On the Retrieval of Cloud Properties Over Snow and Ice-Covered Surfaces Using Satellite Imager Data

Retrieval of cloud properties over snow-covered surfaces is a lingering challenge to the satellite remote sensing community. At visible wavelengths (~0.65 µm), snow and clouds typically have similar reflectances and the latter may often appear darker than the latter. This makes the retrieval of cloud optical depth (COD) using the visible channel an extremely uncertain and often impossible task. A variety of solar reflectance spectral channels that are relatively insensitive to snow offer the opportunity to address this retrieval challenge. Despite their availability on satellite imagers for more than a decade, no systematic evaluation of their utility for cloud retrievals over snow has been conducted. The Clouds and the Earth's Radiant Energy System (CERES) Program Edition 4 cloud analysis system is currently using the MODIS and VIIRS 1.24-µm channels as a substitute for the 0.64-µm channel for retrievals over snow and ice scenes. In the Shortwave-infrared Infrared Near-Infrared Technique (SINT), the 1.2, 3.8, and 10.8-µm channels provide the primary information about COD, cloud effective radius (CER), and cloud effective temperature (CET), respectively. The SINT can also use the 1.6, 2.1, and 2.25-µm channels in place of the 1.24-µm channel. In this study, the theoretical capabilities of these channels are evaluated and compared; the SINT is applied using the various channels from MODIS and VIIRS, and the results are compared to each other, to estimates of ice cloud COD from CALIPSO, and to other retrievals from surface-based measurements. The shortcomings and potential for an optimal method are discussed.