Abstract
The spatial and spectral characteristics of the GOES visible and infrared images are examined. From that analysis, a scheme is developed which identifies and separates the following classes: clear skies/no snow cover, clear skies/snow cover and clouds. Clouds are then classified as high or low broken clouds and overcast. The scheme is tested for various weather situations. Comparison of the classification results with reports from ground synoptic stations and maps reflect an average accuracy of approximately 72%, and a higher accuracy (∼87%) when high or low broken clouds and overcast are considered as one class (i.e., clouds). The differentiation between clouds and snow, or no snow-covered ground has been found to be very satisfactory, even in cases of temperature inversions.
Keywords

This publication has 0 references indexed in Scilit: