地理研究

地理研究 ›› 2011, Vol. 30 ›› Issue (3): 555-563.doi: 10.11821/yj2011030017

• 土地资源与利用 • 上一篇    下一篇

全国1∶25万土地覆被遥感制图精度在内蒙古东部的实地验证

廖顺宝1, 刘睿1,2, 尹芳1,2   

  1. 1. 中国科学院地理科学与资源研究所,北京 100101;
    2. 中国科学院研究生院,北京 100049
  • 收稿日期:2010-05-08 修回日期:2010-11-16 出版日期:2011-03-20 发布日期:2011-03-20
  • 作者简介:廖顺宝(1966- ),男,四川德阳人,博士,副研究员,主要从事遥感与GIS应用及地学数据质量等方面的研究。E-mail:liaosb@igsnrr.ac.cn
  • 基金资助:

    资源与环境信息系统国家重点实验室自选课题;中国科学院信息化专项(INFO-115-C01-SDB3-02)

Validation for accuracy of land cover remote sensing mapping of China at scale of 1: 250,000 in East Inner Mongolia

LIAO Shunbao1, LIU Rui1,2, YIN Fang1,2   

  1. 1. Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China;
    2. Graduate University of Chinese Academy of Sciences, Beijing 100049, China
  • Received:2010-05-08 Revised:2010-11-16 Online:2011-03-20 Published:2011-03-20

PDF (PC)

赞

被引次数

35 | 5

摘要： 土地覆被是全球变化和区域可持续发展研究的重要内容,土地覆被遥感制图是支持全球变化和区域可持续发展研究的基础数据,制图质量是相关研究成果科学性的重要保证。本文针对全国1∶25万土地覆被遥感制图的精度评价,用大类分错扣1分,同一大类中的小类分错扣0.5分的计分评价方法,共选取190个样点,在内蒙古东部进行了实地验证。验证结果表明:制图的总体精度为84.21%。其中,聚落10个样点,精度为100%;农田67个样点,精度为92.54%;湿地、水体20个样点,精度为87.50%;草地59个样点,精度为81.36%;森林34个样点,精度为66.18%。土地覆被类型错分的主要情况包括:将河湖滩地解译成沼泽、水浇地解译成水田、草甸草地解译成典型草地、典型草地解译成沙漠、森林解译成草地等。

关键词: 土地覆被, 遥感制图, 精度, 验证, 内蒙古东部

Abstract: Land cover is one of the important research fields of global change and regional sustainable development. Land cover data are the basis for supporting studies on global change and regional sustainable development. High quality of basic data is the fundamental guarantee of reliability of production of science research. From 2007 to 2009, two sets of map of land cover in China at a scale of 1:250000 respectively in the 1980s and 2005 were produced jointly by eight institutes of Chinese Academy of Sciences (CAS), including Institute of Remote Sensing Applications and Institute of Geographic Sciences and Natural Resources Research. In order to evaluate accuracy of the land cover remote sensing mapping, a field survey was carried out in Hulun Buir in east Inner Mongolia and information from 190 sampling sites was collected in August, 2009. The one or half point deduction method, which means that 1 point is deducted if the first level of category is classified incorrectly and 0.5 points deducted if the second level of category inside the same first category incorrectly for each sampling site, was applied to validate the accuracy of the mapping. The results from verification show that overall accuracy of data based on 190 sampling sites reaches 84.21%, and the accuracies of settlement, farmland, wetland/water body, grassland and forest with 10, 67, 20, 59 and 34 sampling sites reach 100%, 92.54%, 87.50%, 81.36%, and 66.18% respectively. The incorrect classification cases mainly include that (1) Bottomland of river and lake was interpreted to marsh; (2) Irrigable land was interpreted to paddy field; (3) Meadow grassland interpreted to typical grassland; (4) Typical grassland interpreted to desert; (5) Forest interpreted to grassland.

Key words: land cover, remote sensing mapping, accuracy, validation, Inner Mongolia