设为首页
加入收藏
Interpretation of Cloud Classification Using the Color Image Composed by Three-channel Data
-
摘要: 为更好地利用FY-3A气象卫星上中分辨率光谱成像仪 (MERSI) 资料高空间分辨率及多光谱的优势，细致分析云系在宏、微观方面的多重特征，首先利用平面平行辐射传输模式 (SBDART) 证明了MERSI的0.65，1.6 μm和11.25 μm通道能够分别反映云光学厚度、云粒子大小、云顶高度的信息，然后采用三通道合成彩色图像的方法，对FY-3A气象卫星云图进行云的分类解释判读。该技术可直观区分有云区、无云区，显示海陆分界，并且使不同云类在云图上体现为不同颜色。同时，FY-3A气象卫星高达1000 m甚至250 m的空间分辨率也使云的细致结构更为清晰，两项优势的叠加，大大提升了云的分类解释判读的准确程度和精细化水平。文中还尝试寻找典型云系的三通道特征值，以期为模糊C均值聚类 (FCM) 方法中聚类中心的选定提供经验值参考。Abstract: Clouds are the result of atmosphere dynamical and thermo-dynamical process. Different cloud types reflect different weather situations and cloud microphysical structure features. Satellite image offers large-area and all-day information of the cloud formation, gathering or disperse. Cloud classification is one of the most important objectives of the satellite image research.At present, only one single channel can be used to identify cloud category at one time. A quick, direct and accurate method of the interpretation of cloud classification has not yet been developed. It becomes particularly important to get the composed information of cloud optical thickness, effective particle radius and cloud top height information quickly.In order to better use the multiple spectral data of Medium Resolution Spectral Imager (MERSI) on FY-3A meteorological satellite to carefully analyze the macro and micro physical parameters of cloud system, according to the fact that the 0.65, 1.6 μm and 11.25 μm channel on MERSI is respectively sensitive to cloud optical thickness, effective particle radius and cloud top height based on the Santa Barbara DISORT Atmospheric Radiative Transfer, SBDART, which is that larger 0.65 μm reflectance means larger optical thickness, larger 1.6 μm reflectance means smaller effective particle radius, larger black body temperature of cloud top means lower cloud top height or warmer surface under clear air, the method of three color compositions is used to the explanation and interpretation of the cloud classification by FY-3A meteorological satellite data on 20 June 2009. The color is composed of red for visible reflectance, green for near infrared reflectance, and blue for the infrared brightness temperature. Redder means larger optical thickness, greener represents the smaller cloud top particles, and bluer means lower cloud tops or warmer surface under clear air.The technology makes different cloud types show in different colors, which is beneficial to directly distinguish cloud pixel from clear sky area, and the land boundary also can be recognized. In this case, cloud located at typhoon eye mainly shows orange-red, while spiral clouds band shows orange-yellow, indicating the reduced optical thickness compared with the typhoon eye. Mixed cloud system covering the Yangtze River Basin shows clear multiple-layer features, low layer cloud is yellow-green, middle layer cloud is dark-red, and high layer cloud shows orange-red. Over the sea, thin dark-yellow cirrus covers the marine stratiform cloud, which is white. Blue means clear air, but there is difference between sea and land because of the different temperature and reflectance.Besides the advantage of multiple spectrums, the 1000 m even 250 m resolution of FY-3A meteorological satellite makes the cloud detailed structure more clearly. The two advantages have greatly improved the accuracy of cloud classification.Meanwhile, the characteristic values for typical cloud, providing empirical values for the initial clustering center of fuzzy clustering method.
-
图 1 .65, 1.6 μm通道反射函数对云光学厚度和有效粒子半径 (单位：μm) 的敏感性分析
(θ0=20°，θ=40°，ϕ=0°)
Fig. 1 Reflection function as a function of optical thickness and effective particle radius (unit:μm) at a wavelength for 0.65 μm and 1.6 μm when θ0=20°, θ=40°, ϕ=0°
图 2 2009年6月20日10:45 FY-3A三通道合成彩色图像
Fig. 2 Color image composed by three-channel data at 10:45 20 June 2009
