下载中心
优秀审稿专家
优秀论文
相关链接
摘要
河流网络是地表水循环的重要组成部分,如何实现河流网络动态监测已成为河流遥感研究的热点。近年来,以PlanetScope为代表的CubeSat小卫星已具备了米级空间分辨率、1 d重访周期的优势,这为河流网络高时空分辨率动态监测提供了可能。本文以青藏高原长江源区的通天河流域(227 km2)为研究区,选取2017-05—2017-10 5期3 m空间分辨率CubeSat遥感影像,增强河流横纵剖面特征自动化提取了河流网络,研究了通天河流域河流网络动态变化,对比分析了3 m CubeSat与30 m Landsat 8、10 m Sentinel-2所提取的河流网络,以及5种现有水体数据集(GRWL,GSW,FROM-GLC 2017,OpenStreetMap,HydroSHEDS)。研究结果表明:(1)研究区内河流网络5月水系密度较低(0.38 km-1),7—8月河流网络进入丰水期,水系密度显著增加至0.61 km-1,9月河流网络进入平水期,水系密度趋于平稳(0.53 km-1),随后迅速退化并于10月开始冻结,水系密度迅速降低至0.37 km-1;(2)采用高空间分辨率CubeSat所提取的河流网络能够识别更多细小河流(河宽3—30 m),CubeSat所提取的河流总长分别为Landsat 8、Sentinel-2所提取河流总长的1.6倍和1.3倍;(3)CubeSat所提取的河流网络水系密度高于现有水体数据集(2.9—12.4倍),弥补了现有水体数据集无法反映细小河流的不足。
River networks play an important role in the terrestrial water. They have become a hotspot in river remotely sensed studies on using remotely sensed imagery to monitor river dynamic changes. Recent development of CubeSat satellite, such as PlanetScope, allows monitoring of river networks at high spatial and high temporal resolution by providing near-daily revisit time imagery at 3 m spatial resolution. We selected the Yangtze headwaters (Tongtian river basin, ~227 km2) located in Tibetan Plateau as the study area. Five CubeSat images from May to October in 2017 were selected to extract river networks at 3 m resolution by enhancing the river cross sectional and longitudinal features, in order to monitor dynamic changes of in river networks at high-spatial resolution. In addition, we compared the 3 m CubeSat river networks with 30 m Landsat 8 and 10 m Sentinel-2 river networks, and the five existing hydrography data products including GRWL, GSW, FROM-GLC, OpenStreetMap, and HydroSHEDS. We concluded that: (1) Rivers in the study area begin to develop in May with drainage density of 0.38 km-1. July and August are the wet seasons, and the drainage density reaches the peak (0.61 km-1). In September, rivers reach the mean discharge with drainage density of 0.53 km-1, and then the rivers degrade gradually with drainage density of 0.37 km-1 and begin to freeze in October. (2) The high spatial resolution CubeSat river networks include more small rivers (3—30 m wide), and the CubeSat river length is 1.6 and 1.3 times larger than Landsat 8 and Sentinel-2 river networks, respectively. (3) The drainage density of CubeSat river networks is 2.9 to 12.4 times larger than existing hydrography data products, thereby compensating for any lack in the spatial and temporal resolution of the existing river network products.