[M-GI29] Data assimilation: A fundamental approach in geosciences

[MGI29-07] Inverse analysis of atmospheric CO₂ for elucidating seasonal and interannual variations of natural carbon fluxes

★Invited Papers

*丹羽 洋介^1,2、伊藤 昭彦¹、飯田 洋介³ (1.国立環境研究所、2.気象研究所、3.気象庁)

キーワード：逆解析、二酸化炭素、炭素循環

A number of ground-based stations and observational mobile platforms have conducted atmospheric measurements of carbon dioxide (CO₂) for several decades. Those long-term observations have revealed significant interannual variations of the growth rate of the atmospheric CO₂, which suggests that the natural CO₂ flux at the earth surface have been varying in conjunction with some climate key parameters (e.g., El Niño Southern Oscillation index). Furthermore, distinct seasonal variations of atmospheric CO₂ observed by those measurements give evidence of dynamic photosynthesis and respiration activities of terrestrial biospheres. An atmospheric inverse analysis, in which an atmospheric transport model is used to connect surface fluxes with atmospheric mole fractions, is useful in that it quantitatively gives spatiotemporal variations of surface fluxes. In order to obtain implications of surface flux mechanisms, we should conduct a long-term inverse analysis so that we could analyze flux variations in correspondence with interannual climate variations. Furthermore, those flux estimations had better be made with high spatial resolutions, because a spatially small scale but quantitatively significant source/sink event such as biomass burning often plays an important role on interannual variations of atmospheric CO₂. However, there are very few global inverse analyses that can estimate fluxes at high-resolution for a long-term; this is probably because of heavy computational burden required. Recently, we have developed a sophisticated and computationally efficient inverse analysis system, which is expected to contribute such a long-term carbon budget study. This inverse analysis system is based on Nonhydrostatic Icosahedral Atmospheric Model (NICAM) and the four-dimensional variational method (4DVar). Using this system, we performed a 12-year-long inverse analysis with synthetic observations to elucidate its capability for a long-term inverse analysis. By this experiment, we found that, though certain degree of biases remain, the system could reproduce better interannual variations of CO₂ fluxes than prior estimates.