During the Quality Measurement Experiment, a detailed analysis of the output from the AERI Line-by-Line Radiative Transfer Model (LBLRTM) indicated that the error between the observed radiance from the AERI and that calculated by the LBLRTM is primarily due to errors in specifying the atmospheric state. In particular, errors in determining the water vapor profile have a profound influence on the radiance residuals, introducing both a bias and increasing the variability. The radiosonde’s water vapor calibration has been observed to fluctuate between radiosonde calibration lots (called batches), as well as within each calibration batch. To reduce the variability and bias introduced into the QME AERI/LBLRTM radiance residuals, the moisture profiles from each radiosonde are scaled such that its total precipitable water vapor matches that retrieved from the microwave radiometer (MWR), and these scaled profiles are used to drive the model.
For more details, see http://science.arm.gov/vaps/lssonde.stm