卫星云图云检测的一种综合优化方法

Integrated optimal method of cloud detection with meteorological satellite data

为了提高云检测过程的自动化程度和云检测精度,利用FY2C、MTSAT等卫星资料以及其他辅助资料,综合采用优化的ISCCP法、多通道阈值法、空间纹理法等,找到了卫星云图云检测的一种综合优化方法,经过近两年的实际研究和业务化应用,判云的准确率达到90%以上,并得出以下结论:(1)优化的ISCCP法在云检测中的贡献最大,消除了由于地表差异带来的阈值选取困难;(2)红外、可见光合成底图对于中高云检测较有效,而红外与近红外的差值合成底图,无论对白天还是夜间的低云检测都有较好的效果;(3)空间纹理方法对海上积云的检测效果较好;(4)对于洋面上的中高云应用基于气候SST数据的阈值方法可得到较好效果;(5)严格的多通道阈值方法,可以起到有效的补充作用。

To improve the precision and to increase the automation of cloud detection, based on FY2C, MTSAT and some other assistant data, the optimal ISCCP,the multi-channel threshold, the spatial texture and some other methods were integrated to find an optimal cloud detection method. Nearly two years＇ application at the remote sensing center of Institute of Meteorology, PLA University of Science and Technology indicated the accuracy of 90%. Conclusions are drawn as follows： （1） The optimal ISCCP plays a major role in cloud detection where it relieves the difficulty in finding threshold caused by different land covers. （2） Infrared and visible statistic images are effective when detecting mid-high clouds, while the infrared-near infrared difference static images can detect low clouds at both day time and night time. （3） Spatial texture method detects clouds over the sea effectively. （4） As a result of the temperature difference between the sea and the mid-high clouds, the threshold method based on climate SST data achieves high accuracy. （5） Strict multi-channel threshold method complements this integrated method.