Assimilation of AIRS and conventional CO2 observations with an EnKF

The air samples from ~100 stations of the in situ CO2 monitoring stations around the globe most of yield biweekly CO2 concentration close to the surface. The data density is inadequate for inferences of regional and/or local sources and sinks of CO2. AIRS satellite gives around 2000 observations every 6 hours, with twice-daily overpasses. The monthly mean accuracy of AIRS CO2 is around 1.3ppm (Chahine et al., 2005), and its weighting function peaks around upper troposphere (approximately 450hPa). Therefore, AIRS CO2 observations and conventional CO2 observations are complementary with each other. In this study, for the first time, we estimate the 6-hourly global CO2 distribution by simultaneously assimilating both AIRS and conventional CO2 observations with a 64-member Ensemble Kalman Filter (EnKF) coupled with the NCAR climate-carbon model. Furthermore, meteorological observations (e.g., winds) are assimilated in conjunction with CO2 observations, which provide additional information to constrain CO2 estimation. We present results for year 2003, and emphasize the uncertainty as well as the mean states of the CO2 and meteorological analysis.