基于吉林一号光谱星的东北地区春季水体识别方法对比研究

Comparative Study of Spring Water Identification Methods in Northeast of China Based on Jilin-1 Spectrum Satellite

以吉林省敦化市为研究,基于吉林一号5 m分辨率遥感影像选取单波段阈值法、多波段谱间关系法、NDWI水体指数、SWI水体指数、监督分类法和非监督分类法共6种常见的水体识别方法对东北地区春季冰雪交融期最适宜的水体识别方法进行探究。通过Kappa系数、错分误差、漏分误差、面积精度等5种指标对水体识别效果进行对比分析,结果表明:最适宜的提取方法为监督分类法,提取精度较高的方法还有NDWI指数法、多波段谱间关系法。

Taking Dunhua in Jilin province as the research area,based on Jilin-15-meter resolution remote sensing images,this paper uses six common water recognition methods,including single-band threshold method,multi-band spectral relationship method,NDWI water index,SWI water index,supervised classification method and unsupervised classification method to explore the most suitable water recognition method for spring ice-snow melting period in northeast China.The effect of water identification was compared and analyzed by five indexes,including Kappa coefficient,commission,omission,area accuracy,etc.The results show that the most suitable extraction method is the supervised classification method,and NDWI water index method and multi-band spectral relationship method are the other methods with high accuracy.