Estimating biogenic CO2 fluxes for different vegetation types using CarboScope-Regional inversion (CSR) over Europe

CO₂ atmospheric inversions are typically used to derive Net Ecosystem Exchange (NEE) estimates from atmospheric observations, while prescribing the anthropogenic component or resolving for CO₂ emissions using proper proxy data. Year-to-year variability of the biosphere sink is primarily caused by climate variations such as drought events, heat waves, and seasonal changes. However, the response of biospheric sink to climate conditions varies spatially across lands depending on vegetation type. In this study, we modified our CSR experiment by augmenting the state space of a standard CO₂ inversion enabling the separate optimization of seven vegetation types within CSR, which increases the state space by a factor seven. The diagnostic biosphere model VPRM is used to provide a priori fluxes of CO₂ for the targeted vegetation classes, which are derived from the remote sensing data. Posterior flux estimates demonstrate the contributions of CO₂ estimates from evergreen, deciduous, mixed forests, as well as shrublands, savanna, croplands, and grasslands. The results indicate that the largest flux adjustments are associated with croplands over Europe, suggesting a shift from being largest sink in the prior model to the largest source of CO₂. A similar analysis at the national scale over Germany shows substantial flux adjustments in croplands, which also exhibit the dominant interannual variability. Although the inversion suggests smaller corrections for mixed, evergreen, and deciduous forests, these vegetation types contribute substantially to total fluxes over Germany and display lower interannual variability than croplands. Additionally, we assess the sensitivity of the system to choices of the prior uncertainty settings.