Flux estimates of CO2 reported in literature still indicate divergence amid different inverse models. The objectives of this thesis are centered about identifying and quantifying the major drivers of the discrepancies of estimates in inversions. The methodologies applied in the experiments throughout this study relied mainly on the Bayesian inversion system CarboScope-Regional (CSR). Due to the heterogeneity of the biosphere, CO2 terrestrial fluxes are targeted to be optimised against observations of mole fractions, while ocean fluxes and anthropogenic emissions are prescribed in the inversions. Results showed that the inversions can be more sensitive to the choice of stations than to the choice of the a priori fluxes, highlighting the importance of assessing the uncertainty of flux estimates against the station set. In the second objective, the ability of atmospheric inversions to capture changes in Net Ecosystem Exchange (NEE) under climate variations was examined over Europe. Estimated NEE indicated that the biosphere productivity declined in 2018 and 2019 compared to 2006-2017, particularly during the growing season, concurrent with unprecedented increase of temperature and a decrease of soil water content. The impact of atmospheric transport, lateral boundary conditions, and configurations of inversions was investigated as the third objective. The outcome pointed to a dominant impact of atmospheric transport models on estimated fluxes. The largest contribution of transport error results from the transport schemes, while a smaller contribution is caused by differences in meteorology forcing data. Additionally, lateral boundary conditions and different inversion configurations led to a smaller but non-negligible impact in regional inversions. Finally, the effect of CO2 diurnal cycle on CO2 flux estimates was assessed in CarboScope. The findings demonstrated a substantial impact on continental and regional estimates of CO2 fluxes.
Use and reproduction:
All rights reserved