文章摘要
陈燕芬,牛振国,胡胜杰,张海英.基于MODIS时间序列数据的洞庭湖湿地动态监测[J].水利学报,2016,47(9):1093-1104
基于MODIS时间序列数据的洞庭湖湿地动态监测
Dynamic monitoring of Dongting Lake wetland using time-series MODIS imagery
投稿时间:2015-11-20  
DOI:10.13243/j.cnki.slxb.20151245
中文关键词: 洞庭湖湿地  MODIS  时间序列  波谱匹配  物候特征
英文关键词: Dongting Lake  MODIS  time series  spectral matching  phenological character
基金项目:国家自然科学基金项目(41274123)
作者单位E-mail
陈燕芬 中国科学院 遥感与数字地球研究所 遥感科学国家重点实验室, 北京 100101
中国科学院大学, 北京 100049 
 
牛振国 中国科学院 遥感与数字地球研究所 遥感科学国家重点实验室, 北京 100101 niuzg@radi.ac.cn 
胡胜杰 中国科学院 遥感与数字地球研究所 遥感科学国家重点实验室, 北京 100101
中国科学院大学, 北京 100049 
 
张海英 中国科学院 遥感与数字地球研究所 遥感科学国家重点实验室, 北京 100101  
摘要点击次数: 1577
全文下载次数: 69
中文摘要:
      本文以洞庭湖湿地为研究对象,以2001-2014年MODIS时间序列数据产品MODIS_13Q1为主要数据源,利用一年内23景时间序列增强型植被指数(EVI)数据建立不同地表类型(包括各种湿地类型)的时间序列特征参考曲线;在此基础上采用波谱匹配的最小距离分类方法对研究区进行湿地分类和制图。研究结果表明:(1)2014年研究区总体分类精度和Kappa系数分别为87.87%和0.85,表征湿地动态性特征的季节性湿地分类精度为84.85%,具有较好的分类精度,说明基于时间序列遥感数据的分类方法可以满足湿地动态性特征监测的需要;(2)2001-2014年洞庭湖天然湿地(包括永久性水域、永久性沼泽和季节性沼泽、泥滩)呈现了波动减少的趋势,其中永久性水域减少约11%,永久性沼泽减少约22%,而季节性湿地减少约13%;同时,林地则呈现了持续增长的特征,到2014年成为保护区内面积最大的覆盖类型,增长了29%;而农田面积保持了相对的稳定;(3)洞庭湖湿地水域面积缩小、季节性湿地空间位置的变化及沼泽区人工林地的大面积种植改变了洞庭湖原有生态系统格局,进而对洞庭湖湿地的生态服务功能产生潜在影响。
英文摘要:
      The popular method of wetland mapping, which is usually based on single-date satellite imagery, cannot meet the requirements of wetlands monitoring because of the highly dynamic features of wetlands. Using multi-temporal or time series satellite imagery is a good choice to monitor wetlands at large scale, and time series MODIS products with highly temporal revisiting frequency have been employed in many cases of land use/land cover mapping. Different phonological characteristics of various land cover types laid a foundation for wetlands classification using time series MODIS Enhanced Vegetation Index (EVI)products. The standard phonological curves of various wetlands categories were developed firstly on the base of training samples selected from MODIS images and Google Earth. And then, Spectral Matching of Minimum Distance(SMMD)was employed by taking temporal EVI curve of each pixel as substitute of spectral curve to classify the time series MODIS images from 2001 to 2014 in the Dongting Lake. The results show that:(1)the overall classification accuracy and Kappa coefficient of the study area was 87.87% and 0.85 respectively in 2014, and the accuracy of the seasonal wetlands was 84.85%. The good performance of SMMD approach indicates that time series MODIS imagery can be used to monitor the dynamic characteristics of wetland ecosystem. (2)There is a declining trend for those natural wetlands in the Dongting Lake using 2001-2014 MODIS imageries, including waters, seasonal wetlands and permanent marshes. Among them, Permanent water decreased by 11%, permanent marshes by 22% and seasonal wetlands by 13%,while forest land increased continually by 29%. (3)The decreased amount of water input and altered date of drying and wetting in the Dongting Lake has affected the spatio-temporal landscape patterns of Dongting wetlands ecosystem, such as the shrinkage of permanent waters, spatial transformation of seasonal wetlands and increasing forest land, which undoubtedly will give rise to changes of wetlands functions and services. Those new phenomenon should be paid more attention in future to protect biodiversities and agriculture in this region.
查看全文   查看/发表评论  下载PDF阅读器
关闭