摘要
【目的】研究棉花冠层叶片叶绿素含量与高光谱参数的相关性,建立叶绿素含量估算模型。【方法】2014年,以鲁棉研28号为研究对象,测定不同施氮水平和生育期棉花冠层叶片叶绿素含量及350~2 500nm光谱反射率,以棉花冠层高光谱反射率与冠层叶片叶绿素含量为数据源,在分析叶绿素含量与原始高光谱反射率（R）、一阶导数光谱反射率（DR）、光谱提取变量和植被指数相关性的基础上,采用一元线性与多元逐步回归的方法构建了叶绿素含量估算模型,并对从中筛选的6种棉花冠层叶片叶绿素含量估算模型进行精度对比。【结果】1）棉花冠层叶片叶绿素含量在反射光谱766nm处相关系数达到最大值,相关系数r=0.836;对于一阶导数光谱,叶绿素含量的敏感波段发生在753nm处,r=0.878;2）以9种光谱提取变量与8种植被指数为自变量,建立叶绿素含量的估算模型,筛选出的特征变量为红边面积（SDr）、绿峰与红谷的归一化值（（Rg-Rr）/（Rg＋Rr））、绿峰幅值（Rg）,仅采用8种常用植被指数建立估算模型,筛选出的变量为比值植被指数（RVI）;3）所建立的6种模型中以基于一阶导数光谱反射率建立的多元逐步回归估算模型精度最高,均方根误差（RMSE）为1.075,相对误差（RE）为2.22%,相关系数（r）为0.952。【结论】采用原始光谱、一阶导数光谱、光谱提取变量及植被指数均可对棉花叶绿素含量进行监测,其中基于一阶导数光谱的多元逐步回归模型对叶绿素含量的估算效果最优。
【Objective】This study investigated the relationship between chlorophyll content and high spectral parameters of cotton canopy leaves and established a model for estimating chlorophyll content.【Method】In 2014,Lumianyan 28 was selected to measure chlorophyll content and canopy hyperspectral reflectance in band of 350-2 500 nm of cotton at six different growth stages and nitrogen levels in a field experiment.The correlations between single narrow band raw reflectance（R）,the first derivative spectral reflectance（DR）,commonly used spectral variables,vegetation index,and chlorophyll content were analyzed and determined.Basing on the results of correlation analysis,the estimation models of chlorophyll content were established using linear regression and multiply stepwise regression methods,and then the precise of six models was analyzed.【Result】1）The maximum correlation coefficient of chlorophyll content occurred at the reflectance band of 766 nm with r=0.836 and the highest correlation coefficient between the first de-rivative spectral data and chlorophyll content occurred at band of 753 nm with r=0.878.2）The variables screened were SDr,（Rg-Rr）/（Rg＋Rr）,and Rgfor the regression models of 9spectral variables and 8vegetation indexes.RVI was also screened by the regression equation of 8vegetation indexes.3）The model based on the first derivative spectral reflectance using multiply stepwise regression method obtained the most satisfied results for the estimation of chlorophyll content with RMSE=1.075,RE=2.22%,and r=0.952.【Conclusion】It is feasible to monitor the cotton growth by the raw spectral,the first derivative spectral reflectance,the extract spectral variables,and the vegetation index.The model based on the first derivative spectral reflectance was optimal to monitor the chlorophyll content of cotton canopy leaf.
作者
楚万林
齐雁冰
常庆瑞
蒲洁
CHU Wanlin QI Yanbing CHANG Qingrui PU Jie(College of Natural Resources and Environment, Northwest A ＆F University , Yangling , Shaanxi 712100, China)
出处
《西北农林科技大学学报（自然科学版）》
CSCD
北大核心
2016年第9期65-73,共9页
Journal of Northwest A&F University(Natural Science Edition)
基金
"十二五"农村领域国家科技计划课题(2013AA102401-2)
关键词
棉花
叶绿素含量
高光谱
植被指数
cotton
chlorophyll content
hyperspectrum
vegetation index
作者简介
楚万林（1987-），男，河南淮阳人，在读硕士，主要从事地图学与地理信息系统研究。E-mail：chuwanlin0512@126．com
[通信作者]齐雁冰（1976-），男，河南淮阳人，副教授，主要从事3S技术及其应用研究。E-mail：ybqi@nwsuaf．edu．cn
