting East Antarctic sites: D47, a high-accumulation site, and Little Dome C (1km from BEOIC), a low-accumulation site. We observe substantially lower δO2/N2 in closed porosity at LDC compared to D47, but much stronger enrichment of δO2/N2 in open porosity near close off at D47. These observations support findings from our earlier study that temperature and accumulation rate play a key role in modulating δO2/N2. Additionally, inert gas isotopes help evaluate site-specific factors contributing to the δ15N model-data mismatch, building on findings from our seasonal firn air study.