图 4 台风云系 (a)、大陆性多层云系 (b) 及海洋性层云、卷云及晴空陆地、海洋 (c) 图像
Fig. 4 Images of typhoon cloud system (a), continental multiple-layer cloud system (b) with matine stratiform, cirrus and clear-air land, ocean (c)
表 1 11.25 μm通道云顶黑体亮温对云顶高度的敏感性分析 (太阳天顶角θ0为20°，卫星天顶角θ为40°，相对方位角ϕ为0°)
Table 1 Brightness temperature as a function of cloud top height solar zenith angle θ0=20°, satellite zenith angle θ=40°, relative azimuth angle ϕ=0°
云底高度~云顶高度/km 1~4 2~5 3~6 4~7 5~8 6~9 7~10 云顶黑体亮温/K 272.05 266.14 260.03 253.85 247.63 241.4 235.15 
下载: 导出CSV
表 2 一些主要云和地面目标物的反射率[15]
Table 2 Reflectance of some main types of the cloud and underlying surface[15]
云和地面目标物 可见光通道反射率 积雨云 (大而厚) 0.92 积雨云 (小，云顶在6 km左右) 0.86 卷层云 (厚，下面有中低云和降水) 0.74 卷层云 (单独出现在陆地上空) 0.32 积云 (出现在陆地上空，云量＞80%) 0.69 中云 (高层、高积云，中等厚度) 0.68 层积云 (出现在陆地上空，云量＞80%) 0.68 层积云 (出现在洋面上空，成片) 0.60 层云 (厚，出现在洋面上空) 0.64 层云 (薄，出现在洋面上空) 0.42 卷云 (薄，单独出现在陆地上空) 0.36 晴天积云 (出现在陆地上空，云量＞80%) 0.29 陆地 (植被) 0.18 海洋、湖泊、河流 0.07~0.09
下载: 导出CSV
表 3 各类云的有效粒子半径参数
Table 3 Effective particle radius of different cloud types
云类 层云 (海洋上空) 层云 (陆地上空) 晴天积云 海洋积云 积雨云 浓积云 高层云 re/μm 17 10 6.7 25 33 40 8
下载: 导出CSV
表 4 典型云系像素的三通道特征值
Table 4 Characteristic three-channel values of the typical cloud system pixel
编码 RGB色彩模式强度值 R0.65 R1.64 T11.25 云的分类解释判读 A (248，98，37) 0.975 0.311 209.298 积雨云(台风眼墙) B (139，98，36) 0.532 0.311 208.344 厚卷云(台风螺旋云带) C (234，92，30) 0.918 0.289 202.21 积雨云(局地对流) D (182，75，69) 0.707 0.227 233.568 中云(浓积云) E (178，229，128) 0.691 0.793 263.753 厚低云(云体深厚的水云) F (108，95，60) 0.406 0.3 227.657 薄卷云 G (78，122，193) 0.284 0.4 288.383 薄低云(水云) H (35，42，192) 0.110 0.105 288.044 晴天海洋 I (44，89，218) 0.146 0.278 296.566 晴天陆地
下载: 导出CSV
-
[1] Kokkler R, Decotiis A G, Rao P K. A procedure for estimating cloud amount and height from satellite infrared radiation data. Mon Wea Rev, 1973, 101(2):240-243. doi: 10.1175/1520-0493%281973%29101<0240%3AAPFECA>2.3.CO%3B2 [2] Desbois M, Seze G, Szejwach G. Automatic classification of clouds on METEOSAT imagery application to high-level clouds. J Appl Meteor, 1982, 21(3):401-402. doi: 10.1175/1520-0450(1982)021<0401:ACOCOM>2.0.CO;2 [3] Welch R M, Navar M S, Sengupta S K. The effect off resolution upon texture-based cloud field classification. J Geophys Res, 1989, 94(6):14767-14781. [4] 师春香, 吴蓉璋, 项续康.多阈值和神经网络卫星云图云系自动分割试验.应用气象学报, 2001, 12(1):70-78. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=20010108&flag=1 [5] 白慧卿, 方宗义, 吴蓉璋.基于人工神经网络的GMS云图四类云系的识别.应用气象学报, 1998, 9(4):402-409. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=19980460&flag=1 [6] 李伟钢, 石立华.气象卫星云图的多分辨小波分解及人工神经网络降水估计研究.南京气象学院学报, 2000, 23(2):277-282. http://www.cnki.com.cn/Article/CJFDTOTAL-NJQX200002017.htm [7] 吴咏明, 张韧.多光谱卫星图像的一种模糊聚类方法.热带气象学报, 2004, 20(6):689-696. http://www.cnki.com.cn/Article/CJFDTOTAL-RDQX200406008.htm [8] 孙学金, 刘磊, 高太长, 等.基于模糊纹理光谱的全天空红外图像云分类.应用气象学报, 2009, 20(2):157-163. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=20090204&flag=1 [9] 周著华, 白洁, 刘健文, 等.MODIS多光谱云相态识别技术的应用研究.应用气象学报, 2005, 16(5):678-684. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=20050587&flag=1 [10] 洪梅, 张韧, 孙照渤.多光谱卫星云图的高维特征聚类与降水天气判别.遥感学报, 2006, 10(2):184-190. http://www.cnki.com.cn/Article/CJFDTOTAL-YGXB200602005.htm [11] 马芳, 张强, 郭妮, 等.多通道卫星云图云检测方法的研究.大气科学, 2007, 31(1):119-128. http://www.cnki.com.cn/Article/CJFDTOTAL-DQXK200701011.htm [12] 刘希, 许健民, 杜秉玉, 等.用双通道动态阈值对GMS-5图像进行自动云检测.应用气象学报, 2005, 16(4):434-444. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=20050454&flag=1 [13] [2010-07-12].http://fy3.satellite.cma.gov.cn/portalsite/default.aspx. [14] Rosenfeld D, Lensky I M. Satellite-based insights into precipitation formation processes in continental and maritime convective clouds. Bull Amer Meteor Soc, 1998, 79: 2457-2476. doi: 10.1175/1520-0477(1998)079<2457:SBIIPF>2.0.CO;2 [15] 陈渭民.卫星气象学.北京:气象出版社, 2003. -
点击查看大图
计量
- 摘要浏览量: 3374
- HTML全文浏览量: 1007
- PDF下载量: 1363
- 被引次数: 0
