CO
2 concentrations in the upper troposphere were retrieved from thermal infrared spectra as observed by the only spaceborne hyperspectral sounder launched in the 1990s: the Interferometric Monitor for Greenhouse gases (IMG) onboard the Advanced Earth Observing Satellite (ADEOS). First, the effective optical path difference of the IMG was evaluated because the actual instrumental line shape function of the interferometer component has not been evaluated for technical reasons in the orbit. The CO
2 retrieval method was based on the maximum a posteriori (MAP) retrieval method and on procedures to decrease errors that obstruct CO
2 signal detection. For the retrieval analysis, ERA-40 re-analysis meteorological data were used as temperature field data. A method of selecting effective channels for CO
2 retrieval was used to remove channels with a high temperature dependency and to reduce errors in estimating the water vapor, ozone, and surface temperature. Furthermore, uncertainties in temperature and other error factors, which cannot be removed through channel selection, were evaluated and optimized by treating them as components of measurement errors in the MAP retrieval. CO
2 retrieval noises of the MAP retrieval were estimated as 2.5 % and 2.0 % at pressure levels of 500 and 300 hPa, respectively. CO
2 concentrations retrieved from IMG data were compared with aircraft measurement data. Results showed that the random error in the IMG retrieval was smaller than that estimated as the a posteriori error of the MAP retrieval. No significant biases were shown compared with the margin of random errors. The CO
2 retrieval method was applied to IMG data measured in April, 1997. Although assuming a uniform CO
2 concentration as a priori, the latitudinal gradient of the zonal mean concentration was consistent with climatological features presented by previous studies at pressure levels of 500 and 300 hPa. These results suggest that thermal infrared observation by the IMG is effective for evaluating the upper tropospheric CO
2 concentration in the 1990s.
View full abstract