地球信息科学学报 ›› 2016, Vol. 18 ›› Issue (4): 553-563.doi: 10.3724/SP.J.1047.2016.00553

• • 上一篇    下一篇

多分辨率遥感土地覆被数据质量综合评价——以湖南省桃源县为例

许光明1,2(), 杨雅萍2,4,**(), 杨飞2,4, 荆文龙2,3, 常中兵2,5   

  1. 1. 陕西师范大学旅游与环境学院,西安 710062
    2. 中国科学院地理科学与资源研究所,北京 100101
    3. 中国科学院大学,北京 100049
    4. 江苏省地理信息资源开发与利用协同创新中心,南京 210023
    5. 河南大学环境与规划学院,开封 475001
  • 收稿日期:2015-04-16 修回日期:2015-11-19 出版日期:2016-04-20 发布日期:2016-04-19
  • 通讯作者: 杨雅萍 E-mail:xgming_0609@163.com;yangyp@igsnrr.ac.cn
  • 作者简介:

    作者简介:许光明(1990-),男,硕士生,主要从事资源开发与GIS研究。E-mail: xgming_0609@163.com

  • 基金资助:
    国家科技基础性工作专项项目“格网化资源环境综合科学调查规范”(2011FY110400);国家地球系统科学数据共享平台、中国工程科技知识中心地理资源分中心建设项目;江苏省地理信息资源开发与利用协同创新中心资助项目

Quality Assessment of Multi-resolution Remote Sensing Land Cover Data: A Case Study in Taoyuan County of Hunan Province

XU Guangming1,2(), YANG Yaping2,4,*(), YANG Fei2,4, JING Wenlong2,3, CHANG Zhongbing2,5   

  1. 1. College of Tourism and Environment, Shaanxi Normal University, Xi′an 710062, China
    2. Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China
    3. University of Chinese Academy of Sciences, Beijing 100049, China
    4. Jiangsu Center for Collaborative Innovation in Geographical Information Resource Development and Application, Nanjing 210023, China
    5. College of Environment and Planning, Henan University, Kaifeng 475001, China
  • Received:2015-04-16 Revised:2015-11-19 Online:2016-04-20 Published:2016-04-19
  • Contact: YANG Yaping E-mail:xgming_0609@163.com;yangyp@igsnrr.ac.cn

摘要:

评价土地覆被数据质量是正确、合理使用数据的前提和保障,有助于遥感制图方法的改进。本文以1:10万土地利用数据为参考数据,选取2010年RapidEye_5 m、FROM_GLC(30 m)、MODIS_V005(500 m),以及2009年GlobCover 2009(300 m)土地覆被数据,以湖南省桃源县为例对4种不同分辨率的遥感土地覆被数据质量,引入窗口的分类类别统计方法,进行了综合评价,并分析了其误差和空间分布。结果表明:(1)RapidEye_5 m数据总体精度最高,MODIS_V005和FROM_GLC次之,GlobCover 2009数据相对最低。高分辨率土地覆被数据对于居民地、交通用地、水体等精细地物分类较好,具有一定优越性,各数据在一级类上的面积相关性和一致性总体高于二级类;(2)各数据在建筑用地和其他未利用地类型上的生产者精度均较低,FROM_GLC和MODIS_V005数据,在灌木草地上的空间一致性较差,4种数据在以耕地为主的平坦地区空间一致性较好,混淆主要发生在灌木草地、乔木林地和耕地之间;(3)随着土地覆被数据分辨率的提高,分出较多地物类型数的面积比例也随之增加,较高分辨率的RapidEye_5 m和FROM_GLC分出的类别数集中在7-16种较高水平上,低分辨率数据集中在1-5种较低水平上,在丘陵山区差异显著,而高分辨率数据对地物区分更好,集中于11-16种地物。

关键词: 土地覆被数据, 多分辨率, 遥感, 质量评价, 桃源县

Abstract:

The quality assessment of remote sensing land cover data is the premise and guarantee of using it reasonably, and it's helpful to improve remote sensing mapping methods. In this study, the 1:100000 land use data of Taoyuan county in 2010 was used as the reference data to validate four different resolution land cover data: RapidEye_5 m, FROM_GLC (30 m), GlobCover2009 (300 m) and MODIS_V005 (500 m). We evaluated the four different resolution land cover data from three aspects, including area relevance and consistency, spatial consistency, and window analysis based on the conversion of classification systems. The results show that: the overall accuracy of RapidEye_5 m data is the highest, MODIS_V005 and FROM_GLC are intermediate, and GlobCover2009 data is relatively lower. The land cover data with higher resolution have a certain superiority for classifications of residence, transportation land, water and other fine material, and the area relevance and the overall consistency of primary types is higher than secondary types. Producer accuracy and user accuracy of the four types of land cover data in crop land, woodland and water is better, while in the construction land and other unused land is lower. Moreover, the spatial consistency of FROM_GLC and MODIS_V005 data is poor in the shrub grassland. The spatial consistency of the four different resolution land cover data is better in the flat areas. Confusion occurred mainly among shrub grassland, woodland and crop land. With the increase of resolution for land cover data, more and more land cover types can be distinguished. The number of land cover types separated from RapidEye_5 m and FROM_GLC (30 m) land cover data focuses on the range of 7-16, in contrast, data with lower resolution focuses on the range of 1-5. Furthermore, data with higher resolution are better to distinguish the grand objects in hilly and mountainous areas.

Key words: land cover data, multi-resolution, remote sensing, quality assessment, Taoyuan county