摘要
选取云南省内典型的高原湖泊——滇池、洱海为研究区,以TM影像、OLI影像与GF-1影像为数据源.对于TM和OLI影像,采用NDWI、MNDWI、EWI、NWI和AWEIsh指数法提取水体,结果表明五种方法均提取出水体轮廓,AWEIsh对滇池水体的提取效果最好,提取水体完整,并有效抑制了山体阴影,其总体精度为91.71%,Kappa系数为0.88;EWI与AWEIsh对洱海水体提取效果较好,水体边界清晰,周边细小水体没有漏提,并且有效抑制了阴影及云层,其总体精度达到90%以上,Kappa系数分别为0.87和0.88.对于GF-1影像,采用单波段阈值、多波段谱间关系、NDWI、SWI指数法提取滇池流域的水体,对四种方法提取结果进行对比,结果表明NDWI法的提取效果最好,水体信息提取完整、边界清晰,有效抑制了阴影的影响,其总体精度为90.36%,Kappa系数为0.82.
This study selected Dianchi Lake and Erhai Lake,the typical plateau lakes in Yunnan Province,as the study area,and used TM OLI and GF-1 as data sources.For TM and OLI images,we extracted the information from Dianchi and Erhai by Normalized Difference Water Index（NDWI）,Modified Normalized Difference Water Index（MNDWI）,Enhanced Water Index（EWI）,New Water Index（NWI）and Automated Water Extraction Index（AWEI）.The results showed that all of these five methods can extract the contour lines of water.For DianChi Lake,with 91.71% overall accuracy and0.88 Kappa,AWEIsh is better than all the other methods,and it makes the water shape completely and eliminates shadows effectively.For Erhai Lake,EWI and AWEIsh have better effect on water information extracted.With over 90% overall accuracy and 0.87 Kappa in EWI and 0.88 Kappa in AWEIsh,they obtain clear contour lines without omissions and also eliminate the shadows and clouds.In the case of GF-1,comparing the results of single-band threshold method,multi-band spectral correlation method,NDWI and Shadow Water Index（SWI）method.With 90.36% overall accuracy and 0.82 Kappa,NDWI is a more effectively method to extract the information form water,which keeps the images complete and sharp with less effect of shadows.
作者
肖茜
杨昆
洪亮
XIAO Qian;YANG Kun;HONG Liang(School of Tourism and Geography, Yunnan Normal University, Kunming 650500, China;School of Information Science and Technology, Yunnan Normal University, Kunming 650500,China;GIS Technology Research Center of Resource and Environment in Western China of Ministry of Education,Kunming 650500, China)
出处
《云南师范大学学报（自然科学版）》
2018年第2期73-78,共6页
Journal of Yunnan Normal University:Natural Sciences Edition
基金
国家自然科学基金资助项目(41201463
41371514
41561086)
江苏省资源环境信息工程重点实验室开放基金资助项目(JS201301)
