To improve the accuracy of spatial carbon exchange estimates, we evaluated simple linear models for net ecosystem exchange (NEE) and gross primary productivity (GPP) and how they can be used to upscale the CO₂ exchange of agricultural fields. The models are solely driven by Sentinel-2-derived vegetation indices (VIs). Evaluations show that different VIs have variable power to estimate NEE and GPP of crops in different years. The overall performance is as good as results from complex crop models.