基于多源遥感数据的水体提取方法研究
2021-08-11白翠向洋邱春霞赵贝贝张巧玲
白翠 向洋 邱春霞 赵贝贝 张巧玲



摘 要:快速准确地获取水体信息对于水资源管理利用以及灾害防治具有重要意义。利用不同分类方法探讨多种数据源在不同天气场景下水体提取的最优技术,结合Sentinel和Landsat系列数据,对红碱淖1973—2018年湖泊面积变化进行分析。结果表明:在无云的情况下,使用Sentinel-2结合最大似然分类方法提取效果最好,精度为99.30%;Sentinel-1利用面向对象分类精度最高,总体精度为95.70%,最大似然分类法次之;融合数据利用最大似然法精度最高,比Sentinel-1数据总体精度提高了2.00%;有云情况下,Sentinel-1数据通过融合同期Sentinel-2光学数据能有效提取被云覆盖的区域,其提取水体精度比仅使用Sentinel-1雷达数据精度提高了2.50%。
关键词:光学数据;雷达数据;水体提取;影像融合;红碱淖
中图分类号:TV213.4;P407.8 文献标志码:A
doi:10.3969/j.issn.1000-1379.2021.07.015
引用格式:白翠,向洋,邱春霞,等.基于多源遥感数据的水体提取方法研究[J].人民黄河,2021,43(7):78-83.
Abstract:Extraction water body information quickly and accurately has great significant to the management and utilization of water resources and the prevention and control of disasters. This study used different classification methods to explore multi-source data sources in different weather scenarios. The results show that, in the case of no cloud, for Sentinel-2 data using the maximum likelihood classification method is the best method to extract water body, with an accuracy of 99.3%; for Sentinel-1 synthetic aperture radar data using the object-oriented classification method is the best method to extract water body, with overall accuracy of 95.7%, the maximum likelihood classification is the next; for fusing data using the maximum likelihood classification method is the highest and its overall accuracy increased by 2%, compared with the Sentinel-1 data. In the case of cloud, by fusing the Sentinel-2 data over the same period, the accuracy of water extraction has been improved by 2.5% compared with using only the sentinel-1 data.
Key words: optical data; radar data; water extraction; images fusing; Hongjian Lake
水資源是人类生存的重要条件之一。我国水资源严重短缺并且严重污染[1],西北干旱地区降雨稀少。水资源是制约经济社会发展、影响生态安全的主要因素[2]。
利用遥感技术快速准确地获取水体信息,对水资源管理利用以及灾害防治具有重要意义[3]。不同传感器可以获取不同的影像,同时有许多不同的分类方法。Mcfeeters[4]提出了利用归一化差异水体指数(NDWI)来提取水体信息;徐涵秋[5]提出了改进的归一化差异水体指数(MNDWI);Feyisa等[6]提出了自动提取水体指数(AWEI)方法;骆剑承等[7]提出了“全域-局部”分步迭代的多光谱遥感水体信息高精度自动提取模型,使遥感影像的水体提取达到初步的自动化。……
