一种基于Landsat 8 OLI影像的水体信息提取方法研究

提出一种基于Landsat 8 OLI影像提取水体信息的斜率比值法（SR），比对不同方法、不同类型水体和不同季节影像的数据，结果表明：SR法相比归一化差异水体指数法（NDWI）和改进的归一化差异水体指数法（MNDWI），水体信息提取精度更高，且对阈值精确度的要求低，阈值一般设定为2.0；SR法适用于清澈水体、浑浊水体、浅水河滩等水体，对河流干流和较大支流的提取精度达到99%，细小支流、沟渠的提取精度为66%，面积较小的水体提取精度为65%；在冬季太阳高度角较低时，SR也能较好地去除阴影。

Research on A Method for Water Information Extraction by Landsat 8 OLI Image

By comparing the data from different methods, different types of water bodies and images of different seasons, a slope ratio method (SR) for extracting water information by Landsat 8 OLI image was established. Results showed that the SR method had higher accuracy in water information extraction and lower requirements on threshold accuracy than normalized differential water index method (NDWI) and improved normalized differential water index method (MNDWI). The threshold value was generally set to 2.0. The SR method was applicable to clear water, turbid water and shallow river beaches. The extraction accuracy of main stream and large tributaries of rivers was 99%, that of small tributaries and ditches was 66%, and that of small water bodies was 65%. When the sun altitude angle was low in winter, the method could also remove shadows better.