The biosphere is a key component of the global carbon cycle and the terrestrial ecosystem has a large spatial heterogeneity. The sparse in-situ stations limit our ability to constrain the continental scale carbon cycle. We use the CO2 fluxes inferred by inverse models and additionally remote sensing based vegetation indices, which can be a proxy for photosynthesis, as atmospheric constraints to select process-based dynamic global vegetation models (DGVMs) from the 17 TRENDYv11 models. DGVMs provide us the component fluxes such as gross primary productivity (GPP) and terrestrial ecosystem respiration (TER) enabling the attribution to variations in CO2 uptake and emission.