地理学报 ›› 2004, Vol. 59 ›› Issue (4): 592-598.doi: 10.11821/xb200404013

• 土地利用 • 上一篇    下一篇

差分主成分分析法在辽河三角洲景观变化中的应用

杨翠芬,田村正行   

  1. 日本国立环境研究所,筑波 305-8506
  • 收稿日期:2003-11-17 修回日期:2004-03-18 出版日期:2004-07-25 发布日期:2010-09-09
  • 作者简介:杨翠芬(1965-),女,博士。主要从事GIS、遥感和污染生态学、环境地学等方面研究。 E-mail:yang.cuifen@aist.go.jp;Tel:+81-29-861-8190
  • 基金资助:

    日本学术振兴会外国人特别研究员资金资助

The New Method for Detecting Change of the Landscape: The Differencing Image PCA Method and Its Application in the Liaohe River Delta

YANG Cuifen, TAMURA Masayuki   

  1. National Institute for Environmental Studies, Tsukuba 305-8506, Japan
  • Received:2003-11-17 Revised:2004-03-18 Online:2004-07-25 Published:2010-09-09
  • Supported by:

    apan Society for the Promotion of Science, Postdoctoral Fellowships for Foreign Researchers

摘要:

差分主成分分析法是应用遥感数据检测景观变化的一种新方法。为了提高检测精度,我们利用TM卫星遥感数据,改进了主成分分析法和图像差值法,提出了差分主成分分析法。并以辽河三角洲地区为例,对该方法进行了验证。研究结果表明:(1) 与传统的检测法—分类后比较法相比,差分主成分分析法具有较高的检测精度,总检测精度为0.89,Kappa指数为0.82;(2) 在1984~2000年的16年间,辽河三角洲地区有近22%的景观发生了变化,主要包括芦苇湿地的减少、水稻田的增加以及城镇用地的增加。

关键词: 差分主成分分析法, 景观变化, 分类后比较法, LANDSAT/TM图像

Abstract:

This paper describes a method of the detection for the landscape change by satellite remotely sensed data. In order to improve the accuracy of detection for the landscape change, we presented the differencing image PCA method to improve the Principal Component Analysis (PCA) and the band difference of images method using the multi-temporal remotely sensed data of TM (Thematic Mapper). And the Liaohe River Delta of China was selected as a case to validate this method. The results showed that the differencing image PCA method has higher detection accuracy compared with the conventional method--post-classification change detection and the overall accuracy of the change detection reaches 0.89 and the Kappa coefficient is 0.82. The research result also showed that the landscape changed about 22% in the Liaohe River Delta area during 1984-2000. The main change is the reduction of the reed area, the increase of the paddy field and the city area.

Key words: differencing image PCA, landscape change, post-classification, LANDSAT/TM