摘要
水体边界的精确提取在水资源调查和检测中具有重要意义.由于各种水体提取方法设计时基于不同的多光谱传感器和不同的区域类型,因此研究其在相同影像的适用性很有必要.研究基于Landsat 8 OLI影像数据,以北京及周边为研究区,选取4类常见水体区域类型,即面状水体的水库区域、条带状水体的河道区域、山区阴影水体的山区区域、规则划分水体的规则水体区域,进行水体提取方法研究.通过比较谱间关系法(MSR)、归一化差异水体指数(NDWI)、改进的归一化差异水体指数(MNDWI)、波段特征指数(PRWI)、增强型水体指数(EWI)、城镇水体指数(CIWI)、高斯归一化水体指数(GNDWI)、自动水体提取指数(AWEI)共8类水体提取方法在4个常见水体提取特征区域的一致性及误差.研究得出,水库区域最佳方法为MNDWI、河道区域最佳方法为CIWI、山区区域最佳方法为MNDWI、规则水田区域最佳方法为MSR,以及在4个水体区域综合最佳方法为PRWI.此外,在水体面积较小的研究区域中使用EWI提取水体,可以取得较高的一致性.AWEI方法不适合地形多变的研究区;而在GNDWI方法中水的值域有2段,即需要4个阈值才能将水提出,相较于其他水体指数增加了误差,同时造成该方法在本研究区精度较低.
Accurate extraction of water body boundaries is of great significance in water resources survey. Many water body extraction methods may not be generally applicable,because the design of each method is based on different types of multispectral sensors,and developed in different regions,with different types of water. This study is intended to determine the applicability of these water body extraction methods in different types of water area based on Landsat 8 OLI images of Beijing. Four common types of water area,including reservoir,river,water in mountainous areas and paddy fields are selected.The consistency and errors of eight kinds of water body extraction methods,including multi-band spectral relationship(MSR),normalized difference water index(NDWI),modified normalized difference water index(MNDWI),PRWI,enhanced water index(EWI),combined index of NDVI and NIR for water body identification(CIWI),gaussian normalized water index(GNDWI) and automated water extraction index(AWEI) used in four types of water area are compared. The results show that the best method for reservoir is MNDWI,the best method for river is CIWI,the best method for water in mountainous area and paddy field are MNDWI and MSR,respectively.The best stable method for four kinds of water area is PRWI.In addition,EWI can be used to extract water bodies for smaller water areas,which can achieve high consistency.AWEI method is not suitable for water in variable terrains. When using GNDWI,four thresholds of water need to be determined.Compared with other water indices,GNDWI has higher error and lower accuracy.
作者
邓智天
孙永华
邱琦
孙薇
张冬冬
李明超
DENG Zhitian;SUN Yonghua;QIU Qi;SUN Wei;ZHANG Dongdong;LI Mingchao(College of Geospatial Information Science and Technology,Capital Normal University.Beiing 100048,China;College of Resource Environment and Tourism,Capital Normal University,Beijing 100048,China;Bejing Laboratory of Water Resource Security,Capital Normal University,Beijing 100048,China)
出处
《河北师范大学学报(自然科学版)》
CAS
2020年第3期267-276,共10页
Journal of Hebei Normal University:Natural Science
基金
十三五国家重点研发计划课题(2017YFC0406004,2017YFC0406006)
北京市教委科技计划(SQKM201710028013)。
关键词
水体边界
自动化
水体提取
适用性
水体指数
OLI影像
water body boundary
automated
water body extraction
applicability
water indices
OLI image