分层抽样支持的广州市南沙区湿地景观遥感分类  

Remote sensing classification for wetlands in Nansha district of Guangzhou based on stratified sampling

作　　者：李天翔[1] 龚建周[1] 崔海山[1] 陈晓越[1] 

机构地区：[1]广州大学地理科学学院,广东广州510006

出　　处：《广州大学学报（自然科学版）》2016年第4期89-95,共7页Journal of Guangzhou University:Natural Science Edition

基　　金：国家自然科学基金资助项目(41171070);教育部人文社科资助项目(12YJC790176);广东高校省级重点平台和重大科研资助项目(2014KGJHZ009);广州市属高校科技计划资助项目(12014Z1103)

摘　　要：南沙区是广州市"南拓"战略的重点发展区域,在城市化过程中若能合理利用与保护境内具有重要生态功能的湿地资源,将有利于促进区域可持续发展.基于分层抽样技术,通过使用Erdas Imagine软件的Frame Sampling Tool工具和Landsat OLI影像,对南沙区的湿地景观进行分类.结果表明:1基于分层抽样的分类方法具有较高的分类精度,如湿地景观分类总精度为84%,Kappa系数为0.8;2该方法通过Erdas Imagine软件的Frame Sampling Tool平台可以对样本进行更有效地估计、训练及管理;3广州市南沙区内湿地资源丰富,占研究区总面积的40.84%,主要分布在珠江出海口及各支流的附近.Nansha district is an important area of Guangzhou city, which has rich wetland resources. If these wetlands could be reasonably used and protected, it would benefit the sustainable development of Nansha. Based on the Frame Sampling Tool of Erdas Imagine and data of Landsat OLI imagery, the methodology of stratified sampling was applied to remote sensing image classification of the Nansha wetlands. The research results showed that： ①The stratified sampling methodology has a high classification accuracy （an overall accuracy of 84% and a Kappa coefficient of 0. 8 were achieved in this study） ; ②This method has a high efficiency on sampie estimating, training and management via Frame Sampling Tool of Erdas Imagine software ; ③Nansha district of Guangzhou has rich wetland resources, which occupy 41.05% of the study area, and mainly distribute near the estuary and branches of the Pearl River.

关 键 词：湿地景观 遥感分类 分层抽样技术 帧采样框 广州市南沙区 

分 类 号：K903[历史地理—人文地理学]