欢迎访问《中国农学通报》，
中国农学通报

中国农学通报 ›› 2013, Vol. 29 ›› Issue (1): 29-36.doi: 10.11924/j.issn.1000-6850.2012-2751

• 林学 园艺 园林 • 上一篇    下一篇

基于WorldView 2影像杉木叶面积指数与植被指数相关性研究

陈利 林辉 孙华 严恩萍 王继志   

  • 收稿日期:2012-08-12 修回日期:2012-09-25 出版日期:2013-01-05 发布日期:2013-01-05
  • 基金资助:
    国家“863”课题研究任务“森林资源信息快速提取技术研究”(2012AA102001-4)；湖南省高等学校科学研究项目“高分辨率遥感影像森林结构参数反演研究”(11C1313)。

Leaf Area Index and Vegetation Index Correlation of Cunninghamia Based on the WorldView 2 Images

  • Received:2012-08-12 Revised:2012-09-25 Online:2013-01-05 Published:2013-01-05

PDF (PC)

1079

摘要： 为了探究基于高分辨率遥感影像杉木叶面积指数与植被指数的相关性，以湖南省攸县黄丰桥国有林场为研究对象，采用地面实验与遥感技术相结合的方法，利用WorldView 2遥感数据提取NDVI、SAVI、SARVI、RVI、MSAVI、ARVI等6种植被指数，通过LAI-2000测量的杉木叶面积指数(LAI)建立相关关系，开展WorldView 2遥感影像在估测杉木叶面积指数中的应用研究，分析植被指数对杉木LAI的影响。对不同植被指数分别进行线性模型、二次曲线模型、指数曲线模型和对数曲线模型的LAI反演。结果表明：除DVI与LAI相关性稍低一点外，其他植被指数与LAI都有很高的相关性，高于中低分辨率遥感影像提取的植被指数与LAI的相关性，土壤调节植被指数(SAVI)与LAI的相关性与土壤影响因子L无关。在线型模型中，RVI与ARVI更适合于杉木LAI建立一元线性回归模型，相关系数R分别为0.931、0.895，判定系数R2分别为0.866、0.800，均达到较好的拟合效果。在非线性模型中，反演模型最好的是二次曲线模型，其次是指数模型，最差的是对数模型。拟合效果较好的是NDVI、SAVI和RVI；拟合效果最差的是DVI；最好的拟合模型，其R2高达0.884。杉木LAI具有较佳拟合效果的非线型模型是NDVI和SAVI的二次曲线模型。

关键词: 鸟击防范, 鸟击防范

Abstract: In order to research the correlation between leaf area index and vegetation index of Cunninghamia based on high-resolution remote sensing image data, regarding Huangfengqiao Forestry Farm of Youxian as the research object, the author combined the ground experiment with remote sensing technology, and extracted NDVI, SAVI, SARVI, RVI, MSAVI, ARVI 6 vegetation index from the remote sensing data of WorldView 2, and used leaf area index of Cunninghamia (LAI) that was measured by the LAI-2000 to establish the relation, research the estimation of leaf area index of Cunninghamia based on the remote sensing imaging of WorldView 2, and then analyzed the effects between vegetation index and Cunninghamia LAI. Different vegetation index were 2 curve models, linear model, exponential curve model and logarithmic curve model of LAI inversion. The results showed that: except for DVI and LAI correlation was slightly lower, other vegetation index and LAI had high correlation, and were higher than the low resolution image extraction of vegetation index, vegetation index (SAVI) and LAI correlation had unrelated to with the soil factor L. In the linear model, RVI and ARVI were more suitable for Cunninghamia LAI to establish a linear regression model, and the correlation coefficient R was 0.931, 0.895, respectively, the determination of coefficient R2 was 0.866, 0.800, respectively, which had a better fitting effect. In the nonlinear model, the inversion model of the best was 2 curve models, followed by an exponential model, and the worst was the logarithmic model. The model fitting result of NDVI, SAVI and RVI were better than others; the model fitting result of DVI was the worst; the best fitting model of the R2 was as high as 0.884. Cunninghamia LAI had better fitting effect of nonlinear model was NDVI and SAVI2 curve model.