Comparison of TCCON and GOSAT Column Averaged CO2 to Global Atmospheric Transport Modeling

The measurements of column averaged CO2 contain ground surface CO2 flux signals. Comparing these observations to modeling results helps us to better understand the distributions and variations of CO2 sources and sinks and to evaluate the model transport. In this study, we ran a global atmospheric transport model (PCTM) from 2003 to 2011, using a near-balanced bottom up CASA/GSFC-GFED2 CO2 emissions and uptake as the biospheric part of total CO2 flux. All the transport modeling and biospheric CO2 fluxes are forced or driven by NASA MERRA reanalysis data. The PCTM model captures the observed CO2 seasonal cycles and inter-hemispheric gradients at TCCON sites well, within about a ppmv in most instances. The model also agrees very well with observations in the phase and amplitude of synoptic variations, showing high spatial and temporal correlations. These results suggest that the PCTM model has a good skill in capturing variable processes at different atmospheric levels, including the surface level CO2 signals. This study also shows that the CO2 fluxes used in the modeling (primarily the biospheric ones) provide a reasonably good prior representation of the CO2 flux distribution globally. Comparison analysis of GOSAT XCO2 data to PCTM modeling is mainly done on the monthly basis. The GOSAT XCO2 data used are from the JPL ACOS team's version 2.9. We co-sampled the modeling results with GOSAT XCO2 measurements spatially and temporally. Results show the GOSAT retrievals have the ability to capture the seasonal cycles globally, generally presenting reasonable positive correlations with modeling, with some persistent negative biases. On the synoptic scale, GOSAT XCO2 shows obvious discrepancies with modeling in some areas, suggesting possible large uncertainties in the XCO2 retrievals. We also compared the column CO2 data to output from a perturbed CO2 flux model to test the sensitivity of the observations in detecting small flux changes. These sensitivity experiments will help us learn how CO2 flux variations impact the CO2 distributions in the atmosphere and provide us some insights in inverse modeling of GOSAT CO2 to characterize CO2 sources and sinks.