A Pattern Recognition Technique for Distinguishing Surface and Cloud Types in the Polar Regions

摘要: Abstract Measurement of polar cloud cover is important because its strong radiative influence on the energy balance snow and ice surface. Conventional satellite detection schemes often fail in regions visible thermal contrasts between surface are typically small. Nevertheless, experts looking at imagery can distinguish clouds from by examining textural characteristics scene. This paper describes an automated pattern recognition algorithm winch identities various types high latitudes visible, near-infrared, infrared AVHRR data. Five spectral features give information about magnitude albedos brightness temperatures, while three describe variability “bumpiness” a The maximum likelihood decision rule used to classify that region into one seven categories or 11 categories. was able 870...