欢迎访问林草资源研究

林业资源管理 ›› 2017›› Issue (4): 103-109.doi: 10.13466/j.cnki.lyzygl.2017.04.016

• 科学研究 • 上一篇    下一篇

基于时序NDVI数据的洞庭湖区湿地植被类型信息提取

刘晓农1(), 邢元军1, 罗鹏2   

  1. 1.国家林业局中南林业调查规划设计院,长沙 410014
    2.中国林业科学研究院资源信息研究所,北京 100091
  • 收稿日期:2017-04-12 修回日期:2017-05-12 出版日期:2017-08-28 发布日期:2020-09-24
  • 作者简介:刘晓农(1963-),男,湖南郴州人,高工,学士,主要从事遥感与信息技术在林业中的应用研究。Email:714234493@QQ.com
  • 基金资助:
    国家高技术发展计划(863计划)(2013AA102605);国家自然科学基金(31170637)

Wetland Plant Extraction Based on the Time Series Landsat NDVI in Dongting Lake Area

LIU Xiaonong1(), XING Yuanjun1, LUO Peng2   

  1. 1. Central South Forest Inventory and Planning Institute of State Forestry Administration,Changsha 410014,China
    2. Research Institute of Forest Resource Information Techniques,CAF,Beijing 100091,China
  • Received:2017-04-12 Revised:2017-05-12 Online:2017-08-28 Published:2020-09-24

摘要:

洞庭湖湿地是我国及国际重要的湖泊湿地,基于遥感时空融合模型,通过融合高时间分辨率的MODIS数据与中等空间分辨率的Landsat数据,得到时序Landsat NDVI数据,并利用时序Landsat NDVI数据对湿地植被信息进行提取。研究结果表明,该方法能够有效提取研究区湿地植被类型,总体分类精度与Kappa系数分别为91.52%与0.85,较单时相Landsat8 OLI光谱影像总体分类精度与Kappa系数分别提高了4.16%和0.03。苔草沼泽、芦苇沼泽、杨树林沼泽和水稻田几种湿地植被的分类精度提高较为明显,用户精度分别提高了2.35%,0.67%,10.47%和4.75%,生产者精度则分别提高了3.57%,2.31%,10.11%和6.21%。研究结果可为阴雨天气较多的南方地区的湿地信息提取提供有效的技术和方法。

关键词: 时序序列, NDVI, STARFM, 洞庭湖区, 湿地植被

Abstract:

As an important ecological system,wetland of lake groups and river system in Dongting Lake area is essential for the ecological environment.Due to the continuous disturbance of human activities and globe climate change,wetland in Dongting Lake area has degraded and it’s urgent to monitor the wetland change timely.In this paper,we used Landsat8 OLI data and MODIS data to get the time series Landsat NDVI data based on spatial and temporal adaptive reflectance fusion model (STARFM).Then,the Savitzky-Golay (S-G) filter was employed to smooth the time series Landsat NDVI data.With the phonological calendar of plant wetland and the computation of Jeffries-Matsushita distance (J-M),and through selecting validation data randomly throughout the study area for many times,we got the best J-M distance and the optimal Landsat NDVI data combination.Support vector machine was used to map wetland distribution of study area.Results showed that this method could map wetland fields effectively,and get a high overall precision of 91.52% with the Kappa coefficient of 0.85,and overall accuracy and Kappa coefficient were improved about 4.16% and 0.03,respectively,compared with using single date Landsat8 OLI spectral data.Especially,the precision of plant wetland,such as sedge,reed,polar and paddy,were improved about 2.35%,0.67%,10.47% and 4.75% for user accuracy and 3.57%,2.31%,10.11% and 6.21% for producer accuracy.The research can provide an important way to solve the problem of missing data on monitoring wetland.

Key words: time series, NDVI, STARFM, Dongting Lake area, wetland vegetation

中图分类号: