Water Resource Monitoring Exploiting Sentinel-2 Satellite and Sentinel-2 Sat-ellite Like Time Series;Application in Yangtze River Water Bodies

• 作者: Huber CLAIRE Julien BRIANT Kunpeng YI Lei CAO Studer MATHIAS Yésou HERVé
• 来源期刊: 测绘学报（英文版）
• 年，卷（期）: 2020,（4）
• 页码范围: 41-49
• 页数: 9页
• 语种: 英文

摘要

As part of the Dragon 4 project, the water extents of Wuchang and Shengjin lakes have been extracted from Sentinel-2 time series, using all exploitable images since the beginning of the acquisitions in 2015. The aim of the study is to assess the capability of the Sentinel-2 constellation and Landsat 8 over the Anhui region, especially the high temporal resolution. A total of 32 dates have been used and 10 Landsat 8 images ( Libra) have been added to try to reduce the temporal gaps in the Sentinel-2 acquisitions caused by cloudy conditions. Extractions were done using a SERTIT-ICube automatized routine based on a supervised Maximum Likelihood Classification. These extractions allow to recreate the dynamic of the two lakes and show the drought and wet periods. During the 3 years interval, the surface peaks in July 2016 for both lakes. The lowest level appears at two different dates for each lake;in January 2018 for Wuchang, in February 2017 for Shengjin. Wuchang Lake surface area appears to be more variable than Shengjin Lake, with many local maximum and minimum between the end of 2017 and April 2018. In the case of Wuchang Lake, floating vegetation is a problem for automatic water surface area extraction. The lake is covered by vegetation during long periods of time and the water below can' t be detected by automatic radiometric means. Nevertheless, Sentinel-2 stays a pertinent and powerful tool for hydrological monitoring of lakes confirming the expectation from the remote sensing wetland community before launch. The presence of NIR and SWIR bands induces a strong discrimination between water and other classes, and the systematic acquisitions create dense time series, making analysis more consistent. It makes possible to sensor events occurring over short periods of time. Thanks to this a link can be done between endangered bird species, such as the Siberian Crane and the Lesser White-Fronted Goose and periodically flooded areas. These midterm results illustrated the pertinence and powerful of multi-source optical satellite data for environmental analysis.