Evaluation of Arctic Meteorological Fields in Reanalyses Using Satellite Observed Clear-sky Hyperspectral Radiances

Abstract:

Reanalysis has been widely used in meteorological and climate studies. Thus, it is critical to assess reanalysis meteorological fields against observations as much as possible. Given the scarcity of in-situation observations in the polar regions, satellite observations offer best hope for such assessment. Following Chen et al. (JGR, 2018), this study evaluates the temperature and specific humidity profiles in three reanalyses, ECMWF ERA-Interim, ERA-5, and MERRA-2, by comparing observed channel-averaged AIRS L1b spectra with simulated counterparts for all clear-sky AIRS footprints collocated with the CERES on Aqua observations over the Arctic (60°N-90°N) in January 2005. A subset of AIRS channels sensitive to temperature, CO₂, or H₂O but not to other trace gases are grouped based on the peak pressure level of their weighting function, and based on their sensitivities to temperature (T) and to relative humidity (RH). The brightness temperature (BT) difference between synthetic AIRS spectra using AIRS L2 and AIRS L1 spectra is used as a representation of observation uncertainty. Our results show that the radiometric uncertainties are within ± 0.35 K for all groups. For three reanalyses, the BT difference between synthetic and observed AIRS spectra are generally within the uncertainty range for three T-sensitive groups except for ECMWF-Interim between 200 and 400 hPa. In contrast, the BT difference for RH-sensitive groups are all negative and are generally outside of the uncertainty range. To correct the negative bias in BT differences, the specific humidity profiles over 200-600 hPa need to be reduced by 8-13%, 6-16%, and 6-26% in ERA-Interim, ERA-5, and MERRA-2, respectively. The correction for the BT difference over 600-800 hPa may rely on the accuracy of surface skin temperature.