欢迎访问《生态学杂志》官方网站,今天是 分享到:

生态学杂志

• 研究报告 • 上一篇    下一篇

基于阈值分割的京津唐城市群生态用地多源遥感识别

彭建1,2**,吕慧玲2,马晶1,刘焱序1   

  1. 1北京大学城市与环境学院,  地表过程分析与模拟教育部重点实验室,  北京 100871; 2北京大学深圳研究生院城市规划与
    设计学院, 城市人居环境科学与技术重点实验室, 广东深圳 518055)
  • 出版日期:2015-01-10 发布日期:2015-01-10

Multi-source remote sensing of ecologica land in BeijingTianjinTangshan Metropolitan based on threshold segmentation method.

PENG Jian1,2**, LU Hui-ling2, MA Jing1, LIU Yan-xu1   

  1. (1Laboratory for Earth Surface Processes, Ministry of Education, College of Urban and Environmental Sciences, Peking University, Beijing 100871, China; 2Key Laboratory for Environmental and Urban Sciences, School of Urban Planning and Design, Shenzhen Graduate School, Peking University, Shenzhen 518055, Guangdong, China)
  • Online:2015-01-10 Published:2015-01-10

摘要:

当前中国快速城市化进程中的生态环境问题日益突出,在国家生态文明建设理念及新型城镇化战略指引下,有必要通过区域生态用地的快速遥感识别为城市群生态环境问题的有效解决提供基础数据支撑。鉴于不同类型遥感数据在表征特定地物类型的精度上各有优劣,综合多源信息的区域生态用地遥感反演是当前研究的趋势所在。本研究以京津唐城市群为例,综合考虑DMSP/OLS夜间灯光数据和SPOT/VGT数据在建设用地、植被覆盖度识别方面的优点,探讨基于阈值分割法的大尺度区域生态用地多源遥感快速识别方法。研究采用SPOT/VGT数据定量识别林地、草地与耕地,进而基于DMSP/OLS夜间灯光数据区分水体与建设用地。结果表明:京津唐城市群土地覆被整体识别精度达到85.64%,Kappa系数0.771;各类生态用地识别精度均较高,其中林地识别精度最高(90.87%)、水体次之(78.33%)、草地最低(70.97%)。该方法较好地解决了单一遥感数据难以快速区分所有生态用地类型的不足,是基于全球开源数据进行大尺度生态用地快速识别的有效手段。

 

关键词: 半干旱区, 茎流, 土壤含水量, 蒸腾, 降雨

Abstract:

Eco-environmental problems have become increasingly prominent during the process of rapid urbanization in China. Quick remote sensing identification of regional ecological land can provide the basic data for solving the problems in urban agglomeration. In view of the merits and defects of different types of remote sensing data on characterization of a particular type of feature, remote sensing retrieval of regional ecological land with integrated multisource information has become a hot direction of current research. Therefore, this paper took BeijingTianjinTangshan Metropolitan as an example, considering the advantages of DMSP/OLS nighttime lights data and SPOT/VGT data in identifying construction land and vegetation coverage. The study first used SPOT/VGT data to quantitatively identify the forest land, grassland and cropland, then based on DMSP/OLS nighttime lights data to distinguish water body from construction land. The results showed that the overall accuracy of remotely sensed land cover patterns in BeijingTianjinTangshan Metropolitan reached 85.64%, and the Kappa coefficient was 0.771. The identification accuracy of forest land was the highest (90.87%), water body took the second place (78.33%), and grassland was the lowest (70.97%). This method was proved to be effective to solve the shortage of monosource remote sensing data to quickly distinguish various types of ecologica land.
 

Key words: precipitation regime, soil water content, transpiration, sap flow, semi-aridarea