IGBP云检测网格产品升尺度方法及精度评价  

Up-scaling Method and Accuracy Evaluation of IGBP Cloud Detection Grid Products

作　　者：谢涛 任佳昊 王超[2] XIE Tao;REN Jiahao;WANG Chao(School of Remote Sensing and Geomatics Engineering,Nanjing University of Information Science and Technology,Nanjing 210044,China;School of Electronics and Information Engineering,Nanjing University of Information Science and Technology,Nanjing 210044,China)

机构地区：[1]南京信息工程大学遥感与测绘工程学院,南京210044 [2]南京信息工程大学电子与信息工程学院,南京210044

出　　处：《遥感信息》2022年第6期8-14,共7页Remote Sensing Information

基　　金：国家自然科学基金面上项目(42176180)。

摘　　要：国际地圈生物圈计划(international geo-biosphere program, IGBP)云检测网格产品是利用多特征阈值方法得到的云检测结果,云检测产品作为其他大气产品与陆地产品反演的辅助数据,也是目前云检测研究的检验依据。文章面向IGBP云检测网格产品的特点,提出了迭代聚类的升尺度方法。在此基础上,为在无真实参考数据条件下有效评价升尺度方法的效果,基于欧式距离(Euclidean distance, ED)设计了类内欧氏距离(inner class ED,ICE)和类间欧氏距离(ED between classes, EBC)两种精度评价方法。实验基于风云三号卫星数据,利用ICE、EBC,并结合占比法、相似度、互信息和均方根误差对不同升尺度方法结果开展了综合对比分析与讨论,验证了所提出方法的可行性与有效性。The cloud detection grid product of the international geo-biosphere program(IGBP) is the cloud detection result obtained by using the multi-feature threshold method. The cloud detection product is not only the auxiliary data for the inversion of other atmospheric products and land products, but also the test basis of the current cloud detection research. Aiming at the characteristics of IGBP cloud detection grid products, this paper proposes an up-scaling method of iterative clustering. On this basis, in order to effectively evaluate the effect of scaling up method without real reference data, two accuracy evaluation methods of inner class Euclidean distance(ICE) and ED between classes(EBC) are designed based on Euclidean distance(ED). Based on the data of FY-3 satellite, ICE and EBC are used in the experiment to comprehensively compare, analyze and discuss the results of different scaling up methods by combining the proportion method, similarity, mutual information and root mean square error, and verify the feasibility and effectiveness of the proposed method.

关 键 词：云检测产品 升尺度 迭代聚类 类内欧氏距离 类间欧氏距离 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]