85th AMS Annual Meeting

Monday, 10 January 2005
Estimating clear-sky radiances from the AIRS/AMSU instrument suite
Evan Fishbein, JPL, Pasadena, CA; and L. Chen, E. J. Fetzer, and S. Y. Lee
Geophysical parameters such as temperature, water vapor and ozone profiles are retrieved from cloud-free AIRS infrared radiances. Only three percent of all AIRS footprints are sufficiently clear that successful retrievals are possible, but the fraction of retrievable footprints is increased by more than a factor of ten through a process called “cloud clearing.” Cloud-cleared radiances are the radiances that emanate from the clear portion of a footprint and are estimated using the correlation of atmospheric parameters between adjacent footprints and an estimate of the clear geophysical state from the microwave radiances. With cloud clearing we separate cloud from clear sky and therefore independently characterize the clouds and the clear sky between. In this presentation we characterize the clear-sky and cloud radiances derived by the AIRS retrieval algorithms.

Validating clear-sky radiances is made difficult by the limited correlative information. Comparing spectra from contemporaneous aircraft underpasses is especially difficult owing to the rarity of clear sky, differences in horizontal resolution between instruments and rapid changes in clouds over space and time. Therefore we use a bootstrap approach where first we validate in clear scenes, next in scenes adjacent to clear and finally in other scenes. Clear footprints are identified using empirical discriminants derived from long and short wave radiances that predict the surface skin temperature, estimate the lower tropospheric lapse rates or detect cirrus or dust spectral signatures. We first compare observed and cloud-cleared radiances in footprints we have identified as clear. Next we compare cloud-cleared radiances in footprints near clear regions and examine how cloud-cleared radiances correlate with distance from the clear region. Lastly we compare cloud-cleared radiances to calculated radiances from ECMWF analyses and forecasts. We use the differences in the clear and near clear regions to characterize errors in the calculated radiances and use these to estimate the errors in the cleared-radiances away from clear regions. We report on the spectral signatures of clouds and their dependence on cloud height and latitude. We discuss the occurrences of multiple cloud formations within footprints and the difficulty of interpreting cloud spectral signatures when multiple cloud formations are present.

Supplementary URL: