Impacts of Increasing CO₂ Concentration on One-Dimensional Variational (1D-Var) Temperature Retrievals from AIRS Measurements

Atmospheric temperature retrieval (or temperature sounding) is accomplished with infrared observations in carbon dioxide (CO₂) absorption bands because CO₂ is well mixed gas with known concentration. Because atmospheric CO₂ concentration is steadily increasing, using a fixed CO₂ concentration may lead to an uncertainty on temperature sounding. This paper investigates the impacts of increasing CO₂ concentration on the one-dimensional variational (1D-Var) temperature retrievals from the Atmospheric Infrared Sounder (AIRS) measurements. A priori atmospheric profiles are generated by adding random forecast errors to the reference profile selected from the European Centre for Medium-Range Weather Forecasts 91-level diverse profile datasets and the AIRS observations are similarly generated by random observation errors to the simulated AIRS brightness temperatures given the reference profile. The retrieved temperatures are underestimated (overestimated) in the troposphere (lower stratosphere) with increasing CO₂ concentration. The uncertainties of the retrieved temperatures increase in the troposphere and lower stratosphere with increasing CO₂ concentration. This trend is more evident in the lower troposphere. In moist atmosphere, the accuracy of temperature sounding is slightly higher than that in dry atmosphere because the AIRS lower tropospheric temperature channels have large sensitivities to amount of water vapor. The results suggest that either adjustments to temperature sounding channels or considerations of the uncertainties in temperature retrievals in lower atmospheric layers are needed to obtain more accurate atmospheric temperature information with increasing CO₂ concentrations. In addition, it would be desirable to consider the spatio-temporal variability of atmospheric CO₂ concentration on temperature sounding.