Analyse multi-échelle de la dynamique du carbone naturel et anthropique dans l’océan Atlantique Nord

CO2 emissions over 270 years have increased atmospheric CO2 from 278 to 420 ppm. The ocean hasabsorbed ~25% of these emissions, the North Atlantic (NA) serving as a major carbon sink through deep water formation and high biological activity. Climate change affects these processes, and current modelscannot fully replicate observations, highlighting critical knowledge gaps in sink capacity and spatio-temporal variability. This thesis provides a multiscale assessment of natural (Cnat) and anthropogenic (Cant) carbon concentrations and transports in the NA using multisource data (hydrographic sections, Argo floats, ocean reanalyses) and a multimethod approach combining neural networks with back-calculation methods. After validating this approach, the thesis presents novel results that include the first 30-year observation-based monthly time series of surface-to-bottom transport of Cnat and Cant over the A25-OVIDE section (Greenland to Portugal, 1993-2022). Additional results from multiproduct air-sea CO2 fluxes and 3D basin-scale Cant storage estimates open new research avenues for high-resolution NA storage analysis. By elucidating how tracer concentration changes and ocean dynamics interact to influence NA carbon sink variability, this work contributes significantly to understanding of the ocean carbon cycle, potentially improving future climate predictions.