Research On Cloud Detection And Classification Of Meteorological Satellite Image

Meteorological satellite remote sensing since the1960s, It was a major breakthrough inthe Detection of meteorological technology. Meteorological satellite’s cloud images itprovides in terms of time and space continuity is can not compare with others detectionmeans. Meteorological satellites have some advantages，for instance, Detected in all the time,large area, short period and real-time, Plays a crucial role in the field of weather forecasting,To make up71percent the surface area of the earth’s oceans、plateau and desert、sparselypopulated areas’ regular meteorological data’s shortage.Meteorological satellite plays an irreplaceable role. For example, Use the image ofmeteorological satellite to perform calculations air temperature and humidity profiles,analysis of weather systems, study the relationship between precipitation and moisturetransport. It can continuously track the formation of clouds and the development of weathersystems, research the relationship between them.Use the computer to realize the information in the image meteorological satellite of thecloud detection and classification automatically extracted and quantitatively judgedThis paper introduces Fengyun series meteorological satellite’s basic situation andcharacteristics, Detail of the meteorological satellite’s image processing, Discuss the basicprinciples and the common method of clouddetection and classification of meteorologicalsatellite image.In addition, combined with the practical application of the GuangdongMeteorological Bureau, Detailed summary of how to detect and classify the clouds by theFY-3meteorological satellite’s image. To compare the method of this article form NSMC’sThe main research as follows:（1）to generalize and analysis of domestic and abroad common、sophisticated clouddetection algorithm at this stage, Combined with multi-spectral channels characteristic onFY-3satellite VIRR,Reference the realization process of mature cloud detection algorithm,Design and develop a simple and quick multispectral integrated threshold cloud detectionalgorithm to use, Compared the reflectivity of clouds during the day and night to classify theclouds. （2）Successfully introduced the concept of confidence interval, Establish credibilityrate of clear sky, Description cloud test results, Set the threshold of cloud detection todetermine the by confidence interval，Reduce the misjudgment by use absolutely threshold（3）Generated the cloud classification products by the algorithms in this paper Theproduct can be used as auxiliary data for other remote sensing products.