植物生态学报 ›› 2018, Vol. 42 ›› Issue (6): 640-652.DOI: 10.17521/cjpe.2017.0240

• 研究论文 • 上一篇    下一篇

基于连续统去除法的水生植物提取及其时空变化分析——以官厅水库库区为例

汪星,宫兆宁(),井然,张磊,金点点   

  1. 首都师范大学资源环境与旅游学院, 北京 100048; 三维信息获取与应用教育部重点实验室, 北京 100048; 资源环境与地理信息系统北京市重点实验室, 北京 100048; 北京市城市环境过程与数字模拟国家重点实验室培育基地, 北京 100048
  • 收稿日期:2017-09-13 修回日期:2018-04-04 出版日期:2018-06-20 发布日期:2018-06-20
  • 通讯作者: 宫兆宁
  • 基金资助:
    国家国际科技合作专项资助项目(2014DFA21620)

Extraction of aquatic plants based on continuous removal method and analysis of its temporal and spatial changes—A case study of Guanting Reservoir

WANG Xing,GONG Zhao-Ning(),JING Ran,ZHANG Lei,JIN Dian-Dian   

  1. College of Resource Environment & Tourism, Capital Normal University, Beijing 100048, China; Key Laboratory of 3D Information Acquisition and Application,Ministry of Education, Beijing 100048, China; Key Laboratory of Resources Environment and GIS of Beijing Municipal, Beijing 100048, China; Base of the State Laboratory of Urban Environmental Processes and Digital Modeling, Beijing 100048, China
  • Received:2017-09-13 Revised:2018-04-04 Online:2018-06-20 Published:2018-06-20
  • Contact: Zhao-Ning GONG
  • Supported by:
    Supported by the International Science & Technology Cooperation Program of China(2014DFA21620)

摘要:

光谱特征变量的筛选作为水生植物识别的重要手段之一, 在水生植物种类识别研究中应用广泛。该研究将实测光谱特征提取与多时相Landsat 8 OLI影像数据分析相结合, 找到一种有效识别不同种类水生植物的特征变量。在水生植物反射光谱特征分析中引入矿质分析中普遍使用的连续统去除法, 对光谱重采样结果作连续统去除处理后提取光谱吸收深度特征。采用单因素方差分析法对比7个光谱重采样波段和3个连续统去除吸收深度敏感波段, 发现经连续统去除处理的短波红外1波段(SWIR1CR)对于不同类型的水生植物区分效果最佳。将连续统去除法应用到遥感影像处理上, 发现SWIR1CR波段能较好区分沉水植物和挺水植物; 结合影像归一化植被指数和SWIR1CR波段可较好区分三类水生植物。结合特征波段筛选结果采用支持向量机分类方法, 得到水生植物的分类结果精度为86.33%, 对比全生长期12期影像提取的水生植物分布图, 发现水生植物主要分布于官厅水库库区南北岸浅水区, 水生植物面积最大时约占库区总面积的35.13%; 其中沉水植物年内生长分布变化幅度较大, 6月上旬开始迅速生长; 10月份水生植物开始衰减; 11月份水生植物占库区面积的20%, 沉水、浮水植物大幅衰减消失。

关键词: 连续统去除, 光谱吸收深度, 单因素方差分析, 时空变化, 短波红外1波段

Abstract:

Aims Screening of spectral characteristic variables is one of the important means for aquatic plant recognition, and it is widely applied in aquatic plant species identification. In this paper, a method for identifying aquatic plants species was constructed by combining extracted spectral feature information with the multi-temporal Landsat 8 OLI image data analysis.

Methods In analyzing reflectance spectra of aquatic plants, the method of continuum removal for mineral analysis was introduced. The spectral resampling was performed on the measured spectral curve, and the spectral absorption depth was characterized by the continuous removal of the spectral resampling results. One-way ANOVA method was used to compare the seven spectral resampling bands and the three continuum removal absorption depth sensitive bands. Then the characteristic bands with significant differentiation of different aquatic plants were selected. The continuum removal was applied on remote sensing image processing. The results of the spectroscopic analysis were used to guide the identification of aquatic plants in using Landsat 8 OLI. The classification of aquatic plants was carried out by using support vector machine (SVM) classification.

Important findings The results of the measured spectrum resampling are similar to the atmospheric calibration of Landsat 8 OLI in the same position, and the results of the measured spectral curves can be used to guide the classification of Landsat 8 OLI. The one-way ANOVA method was used to compare seven spectral resampling bands and three continuous systems in absorbing sensitive wavelengths. The results showed that the short wave infrared 1 band, which was processed by continuum removal (SWIR1CR), was the best in distinguishing different types of aquatic plants. In this paper, the continuum removal was applied on remote sensing image processing, and it was found that the SWIR1CR band can better distinguish the submerged plants and the emergent plants. The normalized differential vegetation index and SWIR1CR band were well capable of identifying submerged plants, floating plants and emergent plants. Based on the SVM classification method, the classification accuracies of aquatic plants were 86.33%. The distribution of aquatic plants showed that the aquatic plants were mainly distributed in shallow water areas of the south north bank of Guanting reservoir. When the aquatic plant distribution area reached the peak, it accounted for about 35.13% of the total area of the reservoir. The growth distribution of submerged plants changed significantly during a year. The stem and leaves of submerged plants began to emerge in early June. Aquatic plants began to wither in October, and aquatic plants accounted for only 20% of the total area in November.

Key words: continuous removal, spectral absorption depth, one-way ANOVA, spatiotemporal variation, the short wave infrared I band