Determination of the best optimal estimation parameters for validation of infrared hyperspectral sounding retrievals

Abstract

The availability of hyperspectral infrared remote sensing instruments, like AIRS and IASI, on board of Earth observing satellites opens the possibility of obtaining high vertical resolution atmospheric profiles. We present an objective and simple technique to derive the parameters used in the optimal estimation method that retrieve atmospheric states from the spectra. The retrievals obtained in this way are optimal in the sense of providing the best possible validation statistics obtained from the difference between retrievals and a chosen calibration/validation dataset of atmospheric states. This is demonstrated analytically. To illustrate this result several real world examples using IASI retrievals fine tuned to ECMWF analyses are shown. The analytical equations obtained give further insight into the various contributions to the biases and errors of the retrievals and the consequences of using other types of fine tuning. Retrievals using IASI show an error of 0.9 to 1.9 K in temperature and below 6.5 K in humidity dew point temperature in the troposphere on the vertical radiative transfer model pressure grid (RTIASI-4.1), which has a vertical spacing between 300 and 400 m. The more accurately the calibration dataset represents the true state of the atmosphere, the better the retrievals will be when compared to the true states.