引用本文:
【打印本页】   【HTML】   【下载PDF全文】   查看/发表评论  【EndNote】   【RefMan】   【BibTex】
←前一篇|后一篇→ 过刊浏览    高级检索
本文已被:浏览 168次   下载 121 本文二维码信息
码上扫一扫!
分享到: 微信 更多
基于高分辨率遥感影像的青海湖沙柳河流域土地覆盖监督分类方法对比
成淑艳1,2, 曹生奎1,2, 曹广超1,2, 韩建平3, 汉光昭1,2, 吴方涛1,2
1.青海师范大学 地理科学学院, 青海 西宁 810008;2.青海师范大学 青海省自然地理与环境过程重点实验室, 青海 西宁 810008;3.青海省第二测绘院, 青海 西宁 810008
摘要:
[目的]研究高寒河源区土地覆盖的精准分类方法,为地物分类提供参考,分类结果可为青海湖流域土地资源与生态环境监管提供数据支撑。[方法]运用高分辨率遥感影像,通过6种监督分类器(平行六面体、最小距离、马氏距离、最大似然、神经网络和支持向量机)对青海湖沙柳河流域的土地覆盖进行分类,最后通过最佳提取方法统计得出青海湖沙柳河流域土地覆被概况。[结果]支持向量机、神经网络和最大似然的分类精度较高,其总体分类精度均大于96%,Kappa系数均大于0.95,平行六面体的分类精度最低,误差较大。综合各种分类精度及分类图像局部细节,支持向量机分类效果最佳。通过解译可知沙柳河流域土地覆被以草地为主,占流域总面积的71.09%,裸地和湿地分别占流域总面积的16.26%和10.24%,水体、农田和建筑用地面积较小,共占流域总面积的2.41%。[结论]运用高分辨率遥感影像,通过支持向量机分类器可实现对青海湖沙柳河流域土地覆被的良好分类。整个流域植被覆盖度高,生态环境良好。
关键词:  监督分类  分类精度  高分辨率遥感影像  青海湖沙柳河流域
DOI:10.13961/j.cnki.stbctb.2018.05.042
分类号:P23
基金项目:青海省“高端创新人才千人计划”(青人才字[2016]11号);青海省“135高层次人才培养工程”;高分专项青海高分遥感数据产业化应用项目(94-Y40G14-9001-15/18);青海省重点实验室发展专项资助
Comparisons of Supervised Classification Methods for Land Cover Based on High Spatial Resolution Remote Sensing Images in Shaliu River Basin of Qinghai Lake
CHENG Shuyan1,2, CAO Shengkui1,2, CAO Guangchao1,2, HAN Jianping3, HAN Guangzhao1,2, WU Fangtao1,2
1.College of Geography Science, Qinghai Normal University, Xining, Qinghai 810008, China;2.Qinghai Province Key Laboratory of Physical Geography and Environmental Process, Qinghai Normal University, Xining, Qinghai 810008, China;3.Second Institute of Surveying and Mapping Qinghai Province, Xining, Qinghai 810008, China
Abstract:
[Objective] To study the precise classification method of land cover in the alpine river-source area in order to provide references for future land classification and provide data support for land resources and ecological environment supervision in the Qinghai Lake basin.[Methods] High spatial resolution remote sensing images were used to derive the land cover information through six different kinds of supervised classifiers (parallel hexahedron, minimum distance, Mahalanobis distance, maximum likelihood, neural network and support vector machine). And the land cover status of the Shaliu river basin in Qinghai Lake was statistically obtained through the best extraction method.[Results] The classification accuracy of support vector machine, neural network and maximum likelihood was high, the overall classification accuracy was greater than 96%, and Kappa coefficient was greater than 0.95. The classification accuracy of the parallel hexahedron was the lowest, and the error was largest. Support vector machine showed the best classification effect by combining all kinds of classification accuracy and classification image local details. Through interpretation, it showed that the land cover of the Shaliu river basin was dominated by grassland, and accounted for 71.09% of the total area of the basin. Bare land and wetlands accounted for 16.26% and 10.24% of the total area of the basin, respectively. Water, farmland, and construction land was small, accounted for 2.41% of the total basin area.[Conclusion] By using high spatial resolution remote sensing images, a good classification of land cover in Shaliu river basin of Qinghai Lake was achieved by support vector machine classifier. The whole basin has a high vegetation coverage and good ecological environment.
Key words:  supervised classification  classification accuracy  high spatial resolution remote sensing images  Shaliu river basin of Qinghai Lake