Increasing soil carbon stocks has been identified as a major climate change mitigation solution. As a consequence, an objective of 4/1000 yearly increment in soil carbon stocks has been proposed at the COP21.  Sustainable agriculture provides several solutions to meet this objective and among those solutions, the implementation of cover crops has been identified as most efficient. Currently, a comprehensive modeling tool that takes into account the major bio-geophysical processes with associated uncertainties, while assimilating frequent high-resolution observations at large scale could allow accounting for the effect of cover crops on the carbon budget in a realistic way. In this study, we quantify the components of the carbon budget at high resolution and we analyse the effect of cover crops. Computations are based on the newly developed AgriCarbon-EO tool which assimilates full resolution (10-20m) Sentinel-2 optical data into a radiative transfer model (PROSAIL), and a crop model (SAFYE-CO2). The assimilation scheme is based on a Bayesian approach which provides the retrieved biogeophysical variables with their associated uncertainties. Uncertainties are essential when determining the carbon stocks. For instance, the future European Common Agricultural Practice (CAP) may take into consideration the uncertainty of the determination of the soil carbon stocks changes in the evaluation of the subsidies. The main inputs of the computations are weather data, soil texture maps, crop maps and surface reflectances. The Sentinel-2 Leaf Area Index (LAI) are obtained from those Sentinel-2 surface reflectance by inverting the PROSAIL model. These are then assimilated into the SAFY_CO2 model to determine the carbon budget components. To validate our approach, we implemented the AgriCarbon-EO tool over a set of plots in south-western France over which we dispose of biomass measurements for cover crops in wheat/cover crop/maize rotations for 2017-2018 and 2019-2020 agricultural seasons. Also, the CO2 fluxes are validated against eddy covariance flux measurements in the same context. Our study shows that the cover crops allow on average 250gC/m² of organic carbon with a high spatial heterogeneity. This has important implications regarding the dynamic of carbon storage in agronomic soils and demonstrates the importance of high-resolution agronomic modeling.