Cloud detection and property retrieval across the day/night terminator

The problem of retrieving useful cloud information from satellites in the vicinity of the day/night terminator has been largely ignored. However, in the face of operational requirements that drive, for example, the NPOESS-C3 (0530 ascending node) orbit there is growing recognition that algorithms must be designed to operate in this difficult domain. Recent work in support of Air Force mission planning activities has resulted in development of cloud detection and cloud spatial/microphysical property retrieval algorithms specifically intended to operate across the terminator. To date these have been validated using data from the GOES and Meteosat Second Generation (MSG) geostationary platforms, though they should be generally applicable to multispectral sensor data from polar-orbiting systems as well. Detection algorithm characteristics include use of reflectance-channel data into the nighttime side of the terminator to help detect vertically developed clouds that may be directly illuminated on one side, empirical normalization functions to adjust reflectance values for solar zenith angle dependence in the 80-105° region, and midwave IR algorithms that automatically adapt to the transition from emitted plus reflected during daytime to emitted only at night. Cloud property retrieval algorithms rely on multispectral IR channel pairs at some combination of 6.7, 8.5, 10.5, and 12.0 μm, and as such are insensitive to solar illumination. Time-series plots of cloud masks show high consistency across the terminator for all cloud types including marine stratus. Comparisons of cloud top/base retrievals to ground-based lidar/radar estimates indicate good agreement, particularly for transmissive cirrus.