Impact of independent satellite bias correction method on CO2 flux inversion

Satellite observations are expected to play an important role in studying carbon fluxes. However, it is necessary to properly remove spatiotemporal bias from these observations. Herein, we estimated the spatiotemporal bias in satellite XCO2 data by making an inversion of in-situ observations (JMA XCO2). We performed in-situ inversion (CNTL), inversion of in-situ observation and NIES GOSAT XCO2 ver.2.97-8 (RAW), inversion of in-situ observation and replaced GOSAT XCO2 with JMA XCO2 (JMA), inversion of in-situ observation and fixed value bias corrected version of GOSAT XCO2 (FIX), inversion of in-situ observation and spatial bias correction of GOSAT XCO2 (ALL), and inversion of in-situ and spatial and seasonal bias correction of GOSAT XCO2 (MAV). Using these bias correction methods, we performed inverse analysis using both in-situ and satellite observations. Compared to previous studies, the introduction of satellite observations also suppressed large global variations in the regional CO2 flux, except in regions with few observations. The largest changes tend to be for regions lacking in-situ data (South America, Africa, and the Southern Ocean). As this result was almost the same in the JMA experiment in which satellite observation data were replaced by the JMA XCO2, it is considered that the difference is not due to errors in the bias correction. In contrast, the relatively small changes in the case of the CNTL and RAW experiments indicate that the original satellite data did not efficiently constrain the regional fluxes. Therefore, including satellite data that have been corrected for bias realizes improved estimation of the CO2 fluxes in less-constrained regions. This study shows that after correction using the proposed method, the estimated CO2 flux can be used to provide improved estimates of global CO2 fluxes and concentrations; however, the results also show a need for increasing CO2 observation sites especially less-constrained regions and improving analysis systems.