Pixel‐scale assessment and uncertainty analysis of AIRS and MODIS ice cloud optical thickness and effective radius

Abstract

AbstractComparisons of collocated Atmospheric Infrared Sounder (AIRS) and Moderate Resolution Imaging Spectroradiometer (MODIS) ice cloud optical thickness (τ) and effective radius (re) retrievals and their uncertainty estimates are described at the pixel scale. While an estimated 27% of all AIRS fields of view contain ice cloud, only 7% contain spatially uniform ice according to the MODIS 1 km optical property phase mask. The ice cloud comparisons are partitioned by horizontal variability in cloud amount, cloud top thermodynamic phase, vertical layering of clouds, and other parameters. The magnitudes of τ and re and their relative uncertainties are compared for a wide variety of pixel‐scale cloud complexity. The correlations of τ and re between the two instruments are strong functions of horizontal cloud heterogeneity and vertical cloud structure, with the highest correlations found in single‐layer, horizontally homogeneous clouds over the low‐latitude tropical oceans. While the τ comparisons are essentially unbiased for homogeneous ice cloud with variability that depends on scene complexity, a bias of 5–10 µm remains in re within the most homogeneous scenes identified, consistent with known radiative transfer differences in the visible and infrared bands. The AIRS and MODIS uncertainty estimates reflect the wide variety of cloud complexity, with greater magnitudes in scenes with larger horizontal variability. The AIRS averaging kernels suggest scene‐dependent information content that is consistent with infrared sensitivity to ice clouds. The AIRS‐normalized χ2 radiance fits suggest that accounting for horizontal cloud variability is likely to improve the AIRS ice cloud retrievals.