Improvements to cloud detection and optical properties over snow background from geostationary satellite data

Cloud properties and radiances retrieved from passive satellite measurements are increasingly being used in numerical weather predication models. With improved understanding of cloud physics, many passive retrieval algorithms have matured in recent years. At NASA Langley (LaRC), we have modified the CERES cloud retrieval algorithms used for MODIS analyses to run with geostationary satellite and AVHRR data. Multispectral techniques like the Visible Infrared Solar-infrared Split-window Technique (VISST) and Solar infrared- Infrared-Split window Technique (SIST) currently used for daytime and nighttime respectively, retrieve cloud amount, phase, height, thickness, optical depth, effective particle size, and ice or liquid water path. We are applying the algorithm to GOES-11 & 12 imager data to provide cloud properties for most of North America every half hour in real-time. The algorithms are also being applied, at lower temporal resolutions, to data taken from full-disk imagery from GOES-11/12, Meteosat-9, MTSAT, and FY-2D/E and to MODIS data to provide global coverage. LaRC's cloud properties, such as cloud height, liquid water path etc., are currently assimilated hourly into the operational development run of National Oceanic and Atmospheric Administration (NOAA) Rapid Update Cycle and Rapid Refresh (the next-generation version of RUC) weather prediction systems, that cover the continental United States of America and North America, respectively. While the assimilation of the imager cloud products have improved both analyses and forecasts of cloud-related parameters, such as ceiling heights, especially in remote regions like Alaska, the analysis algorithms still have room for improvement. One of the challenges is retrieval over snow backgrounds during daytime. The 0.65-µm spectral signatures are similar for clouds and snow. A critical input dataset for the retrieval algorithm is the clear-sky background reflectance. The accuracy of this dataset affects both the cloud detection and the retrieved optical depth. Currently a monthly static mean overhead albedo map is used and the clear reflectance calculated by applying directional models. This paper reports on the use of an automated scheme, based on results from previous hours, to update the clear-sky reflectance map and on the impact of empirical vegetation-dependent 0.65-µm snow reflectance models. Comparisons of optical depth from the current and improved method are presented. The GOES retrievals are also compared to MODIS retrievals that use other channels more sensitive to snow like 1.6-µm. Finally, the method is applied to Meteosat-9 data and the results are compared to those over snow based on the Meteosat-9 1.6-µm channel.