|
Night-Time Cloud Detection in the Arctic from AVHRR Data |
AS&M, Inc., AS&M, Inc., NASA Langley Research Center
a.mahesh@larc.nasa.gov
The cloud mask algorithm to detect night-time clouds in the Arctic requires knowledge of the radiating temperatures of the cloud, as well as that of the underlying surface(s). These temperatures can then be used as thresholds to separate cloudy pixels from clear-sky ones. Using several combinations of possible values for these temperature thresholds, a "best fit" combination can be chosen. Scene identification maps are first created for the nearest daytime image using both visible and infrared channel data. Then, cloud temperature thresholds are varied until a second scene-ID map, this time produced using only Channels 4 and 5 data, is similar to the earlier one produced from all channels. Temperature thresholds obtained in this manner can then be used to detect clouds in the night-time images. This procedure can be used throughout most of the year, except during the deep polar night, when even the nearest daytime observations can be several days or weeks away. The algorithm is used to analyze a series of NOAA-12 and NOAA-14 images taken during SHEBA in the spring and fall. The results are compared to surface-based data from the SHEBA ship, Des Groselliers, and to an operational algorithm developed to identify clouds for the Clouds and the Earth's Radiant Energy System (CERES) Project.
