Choice of satellite-based CO2 product (XCO¬2, vertical profile) alters surface CO2 flux estimate

The ACOS (Atmospheric CO2 Observations from Space) algorithm provides column-averaged CO2 products in units of dry-air mole fraction (XCO2) based on GOSAT radiances. However, XCO2 is derived from a linear transformation of the CO2 vertical profiles estimated from the ACOS retrieval algorithm. In theory, XCO2 vertical columns should provide no more information than the original CO2 profiles. However, the different sensitivities of either CO2 profiles or XCO2 to transport errors can significantly alter surface CO2 flux estimates. Though it has been argued that XCO2 may be less sensitive to transport error than CO2 vertical profiles, there is no study so far investigating the actual impact on surface CO2 flux estimation due to the choice of observation format, which could have significant impact on future satellite CO2 profile mission concepts. In this presentation, we will present the sensitivity of surface CO2 flux estimation to a suite of CO2 observation products, which includes CO2 vertical profiles, XCO2, and the lowest 3 levels of CO2 from CO2 vertical profiles. The CO2 observations are ACOS products covering from July 2009 to June 2010. We will present both OSSE and real observation experiments. In the OSSE experiments, we will present both perfect model experiments and experiments with model errors that are introduced by changing the planetary boundary height. In the real observations, we will show the annual and seasonal CO2 flux as function of regions from using the three observation products. The accuracy of CO2 flux estimation will be examined by comparing CO2 concentrations forced by posterior CO2 flux to independent CO2 observations. The surface CO2 flux estimation framework is based on GEOS-Chem adjoint model that is developed by the Carbon Monitoring Study flux pilot project.