The carbon intensity (CI) of a biofuel quantifies the cumulative greenhouse gas (GHG) emissions associated with the production of a particular biofuel. Cradle to farm-gate CI assesses the total amount of GHG emitted per unit (mass or volume or energy content) of the biofuel produced from the production of feedstocks. The CI of biofuels provides a standardized and transparent way to compare the climate change impacts of different biofuels and to inform policy decisions related to renewable fuel standards and incentives. However, the uncalibrated process-based model and information gap regarding field scale farming practices lead to uncertainties in LCA inventory, thus leading to significant uncertainties in the cradle to farm-gate feedstock CI quantification at the field-level. We aim to answer the following science questions in this study: (1) How can we systematically consider uncertainties of feedstock cradle to farm-gate CI when using inventory from a process-based model? (2) How much uncertainty in CI is reduced when we have satellite constraints? To answer these questions, we first set up a CI assessment framework with the process-based model, ecosys, and the GHG calculator GREET-FDCIC model to assess the feedstock cradle to farm-gate CI at the field-level. Then we did an uncertainty assessment at a flux-tower site in central Illinois where the ecosys model has been fully calibrated and validated using flux-tower measurements (i.e., GPP, NEE, net radiation and latent heat fluxes). We assessed the uncertainty in feedstock CI caused by unknown management practices (i.e., fertilizer, tillage, cover crops), soil properties (i.e., initial soil organic carbon stock, clay content) and plant parameters with- and without-satellite-constraint conditions using the Bayesian framework with a Monte Carlo sampling method. Finally, we performed a regional assessment of CI for every field in a region in central Illinois. We found that after incorporating satellite-observed management information (i.e., tillage intensity and cover crop existence through satellite observations) and model constraints (i.e., satellite derived GPP), uncertainty in CI could be significantly reduced. Our regional CI assessments revealed significant variability among fields within the same county. This finding stressed the importance of conducting CI assessments at a fine scale to better understand and account for CI spatial heterogeneity.