Simultaneous assimilation of AIRS Xco₂ and meteorological observations in a carbon climate model with an ensemble Kalman filter

Abstract

This study is our first step toward the generation of 6 hourly 3‐D CO 2 fields that can be used to validate CO 2 forecast models by combining CO 2 observations from multiple sources using ensemble Kalman filtering. We discuss a procedure to assimilate Atmospheric Infrared Sounder (AIRS) column‐averaged dry‐air mole fraction of CO 2 (Xco 2 ) in conjunction with meteorological observations with the coupled Local Ensemble Transform Kalman Filter (LETKF)‐Community Atmospheric Model version 3.5. We examine the impact of assimilating AIRS Xco 2 observations on CO 2 fields by comparing the results from the AIRS‐run , which assimilates both AIRS Xco 2 and meteorological observations, to those from the meteor‐run , which only assimilates meteorological observations. We find that assimilating AIRS Xco 2 results in a surface CO 2 seasonal cycle and the N‐S surface gradient closer to the observations. When taking account of the CO 2 uncertainty estimation from the LETKF, the CO 2 analysis brackets the observed seasonal cycle. Verification against independent aircraft observations shows that assimilating AIRS Xco 2 improves the accuracy of the CO 2 vertical profiles by about 0.5–2 ppm depending on location and altitude. The results show that the CO 2 analysis ensemble spread at AIRS Xco 2 space is between 0.5 and 2 ppm, and the CO 2 analysis ensemble spread around the peak level of the averaging kernels is between 1 and 2 ppm. This uncertainty estimation is consistent with the magnitude of the CO 2 analysis error verified against AIRS Xco 2 observations and the independent aircraft CO 2 vertical profiles.