Evaluating AIRS Radiometric Error in Non-uniform Scenes using MODIS

Abstract:

The Atmospheric Infrared Sounder (AIRS) on the EOS Aqua Spacecraft was launched on May 4, 2002. AIRS acquires hyperspectral infrared radiances in 2378 channels ranging in wavelength from 3.7-15.4 um with spectral resolution of better than 1200, and spatial resolution of 13.5 km with global daily coverage. The AIRS was designed to measure temperature and water vapor profiles for improvement in weather forecast and improved parameterization of climate processes. Currently the AIRS Level 1B Radiance Products are assimilated by NWP centers worldwide and have shown considerable forecast improvement. AIRS L1 and L2 products are widely used for studying critical climate processes related to water vapor feedback, atmospheric transport and cloud properties. AIRS trace gas products include ozone profiles, carbon monoxide, and the first global maps of mid-tropospheric carbon dioxide.

The AIRS radiances are calibrated using a uniform on-board blackbody and full aperture space view. For this reason, all radiometric measurements assume a uniform scene. As with most instruments, the AIRS 2D spatial response functions (tophat functions) are not flat for all channels, nor are they the same. When viewing a non-uniform scene, this causes a radiometric error that is scene dependent and cannot be removed without knowledge of the scene response. The magnitude of the error depends on the non-uniformity of the AIRS spatial response and the non-uniformity of the scene, but typically only affects about 1% of the data.

In this effort we use data from the MODIS instrument to provide information on the scene uniformity that can be used to correct the AIRS data. Early results show we can match the AIRS and MODIS radiances to about 0.6K when we include the AIRS tophat functions in the normalization of the MODIS data (Elliott, Proc SPIE 6296, 2006). The method requires use of different infrared bands in MODIS depending on the channels of AIRS being corrected. Resulting improvement in noise and bias will be presented by comparing to the new AIRS Level 1C product that uses PC techniques to correct impacted channels. The method can be used to recover impacted channels, validate the Level 1C product, and identify scene conditions where the error is most significant.