Possible thin cirrus cloud contamination of MODIS clear-sky pixels

Unexplained differences between modeled and satellite-observed clear-sky radiances over oceans prompt this investigation into the possibility of undetected thin cirrus contamination of MODIS clear-sky pixels. Radiances from Moderate Resolution Imaging Spectroradiometer (MODIS) channels 1, 26, 6, 7, 20, 29, 31, and 32 (0.65, 1.38, 1.64, 2.13, 3.78, 8.52, 11.0 and 12.0 µm) are compared to simulated radiances generated using a discrete ordinates radiative transfer (DISORT) code. Several scenes are chosen over the tropical and mid-latitude oceans where thin cirrus clouds are known to be ubiquitous. These scenes are classified as cloud-free by the MODIS cloud mask. Nearby Atmospheric Infrared Sounder (AIRS) profiles are found for each of these scenes, and the radiative transfer calculations are performed using the DISORT code. The correlated k-distribution method is used to correct for gaseous absorption, and the δ-fit phase function truncation method is used for the ice cloud calculations. The viewing geometry and surface properties used in the calculations are chosen to correspond with the MODIS data. This study suggests that the presence of thin cirrus clouds can at least partially explain the differences between the observed and simulated radiances. By matching the simulated radiances to the observed radiances, the optical thickness of thin cirrus clouds can be inferred. Furthermore, we examine the effects of introducing a ±1 K bias in the temperature profile and a ±10% bias in the moisture profile.