Preprints
https://doi.org/10.5194/egusphere-2026-6982
https://doi.org/10.5194/egusphere-2026-6982
20 Nov 2026
 | 20 Nov 2026
Status: this preprint is open for discussion and under review for Atmospheric Measurement Techniques (AMT).

An intercomparison of modern GOSAT XCO2 data products

Christopher W. O'Dell, Thomas E. Taylor, Gregory McGarragh, Peter Somkuti, Hiroshi Suto, Yu Someya, Yukio Yoshida, Hartmut Bösch, Robert Parker, Maximilian Reuter, Stefan Noël, André Butz, Andrew Barr, Jochen Landgraf, Nobuhiro Kikuchi, Kei Shiomi, Sudhanshu Pandey, and David Crisp

Abstract. Space-based observations of column-mean dry-air mole fraction of carbon dioxide (XCO2) from multiple satellites such as GOSAT, GOSAT-2, OCO-2, and OCO-3 are often used to infer global and regional-scale carbon fluxes. Several algorithms have been developed to retrieve XCO2 from these measurements, and they make different choices regarding the retrieval itself, pre- and post-filtering, and post-retrieval bias correction. These choices can yield differences in sounding-matched XCO2 as well as in the location and density of quality-filtered soundings. Both of these traits, actual XCO2 as well as good sounding spatiotemporal distributions, can lead to differences in inferred carbon fluxes. In this work, we present an extensive intercomparison of XCO2 from six retrieval algorithms operating on data from the GOSAT satellite, over the period 2010–2022. The retrieval algorithms include the FOCAL v4 algorithm, the University of Leicester v7.3 algorithm, the NIES v3.05 algorithm, the JAXA EORC v3 algorithm, the RemoTeC v2.4.1 algorithm, and the ACOS v11 algorithm. These algorithms differ in terms of sounding selection, CO2 and meteorological priors, molecular spectroscopy, assumptions regarding cloud and aerosol scattering, and post-retrieval filtering and bias correction. For each XCO2 product, we evaluate throughput, accuracy, temporal stability, and robustness of their posterior uncertainty estimate. These metrics are obtained from comparisons with co-located Total Carbon Column Observing Network (TCCON, GGG2020) data and with XCO2 from a suite of carbon inverse models. The models discriminate among the products more effectively than TCCON because of their superior global coverage. We find that the six products differ significantly in terms of all these metrics, though it is often difficult to identify the cause of the differences (e.g. forward model, filtering, and bias correction) without further research. These large discrepancies suggest that poorer-performing algorithms should be examined with scrutiny, with the hope that all such algorithms can be improved, to maximally leverage observations from both present and future XCO2-sensitive satellites in informing the global carbon cycle.

Competing interests: At least one of the (co-)authors is a member of the editorial board of Atmospheric Measurement Techniques (Andre Butz).

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.
Share
Christopher W. O'Dell, Thomas E. Taylor, Gregory McGarragh, Peter Somkuti, Hiroshi Suto, Yu Someya, Yukio Yoshida, Hartmut Bösch, Robert Parker, Maximilian Reuter, Stefan Noël, André Butz, Andrew Barr, Jochen Landgraf, Nobuhiro Kikuchi, Kei Shiomi, Sudhanshu Pandey, and David Crisp

Status: open (until 26 Dec 2026)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
Christopher W. O'Dell, Thomas E. Taylor, Gregory McGarragh, Peter Somkuti, Hiroshi Suto, Yu Someya, Yukio Yoshida, Hartmut Bösch, Robert Parker, Maximilian Reuter, Stefan Noël, André Butz, Andrew Barr, Jochen Landgraf, Nobuhiro Kikuchi, Kei Shiomi, Sudhanshu Pandey, and David Crisp
Christopher W. O'Dell, Thomas E. Taylor, Gregory McGarragh, Peter Somkuti, Hiroshi Suto, Yu Someya, Yukio Yoshida, Hartmut Bösch, Robert Parker, Maximilian Reuter, Stefan Noël, André Butz, Andrew Barr, Jochen Landgraf, Nobuhiro Kikuchi, Kei Shiomi, Sudhanshu Pandey, and David Crisp
Metrics will be available soon.
Latest update: 20 Nov 2026
Download
Short summary
Satellites measuring sunlight reflected from Earth's surface can estimate atmospheric carbon dioxide, and several groups have developed methods to do this. We compared six methods applied to more than a decade measurements from Japan’s Greenhouse gases Observing Satellite, or GOSAT. They differ far more than expected in how much good data they reported and most importantly the size of their errors.  This is of importance for estimating carbon sources and sinks from satellites.
Share