Journal of Resources and Ecology ›› 2011, Vol. 2 ›› Issue (2): 186-192.DOI: 10.3969/j.issn.1674-764x.2011.02.013

• Reports • Previous Articles    

Identifying Alpine Wetlands in the Damqu River Basin in the Source Area of the Yangtze River Using Object-based Classification Method

ZHANG Jiping1,2, ZHANG Yili1, LIU Linshan1, DING Mingjun1,3, ZHANG Xueru1,2   

  1. 1 Institute of Geographic Science and Natural Resources Research, CAS, Beijing 100101, China;
    2 Graduate University of Chinese Academy of Sciences, Beijing 100039, China;
    3 Key Lab of Poyang Lake and Watershed Research, Ministry of Education (Jiangxi Normal University), Nanchang 330022, China
  • Received:2010-12-30 Revised:2011-04-20 Online:2011-06-30 Published:2011-06-28
  • Supported by:

    This work was financially funded by National Natural Science Foundation of China (Grant No. 40901057) and National Basic Research Program of China (Grant No. 2010CB951704).

Abstract: Alpine wetlands are very sensitive to global change, have great impacts on the hydrological condition of rivers, and are closely related to peoples’ living in lower reaches. It is essential to monitor alpine wetland changes to appropriately manage and protect wetland resources; however, it is quite difficult to accurately extract such information from remote sensing images due to spectral confusion and arduous field verification. In this study, we identified different wetland types in the Damqu River Basin located in the Yangze River source region from Landsat remote sensing data using the object-based method. In order to ensure the interpretation accuracy of wetland, a digital elevation model (DEM) and its derived data (slope, aspect), Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), and Kauth-Thomas transformation were considered as the components of the spectral characteristics of wetland types. The spectral characteristics, texture features and spatial structure characteristics of each wetland type were comprehensively analyzed based on the success of image segmentation. The extraction rules for each wetland type were established by determining the thresholds of the spatial, texture and spectral attributes of typical parameter layers according to their histogram statistics. The classification accuracy was assessed using error matrixes and field survey verification data. According to the accuracy assessment, the total accuracy of image classification was 89%.

Key words: alpine wetland, remote sensing, object-based classification, Damqu River Basin