This presentation explores how CO₂ fluxes can be inferred from the atmospheric perspective. Both biogenic and anthropogenic fluxes cause measurable signals in CO₂ concentration, which can be interpreted using atmospheric transport models. Across continental regions, satellite observations capture broad atmospheric signatures of ecosystem activity. When interpreted with vegetation models, these signals reveal how carbon-water dynamics shape biogenic fluxes. The same principle applies in cities, but at much finer scales: dense sensor networks can record urban enhancements that high-resolution models translate into spatially resolved emissions. The presentation covers optimal sensor network design and first results from the mid-cost CO₂ network in Heidelberg–Mannheim. The results demonstrate how atmospheric signals can be translated into estimates of both biogenic and urban emissions.