基于TM影像的玛多县湿地变化监测方法

Monitoring methods for wetland changes of Maduo County of China based on Landsat TM data

以玛多县为典型研究区,利用2010年Landsat TM数据比较分析了谱间关系法、归一化水体指数(NDWI)、修正归一化水体指数(MNDWI)及非监督分类4种分类方法提取湿地信息的准确度。结果表明,在较大显著湿地区域,4种方法信息提取的准确度都比较高;在海拔4 500m以上且地形复杂地区,MNDWI方法的精度明显高于其他3种方法,达到86%;在中部地势较低区域,MNDWI方法不如NDWI方法的准确度高,但优于谱间关系法及非监督分类方法。总体来说,MNDWI方法能很好地将云和地形对湿地信息提取的影响降到最低,筛选最佳湿地信息提取方法有助于保护青藏高原湿地资源。

The remote sense is a useful method to extract the information of wetland. To select an optimal method for extracting wetland information in the Tibetan Plateau, this study tested the accuracy of spec- trum, normalized difference water index, modified normalized difference water index and unsupervised classification to extract the wetland information with Landsat TM data in the Maduo County of Qinghai Province. This study showed that the four methods successfully extracted the wetland information at a large scale. The accuracy of MNDWI to extract the wetland information was better than that of other three methods when the wetland distrihuted at above 4 500 m elevation regions and its accuracy was 80%. At low and mid regions, the accuracy of NDWI was higher than that of MNDWI, but the latter was higher than those of spectrum and unsupervised classification. This study suggested that the MNDWI was better to extract the wetland information due to minimizing the negative effects of the clouds and terrain.