The AIRS Science Team Version 5 retrieval algorithm has been finalized and is now operational at the Goddard DAAC in the processing (and reprocessing) of all AIRS data. Version 5 contains accurate case-by-case error estimates for most derived products, which are also used for quality control.

We have conducted forecast impact experiments assimilating AIRS quality controlled temperature profiles using the NASA GEOS-5 data assimilation system, consisting of the NCEP GSI analysis coupled with the NASA FVGCM. Assimilation of quality controlled temperature profiles resulted in significantly improved forecast skill in both the Northern Hemisphere and Southern Hemisphere Extra-Tropics, compared to that obtained from analyses obtained when all data used operationally by NCEP except for AIRS data is assimilated.

Experiments using different Quality Control thresholds for assimilation of AIRS temperature retrievals showed that a medium quality control threshold performed better than a tighter threshold, which provided better overall sounding accuracy; or a looser threshold, which provided better spatial coverage of accepted soundings. We are conducting more experiments to further optimize this balance of spatial coverage and sounding accuracy from the data assimilation perspective.

In all cases, temperature soundings were assimilated well below cloud level in partially cloudy cases. The positive impact of assimilating AIRS derived atmospheric temperatures all but vanished when only AIRS stratospheric temperatures were assimilated. Forecast skill resulting from assimilation of AIRS radiances uncontaminated by clouds, instead of AIRS temperature soundings, was only slightly better than that resulting from assimilation of only stratospheric AIRS temperatures. This reduction in forecast skill is most likely the result of significant loss of tropospheric information when only AIRS radiances unaffected by clouds are used in the data assimilation process.