Toward accurate XCO2 level 2 measurements by combining different CO2 retrievals from gosat and sciamachy

Carbon dioxide (CO₂) is one of the most important anthropogenic greenhouse gases causing global warming. Although various space-based observations have been used to derive and identify regional and global distribution of CO₂ source and sink, uncertainties and biases are still subsistent. A high spatial and temporal resolution is important to remove the CO₂ uncertainties and biases. In this study, an improved ensemble median algorithm is developed by fully accounting for the variation of XCO₂ on synoptic timescales. Based on this, XCO₂ data sets from the improved method and EMMA are compared in terms of their spatiotemporal distributions and correlations. Meanwhile, XCO₂ measurements from three TCCON sites are used to compare with these two datasets. The results reveal that although there is a good correlation between the XCO₂ from improved method and that of EMMA at annual scale, large discrepancies can be still detected at most places over the world. It seems that a higher time resolution is necessary to remove the outliers. In addition, the season cycle of the two methods is generally in agreement with ground measurements. However, our improved method shows less divergence compared to that of EMMA at three ground sites. To some extent, these results proved the effectiveness of our new method and thus demonstrated the necessity for refining the Level 2 XCO₂ measurements at a shorter time averaging period.