基于双极化Sentinel-1A影像数据的鄱阳湖水域提取与汛期变化检测

The Water Extraction and Flood Season Changes Detection of Poyang Lake Based on Dual Polarized Sentinel-1A Image Data

摘要: 合成孔径雷达(SAR)卫星影像数据具有不受气候条件限制的优势，且具有较高的重访率，可以在汛期恶劣气候条件下及时获取灾区影像，为科学开展抗洪救灾提供依据和信息保障。面向汛期应急监测和水域变化分析的应用需求，本文选取位于江西省九江市的鄱阳湖区域为研究区域，基于汛前6月25日和7月7日水位高峰期获取的两景双极化SAR影像数据，综合利用H/A/Alpha分解结果开展极化目标分析，并基于Wishart (距离最短原则的)非监督分类提取水体，成功获取了汛期前后空间分辨率为15 m水域空间分布图，经统计分析探明九江市鄱阳湖区域汛期新增面积达56 km2。实验结果表明：极化分解技术的介入，对于波浪较大和含沙量较高的水域识别具有较好的适应性，能正确判识和保留该类水体目标；同时，该途径还能够有效降低雷达阴影区造成的误判。

Abstract: Synthetic Aperture Radar (SAR) satellite image data has the advantage of being unrestricted by climatic conditions, and has a high revisit rate. It can obtain images of disaster areas timely under severe weather conditions in flood season, providing an evidence and information insurance for scientific flood fighting and disaster relief. According to the application requirements of flood emergency monitoring and water area changes analysis, this paper selects the Poyang Lake area, located in Jiujiang City Jiangxi Province, as the research area. Based on the two-view dual-polarization SAR obtained from the peak water level on June 25 and July 7 Image data, by using H/A/Alpha decomposition results comprehensively to carry out polarization target analysis, and depended on Wishart (distance principle) unsupervised classification to extract water bodies, we successfully obtained the spatial distribution map with spatial resolution of 15 m before and after the flood season. By statistical analysis, it shows that the newly increased area of the Poyang Lake area in Jiujiang City has reached 56 km2. The experimental results show that the intervention of polarization decomposition technology has better adaptability to the identification of waters with larger waves and higher sediment concentration, and also can correctly identify and retain such water targets. Meanwhile, this method can also effectively reduce the misjudgment that caused by the radar shadow area.