In this study, the Greenhouse Gases Monitoring Instrument (GMI) data collected from GaoFen-5 (GF-5) satellite was applied to the cloud detection study in O₂ A band. The detection results were compared with the cloud judgment product of the moderate resolution imaging spectroradiometer (MODIS), and the proposed algorithm can filterout 90% of the clear sky data. Based on this, a more in-depth study of atmospheric CO₂ retrieval was carried out. The CO₂ retrieval results were compared with those data collected from the Total Carbon Column Observing Network (TCCON) and Greenhouse Gases Observing Satellite, and the results showed that the average accuracy of CO₂ retrieval results was better than 1%. In addition, the correlation coefficient between the results of CO₂ retrieval method and the data collected from TCCON was 0.85. Due to the effect of vegetation chlorophyll fluorescence, the CO₂ retrieval results based on the GMI data (mainly collected from the Park Falls and Sodankyla sites) were higher than TCCON collected data. After the corrections to reduce the effect of vegetation chlorophyll fluorescence, the correlation coefficient of the CO₂ retrieval results between GMI and TCCON collected data increased to 0.91.