[M-IS05] Environmental, Socio-economic, and Climatic Changes in Northern Eurasia

[MIS05-P14] A global 0.025 x 0.025 degree resolution inverse modeling of biospheric CO₂ fluxes and comparison to the observations

*Shamil Maksyutov¹, Lorna Nayagam¹, Rajesh Janardanan¹, Tomohiro Oda², Jiye Zeng¹, Motoki Sasakawa¹, Tsuneo Matsunaga¹ (1.National Institute for Environmental Studies, 2.Earth from Space Institute, Universities Space Research Association)

Keywords:carbon dioxide, net ecosystem exchange, MODIS , inverse modeling

Simulating atmospheric CO₂ concentration at fine spatial (kilometer) scales is often critical for retrieving anthropogenic CO₂ emissions using in-situ and satellite observations as well as estimating terrestrial biospheric uptake in the presence of anthropogenic CO₂ contributions. Established model-simulated and/or ML-based upscaled ecosystem global CO₂ flux products are only available at 0.05° and coarser resolutions. We prepare a global 10-day mean net ecosystem exchange (NEE) CO₂ flux climatology data at a 0.025° resolution by combining the gross primary productivity (GPP) estimates from the 500 m MODIS v.6.1 GPP dataset and the ecosystem respiration estimates from a 0.1° resolution upscaled flux product. The MODIS GPP annual total is adjusted by 20% globally to match with the upscaled flux product. The respiration estimate is also scaled to ensure the neutral annual NEE at each grid point. The developed flux climatology data is used in the NIES-TM-FLEXPART-VAR (NTFVAR) inverse model, which is applied to estimate carbon fluxes at a 0.025° resolution for 2015-2019 using the Obspack-CO₂ data for observations, together with ODIAC fossil, GFAS fire and NIES ocean flux datasets used as prior emissions. The comparison of the simulated CO₂ to the observations shows a good fit. We find good model performance at continental sites, such as Berezorechka, West Siberia, where prior flux datasets often underestimate the amplitude of the seasonal cycle. Notably, the summertime biases there are below 1 ppm after the flux correction by the inverse model. The NEE flux optimization with the inverse model provides a global 10-day mean 0.025° resolution flux dataset that is consistent with both the observed atmospheric CO₂ variations and fine-scale patterns of the vegetation productivity.