摘要
开展Sentinel-1A SAR数据在洪水淹没范围提取和水体变化监测方面的应用研究,对科学有效地管理洪涝灾害有重要意义。合成孔径雷达以其不受天气影响、能穿透云层、覆盖面积广等特点成为灾害监测的重要数据来源。面向对象的方法能有效解决影像的椒盐现象被广泛运用于信息提取研究。本文基于Sentinel-1A SAR数据,利用面向对象的方法构建洪水淹没范围提取流程,绘制灾前、灾中、灾后水体变化监测图,对比分析基于传统像元的提取方法,实现对广西临桂会仙岩溶湿地区域不同时期洪水动态监测。研究表明,Sentinel-1A SAR数据在洪水监测领域有巨大的应用潜力,相较于传统基于像元的方法,面向对象的方法能有效抑制杂斑生成,提高空间信息的利用效率,具有更好的提取精度。
In recent years, flood disasters have emerged successively, so extracting the disaster range and study- ing detection of the change of water body are of great significance to the scientific and effective management of flood disasters. Synthetic Aperture Radar （SAR） has the ability of penetrating clouds and generating ground in- formation and it can collect data from large areas under any weather conditions. SAR has become an important means of natural disasters observation. Sentinel- 1A is an important part of satellite in ESA＇s Copernicus Pro- gramm for monitoring environment. Therefore, discussions of the application of Sentinel-1A SAR data to the monitoring of flood disaster is meaningful. The traditional pixel-based methods have limitation in image speckle noise suppression, so the object-oriented method is introduced in this study. This study takes Sentinel-1A SAR images of the pre-flood, flooding and after-flood periods as data source. We introduce the object-oriented classifi- cation software- eCognition 9.0 to establish an extraction process of flooding area. Firstly, we use the SNAP software to preprocess the Sentinel- 1A SAR images. Then, to account for objects features, we conduct experi- ments using multi-scale segmentation of eCognition 9.0, comparative analysis of the experimental results and we use the prior knowledge to determine the optimal segmentation scale. Based on the assigned class function of eCognition 9.0, we analyze the VV polarization mean feature to extract flooding area. Meanwhile, we use DEM data to remove the hill-shade in water extraction. Using the methods above, the detection of flood dynamic changes of different periods and flood submerge area extracted in the Huixian karst wetland in Guangxi was real- ized. Maps of the detection of flood changes in the pre-flood, flooding and after-flood periods were plotted. At last, we compared the traditional pixei-based water extraction method （Unsupervised classification and OSTU threshold segmentation）. This research shows that Sentinel-IA SAR data has a great application potential in flood detection field. Compared with the traditional pixel-based method, the object-oriented method can effec-tively avoid the ＂salt and pepper phenomenon＂, have high precision in flooding area extraction and can improve the utilization of spatial information.
作者
汤玲英
刘雯
杨东
陈乐
苏扬媚
徐宪立
TANG Lingying1, LIU Wen1,2, YANG Dong1, CHEN Le1, SU Yangmei1, XU Xianli3(1. College of Resources and Environmental Science, Hunan Normal University, Changsha 410081, China; 2. Key Laboratory of Geospatial Big Data Mining and Application, Changsha 410081, China;3. Institute of Subtropical Agriculture, Chinese Academy of Sciences, Changsha 410125, Chin)
出处
《地球信息科学学报》
CSCD
北大核心
2018年第3期377-384,共8页
Journal of Geo-information Science
基金
国家自然科学基金项目(41501478)
湖南省重点学科地理学(2016001)
湖南省自然科学基金项目(2015JJ6068)~~
