Inversion and Preliminary Validation for Cloud Classification and Cloud Phase Products of Fengyun-3D in CMA-NSMC
Inversion and Preliminary Validation for Cloud Classification and Cloud Phase Products of Fengyun-3D in CMA-NSMC
复制标题
DOI：
10.1109/icmo49322.2019.9026035
复制
发表时间：
2019-12
期刊：
影响因子：
--
通讯作者：
Bo Li;Ruixia Liu;Shihao Tang
中科院分区：
文献类型：
--
作者：
Bo Li;Ruixia Liu;Shihao Tang
风云三号D（以下简称风云三号D）是我国第二代极轨气象卫星的第四颗业务卫星。FY-3D中分辨率光谱成像仪（MERSI）的可见光（0.58-0.68 μm）、近红外（1.55-1.64μm和3.55-3.93 μm）和两个长波红外（10.3-11.3 μm和11.5-12.5 μm）通道可以联合使用，反演FY-3D的云分类和云相产品。利用MERSI L1资料和云掩模产品，结合地表类型、冰雪识别等一系列辅助资料，反演了FY-3D的云分类和云相。利用可见光、红外和近红外通道数据的光谱和纹理特征，采用阈值法对云像元进行云相识别，得到1 km分辨率的云相产品。然后，结合薄卷云和破碎云识别结果，采用阈值法识别高、中、低云像元，得到FY-3D 1 km分辨率的云分类产品。通过与MODIS云相产品的对比分析，可以看出，对于结构简单的云，云相产品的精度较高，而对于结构复杂的云，云相产品的精度稍低。总体而言，FY-3D对水云的识别效果较好，对冰相的识别精度略低。需要指出的是，由于算法的不同，识别混合和不确定位相的方法和标准也不同，导致FY-3D和MODIS云位相输出结果存在一定差异。
Fengyun-3D (thereafter referred as FY-3D) is the fourth operational satellite of the second generation polar orbiting meteorological satellite in China. The visible (0.58-0.68 μm), near infrared (1.55-1.64μm and 3.55-3.93 μm) and two long-wave infrared (10.3-11.3 μm and 11.5-12.5 μm) channels of the FY-3D MERSI (Medium Resolution Spectral Imager) instrument can be combined used to retrieve cloud classification and cloud phase products of FY-3D. Cloud classification and cloud phase of FY-3D are retrieved based on the MERSI L1 data and cloud mask product as well as a series of auxiliary data such as surface type and ice-snow identification. By using the spectral and texture characteristics of visible, infrared and near infrared channels data, threshold method is used to identify cloud phases of cloudy pixels, and 1 km resolution cloud phase product is obtained. Then, combined with the results of thin cirrus cloud and broken cloud identification, high, medium and low cloudy pixels are identified by threshold method and cloud classification product with 1 km resolution of FY-3D is obtained. Compared with MODIS cloud phase products, it can be concluded that the precision of cloud phase product is higher for simplestructured clouds, and slightly lower for complex clouds. In general, FY-3D has a better performance on identification of water clouds and a slightly lower accuracy for the ice phase. It should be noted that due to the different algorithm, the methods and criteria for identifying the mixed and uncertain phases are different, which results in some differences between FY-3D and MODIS cloud phase outputs.