Quantifying the benefit of total column CO2 observations for constraining the global carbon budget: An inter-comparison study of top-down flux estimates based on GOSAT observations with bottom-up flux estimates from MsTMIP

Space-based remote sensing observations, such as those available from the Greenhouse gases Observing SATellite “IBUKI” (GOSAT) hold great promise for improving the scientific understanding of carbon cycle processes and budgets at regional and global scales. The degree to which the GOSAT CO2 total column (XCO2) observations can constrain global fine-scale fluxes with reasonable precision and accuracy, and the degree to which the dense but lower precision GOSAT data provide additional information relative to the high precision but sparse in situ observations, remain topics of ongoing research.

In this study, XCO2 observations retrieved via the GOSAT-ACOS B3.4 algorithm, the Total Column Carbon Observing Network (TCCON) XCO2 retrievals, and CO2 measurements from surface flask sites are assimilated using a geostatistical ensemble square root filter (GEnSRF) to estimate global surface fluxes at high spatial and temporal resolutions (spatial: 1° × 1.25°; temporal: daily). Fluxes are estimated over a period of four consecutive years (January 2009 – December 2012), with only the in situ and TCCON observations constraining the first year surface fluxes, while fluxes for the remaining estimation periods are constrained by all three sets of observations. The estimated fluxes are compared with a suite of bottom-up estimates based on a combination of biospheric fluxes from models participating in the Multi-scale Synthesis and Terrestrial Model Intercomparison Project (MsTMIP) plus anthropogenic flux estimates from the Open-source Data Inventory for Anthropogenic CO2 (ODIAC).

Because GEnSRF has been designed to estimate fluxes independently of any a priori flux estimates from flux models and/or inventories, this data assimilation tool allows for a completely independent comparison with the bottom-up estimates. Evaluation of the posterior CO2 concentrations with independent aircraft observations demonstrate that the information from GOSAT is particularly valuable for constraining fluxes: (a) during the summer season over the land, and (b) across all seasons over the oceans; in both cases, this is due to the large number of good-quality retrievals available from GOSAT relative to the limited data available from the surface network. A time-averaged analysis of the sensitivities of the GOSAT observations to underlying fluxes further confirms that the GOSAT observations contribute more to the assimilation over both land and ocean during the Northern Hemisphere summer than winter. Finally, the inter-comparison to bottom-up flux estimates reveal that the GOSAT XCO2 observations are consistent with a shift in uptake from the tropical to the temperate lands during the boreal summer.