干旱区生态环境知识资源中心(Arid): 多通道卫星云图云检测方法的研究

Arid

	多通道卫星云图云检测方法的研究
其他题名	The Study of Cloud Detection with Multi-Channel Data of Satellite
	马芳 1; 张强2 ; 郭铌 2; 张杰 2
ISSN	1006-9895
出版年	2007
卷号	31 期号:1 页码:119-128
中文摘要	通过对2002年7、8、9三个月，范围为（8．65°N～59．65°N，73．22°E-134．42°E）的GMS-5卫星云图3000多次数字资料的取样，根据遥感原理和样本统计特征，探讨了常用的通道阈值法云检测方法，并尝试建立了红外分裂窗通道差值法和通道综合运算法的云检测方法。通过各种检测方法比较分析后发现：对通道阈值法，只要用红外一和可见光两个通道的阈值，就可得到较好的检测效果，但用这种方法阈值要随着太阳高度角和季节的变化发生相应的变化，虽阈值变化的幅度不大，却会对云检测工作带来很大的不便。此外，该方法云检测的结果存在地理位置的影响，即检测出的云量在中低纬度偏多而较高纬度偏少。作者建立的通道综合运算云检测方法，不仅改善了地理位置的变化对云检测带来的影响，而且通过红外分裂窗通道差值检测，可减弱太阳高度角的影响，减少了检测过程中阈值变化的繁琐，同时得到了更好的检测效果，检测结果与其可见光图像中的云区相比基本符合。
英文摘要	Cloud detection is, by using various algorithms, to determine whether a pixel （or small area that is viewed by the satellite at a given time） is cloudy, clear or undecided. It plays an important role in analyzing satellite images, since the cloud influences should be removed from the images in the earth study, and then the authors can pick up clouds and to know where and how much the clouds are. Moreover, it is useful to determine the possible effects of clouds on climate in remote and mountain regions. In order to do this, the purpose of this study is to systematically present several algorithms of cloud-detection; and to find a more accurate algorithm that is helpful to understand the condition of cloud water over Qilian Mountains. In this study, the cloud detection algorithms established with the multi-channel data of GMS-5 are introduced. There are four channels including two infrared split channels, visual channel and water vapor channel in GMS-5. About 3000 digital-data samples are taken out from the satellite images. The multi-channel cloud detection contains five kinds of samples including cloud, land, water, desert and snow-pack in four satellite channels. The study period is from July 1 to September 30, 2002 and the study area is from 8. 65° N to 59. 65°N and from 73. 22°E to 134. 42°E covering the whole China （1024 * 1024 pixels）. Based on the theory of remote sensing and the statistical characteristics of samples, cloud detection algorithms are established. In the first test, the algorithm is commonly used through utilizing thresholds of the GMS-5 infrared and visual channels. The threshold values are built only from the two channels and to obtain results that contain most pixels of cloud, but the threshold values are changing with the season or solar altitude When comparing the result images with the visual images, it shows that the detected pixels decrease in higher latitudes and increase in lower latitudes. The second algorithm is designed by using the difference of gray levels between two GMS-5 infrared channels, and the threshold of infrared 1 and water vapor channel. In the detected images, the second algorithm is also affected by latitude The detected cloud pixels are less in higher latitudes, and are more in lower latitudes, which is reverse from the first algorithm. Although it has only been tested in the daytime, because the second algorithm derives little effect from visual channel, it should perform quite well in any time as well. Moreover, there are four stable thresholds in the second algorithm. The third algorithm is more accurate, which is based on the two aforesaid algorithms. Because the third algorithm inherits the strongpoint of the two aforesaid algorithms, its results show that it can offer improved technique of the cloud detection applied to GMS-5 imagery. Investigating the quality of the three algorithms, it is showed that the multi-channel incorporate method of cloud detection achieves a good agreement with the visible picture. However, it is difficult for the three algorithms to distinguish clouds from snow packs in GMS-5 images.
中文关键词	卫星云图 ; 云检测 ; 多通道 ; 灰阶值 ; 阈值
英文关键词	satellite image cleaning cloud multi-channel threshold gray levels
语种	中文
国家	中国
收录类别	CSCD
WOS类目	METEOROLOGY ATMOSPHERIC SCIENCES
WOS研究方向	Meteorology & Atmospheric Sciences
CSCD记录号	CSCD:2736034
来源机构	中国气象局兰州干旱气象研究所
资源类型	期刊论文
条目标识符	http://119.78.100.177/qdio/handle/2XILL650/220742
作者单位	1.中国气象科学研究院, 北京 100081, 中国; 2.中国气象局兰州干旱气象研究所甘肃省干旱气候变化, 减灾重点实验室, 兰州, 甘肃 730020, 中国
推荐引用方式 GB/T 7714	马芳,张强,郭铌,等. 多通道卫星云图云检测方法的研究[J]. 中国气象局兰州干旱气象研究所,2007,31(1):119-128.
APA	马芳,张强,郭铌,&张杰.(2007).多通道卫星云图云检测方法的研究.,31(1),119-128.
MLA	马芳,et al."多通道卫星云图云检测方法的研究".31.1(2007):119-128.