基于Landsat 8影像的赣抚尾闾区水体提取研究

Study on the Water Extraction Based on Landsat 8 Imagery in the Area of Tail-Streams of Gan River and Fu River

水体是人类赖以生存的重要资源。随着遥感技术的不断发展,利用遥感对水体各方面的研究也越来越多。赣抚尾闾区作为鄱阳湖入湖区的重要组成区域,包含了许多对鄱阳湖主湖区影响很大的农田和城镇,因此,选择该区域作为研究区域具有重要意义。选取Landsat 8遥感影像作为数据源,结合归一化植被指数、归一化水体指数、近红外波段反射率、自动水提取指数等方法,选择合适的阈值,构建适用于赣抚尾闾区水体提取的决策树模型,实现对赣抚尾闾区水体的提取,其精度达到94.86%。通过对结果分析,发现该方法对植被的浅水区水体的提取存在一定的局限性。

Water is an important resource for human survival. With the continuous development of remote sensing technology, the use of remote sensing in all aspects of water research is increasing. As an important component area of Poyang Lake inlet, there is a lot of farmland and towns which have great influence on the main lake area of Poyang Lake. Therefore, it is of great significance to select the region as the research area. In this paper, Landsat8 remote sensing image is selected as the data source, and the appropriate threshold is selected according to the normalized vegetation index(NDVI), normalized water index(NDWI), near infrared band reflectance and new water index(AWEI) to construct the decision tree model of the water body extraction in the tail area of Gan River and Fu River and to extract the water body in the tail area of Gan River and Fu River area, the accuracy of which is 94.86%. The study also shows that this method has some limitations on the extraction of shallow water in the presence of vegetation.