Cloud Detection in Nonpolar Regions for CERES Using TRMM VIRS and Terra and Aqua MODIS Data

Objective techniques have been developed to consistently identify cloudy pixels over nonpolar regions in multispectral imager data coincident with measurements taken by the Clouds and Earth's Radiant Energy System (CERES) on the Tropical Rainfall Measuring Mission (TRMM), Terra, and Aqua satellites. The daytime method uses the 0.65-, 3.8-, 10.8-, and 12.0-$\mu\hbox{m}$ channels on the TRMM Visible and Infrared Scanner (VIRS) and the Terra and Aqua MODIS. The VIRS and Terra 1.6- $\mu\hbox{m}$ channel and the Aqua 1.38- and 2.1-$\mu\hbox{m}$ channels are used secondarily. The primary nighttime radiances are from the 3.8-, 10.8-, and 12.0- $\mu\hbox{m}$ channels. Significant differences were found between the VIRS and Terra 1.6-$\mu\hbox{m}$ and the Terra and Aqua 3.8-$\mu\hbox{m}$ channels' calibrations. Cascading threshold tests provide clear or cloudy classifications that are qualified according to confidence levels or other conditions, such as sunglint, that affect the classification. The initial infrared threshold test classifies $\sim$43% of the pixels as clouds. The next level seeks consistency in three (two) different channels during daytime (nighttime) and accounts for roughly 40% (25%) of the pixels. The third tier uses refined thresholds to classify remaining pixels. For cloudy pixels, $\sim$4% yield no retrieval when analyzed with a cloud retrieval algorithm. The techniques were applied to data between 1998 and 2006 to yield average nonpolar cloud amounts of $\sim$0.60. Averages among the platforms differ by $≪$ 0.01 and are comparable to surface climatological values, but roughly 0.07 less than means from two other satellite analyses, primarily as a result of missing small subpixel and thin clouds.