摘要
根据杨存建等人发现LandsatTM影像水体具有TM2+TM3>TM4+TM5的特征,本文以2000年福州市Landsat-TM为例,分析了水体及其他几类主要地物的光谱特性在影像中的表现特征,发现除水体外,居民地的影像以及山体影像也都具有TM2+TM3>TM4+TM5(即(TM2+TM3)/(TM4+TM5)>1)的特征,但是三者(TM2+TM3)/(TM4+TM5)的比值却存在着较大的差异,所以辅以适合的阈值(水体(TM2+TM3)/(TM4+TM5)>2.0,TM2>40)可以将水体信息区别于其他背景地物。将该方法在含不同水体类型的福州市Landsat-TM遥感影像上进行实验,表明其可以将水体与全部居民地的阴影和山体阴影有效区分开来;同时可用于快速、简便和准确地提取城市和山区中的湿地水体信息,解决水体提取中难于消除居民地阴影与山体阴影的难题。
With the fast development of aviation and space technology, remote sensing technology became irreplaceable and advanced means in the fields of global changing and population-resources-environment. Utilizing the satellites data, it is possible to draw water information and macroscopic monitored water resource, assess the range of flood, monitor dynamic change of wetland resource and monitor the changes of coastline, which got extensive studies and applications during the last decades. The macroscopic studies of water are closely related to the accurately drawn range and boundary line of water information, which are especially important in the research of monitoring coastline change and flood. The study of the extraction of water information from satellite data has been further investigated by many schol- ars in the past. The most common methods are spectrum classifier （maximum likelihood classifier is most extensive）, normalized difference water index （NDWI）, ratio classifier and so on. However, how to overcome the interference of the shade has always been a difficult problem in the terrestrial water body information extraction. Such as maximum likelihood classifier, which was a traditional statistics categorised method, took all massif shade as lake information. Through experiment, Professor Xu found that the result, using NDWI index to extract water information, contained a lot of city building information, so he believed that the NDWI index was not suitable for extracting water information in the range of the constructed urban areas. The paper analyzed the spectrum characteristic of the water body and several other main surface features based on Yang Cunjian＇s viewpoint of water having characteristics of TM2 ＋ TM3 〉 TM4 ＋ TM5 in TM imagery. It found that water body, residential area and the shadow of mountains all have the characteristic of TM2 ＋ TM3 〉 TM4 ＋ TM5, but the ratio of （TM2 ＋ TM3 ）/（TM4 ＋ TM5 ） differs greatly between them, so it was easy to separate water body from all other surface complement only by assisting suitable threshold value （for example, the suitable threshold value of water body is （ TM2 ＋ TM3 ） / （ TM4 ＋ TM5 ） 〉 2. 0 ）. And the experiment on TM imagery of Fuzhou city which including different types of water bodies was carried out and the result indicated that it can effectively distinguish water information from residents＇s shade or Massif shade. This method emphasized the spectrum difference law among all parts of the background features, it simplified the analytic process among the complicated tables and had reduced analytic difficulty among the tables, so it was a fast, simple and accurate water information extraction technology. It can be used to extract water information from landsate TM not only in the region containing Massif shade but also in the urban built-up land.
出处
《地球信息科学》
CSCD
2008年第5期663-669,共7页
Geo-information Science
基金
福建省科技厅资助项目(2005K024)
福建师范大学地理科学学院研究生创新基金资助项目
关键词
水体识别
光谱特征
比值型指数
TM影像
water information identified
spectral signature of TM
ratio index
fuzhou city
