论文部分内容阅读
利用多通道地球同步卫星资料,采用多通道阈值、基于晴空合成底图和云图时间序列的阈值和相似检测、有限区域动态阈值检测等方法,运用分步决策思路,对(60°S～60°N、45～165°E)区域内卫星云图分中高云和低云进行云地分离综合处理。经人工判断与自动云地分离结果对比分析表明,准确率达到90%以上。分析表明,考虑云和地(海)表在不同通道的亮温差异和随时间演变特点,建立并运用多种云地分离方法,有利于云地分离信息量的增加;提出的考虑云系分布时变特性的相似检测法,以及从有云和无云两个角度优先选择高可信度方法逐步检测,有利于云地分离准确率的提高。
Based on multi-channel geosynchronous satellite data, multi-channel threshold, threshold and similarity detection of clear sky composite base map and cloud time series, dynamic threshold detection of limited area and so on, N, 45 ~ 165 ° E) within the region satellite cloud subdivided high cloud and low cloud for cloud separation. The comparative analysis of the results of artificial separation and automatic cloud separation shows that the accuracy is above 90%. The analysis shows that considering the difference of light temperature and the evolution with time of different channels of cloud and earth (sea) table, establishing and using a variety of cloud land separation methods is conducive to the increase of information amount of cloud land separation; the proposed cloud system distribution Similarity detection methods of time-varying characteristics, as well as the gradual detection of high confidence from the viewpoints of cloudiness and cloudlessness, are conducive to the improvement of the accuracy of cloud separation.