Preprint finds biospheric variability masks regional anthropogenic CO2 signals in CAMS-EDGAR integration

The study integrated CAMS atmospheric reanalysis with EDGAR inventories, GOSIF biosphere data, and the Southern Oscillation Index across global grids. Unsupervised clustering identified five carbon-cycle regimes while persistence analysis quantified how spatial averaging erases local signals. Sample size comprised full global fields from 2000-2023 reanalysis; the key limitation is reliance on model-derived transport that may understate inversion uncertainties.

Findings indicate anthropogenic emission drops remain undetectable in most regimes because ENSO-driven biosphere fluxes and atmospheric mixing overwhelm them. Tropical forest clusters alone retained strong biogenic signatures, while neutral-ENSO 2020 showed no consistent regional CO2 growth-rate dip despite documented inventory declines. This aligns with prior top-down studies but adds regime-specific masking thresholds.

The analysis highlights that bottom-up inventories cannot be regionally validated by current atmospheric observations without denser in-situ networks or higher-resolution transport models. It connects to earlier mismatches reported in OCO-2 and TCCON campaigns where tropical Africa and Amazonia showed persistent inventory-atmosphere gaps exceeding 20%.

Next steps require joint assimilation of upcoming CO2M and MicroCarb satellite data with improved biosphere priors to test whether 2030 emission cuts become detectable above natural noise floors.