Recent estimates of the North American carbon budget have shown a moderate convergence at annual and longer time scales between terrestrial biogeochemical models (TBMs) and atmospheric inversions. However, multi-TBM comparisons revealed large discrepancies both spatially and temporally among net ecosystem exchange estimates, illustrating our limited understanding of the underlying mechanisms. To bridge the gap between processes and atmospheric inversions, we propagated process-based errors in a TBM, here an ensemble of CASA model simulations, into a mesoscale atmospheric system to identify and possibly optimize parameters instead of surface fluxes. Our offline atmospheric-ecosystem coupled model also represent uncertainties from the atmospheric transport in an ensemble-based framework. The unique collection of continental Planetary Boundary Layer measurements of CO2 mole fractions and meteorological variables from the NASA Atmospheric Carbon and Transport-America (ACT-America) mission provides new perspectives on our understanding of transport and fluxes of greenhouse gases across three regions of the U.S., four seasons, and a variety of synoptic weather conditions. We have assembled a calibrated, continental-scale, 27-km resolution atmospheric model ensemble including biospheric and fossil fuel contributions, prescribing the large-scale inflow of CO2 from several global models. The ensemble system can separate and quantify the uncertainties in modeled CO2mole fractions from atmospheric transport, biospheric fluxes, fossil fuel emissions, and boundary inflows. Key parameters of CASA, driving ecosystem respiration and photosynthetic uptake, are constrained using both atmospheric mixing ratio measurements. We identified discrepancies between bottom-up and top-down approaches spatially using aircraft footprints from a backward Lagrangian particle model, to define optimal parameter values for dominant ecosystems across the US. Our joint CO2 mole fraction and flux analysis revealed that CASA underestimated net uptake resulting from missing sink processes that result in overestimation of respiration.