摘要
干旱半干旱地区、低密度不透水面地区、山区等困难地区具有灯光较弱、空间布局分散、周边遮挡物较多等特点,目前已有的大尺度不透水面研究中对困难地区不透水面的提取效果较差。鉴于此,提出了以多时相升降轨Sentinel SAR数据和光学影像为主,辅以灯光、DEM数据为阈值筛选的提取方法。对中国延安、越南南部、阿联酋迪拜、孟加拉国南部等困难地区进行了不透水面提取,并将提取结果与2010年GlobalLand 30产品进行了对比。结果表明,平均精度可达88.58%,平均Kappa系数可达0.76。该方法能增强困难地区不透水面提取效果,并有较高的提取精度,对于大尺度乃至全球不透水面的提取具有重要的指导意义。
Diffcult regions like arid and semi-arid regions,low-density impervious surface regions and mountainous areas have features such as weak lighting,scattered spatial layout and many surroundings obstructions.At present,the largescale extraction researches of impervious surface perform badly on diffcult region.In order to solve these problems,we proposed a threshold ltering method based on multi-temporal Sentinel SAR data and optical images,supplemented by lights and DEM data.Taking Yan’an Province,southern Vietnam,Dubai and southern Bangladesh as research areas,we extracted the impervious surface,and compared the extraction results with GlobalLand 30 products in 2010.e results show that the average precision and Kappa are above 88.58%and 0.76.It is concluded that the proposed method can enhance the effect of impervious surface extraction,and has a key guiding signi cance for the large-scale and global impervious surface extraction.
出处
《地理空间信息》
2019年第9期95-98,102,I0003,共6页
Geospatial Information
基金
国家重点研发计划“全球变化及应对”重点专项课题资助项目(2016YFA0600302)
中国科学院国际合作局对外合作重点资助项目(131C11KYSB20160061)
