基于高光谱数据的病害胁迫下杉木冠层色素含量估算研究

Study on Estimating Pigment Contents in Canopy of Chinese Fir under Disease Stress Based on Hyperspectral Data

分析炭疽病胁迫下杉木冠层的高光谱特征,探索建立病害胁迫下杉木冠层色素含量的高光谱估算模型,从而促进高光谱遥感技术在森林病虫害监测中的应用。于2011年4—6月,在湖南省攸县实地调查杉木炭疽病,并测定不同程度炭疽病胁迫下杉木冠层的光谱反射率及其色素含量。将冠层光谱、一阶微分及相应的色素含量数据分别进行相关分析,采用回归分析方法,选取部分样本建立色素含量估算模型,并利用其余的样本对模型进行精度检验。结果表明,可见光和近红外区域是病害杉木色素反射和吸收的敏感区域,色素含量与一阶微分光谱在红边(695～754nm)内相关性最高,单波段一阶微分光谱741nm处的相关系数最大。其中,以差值植被指数DVI[FD587,FD741]为变量的幂函数模型估算Chla+b、Chla和Chlb含量的精度最高,相对误差均小于15%,均方根误差在0.093～0.241之间。研究表明,受不同程度炭疽病胁迫下杉木冠层的光谱反射差异较大,可利用高光谱信息定量估算病害胁迫下杉木冠层的色素含量,且估算精度较高。

This study analyzed the hyper-spectral characteristics of the canopy of Chinese fir under the anthracnose stress, and explored and built the hyper-spectral estimation models of the pigment content in canopy of Chinese fir under the disease stress, the results will promote the application of the hyper-spectral remote sensing-technology in the forest pest and disease monitoring. From April to June in 2011, the anthracnose of Chinese fir had been investigated at Youxian in Hunan Province, and the pigment content and the spectral reflectance of Chinese fir canopy which was under the different degrees of anthracnose stress had been measured. The correlation between the pigment contents, the first derivative data and the canopy hyper-spectral data of the relatively disease Chinese fir leaves were analyzed respectively. By regression analysis methods, a part of the samples were used for construction of the estimation model about pigment contents of disease Chinese fir, and the rest of the samples were used to test the model accuracy. The result showed that the visible light and near-infrared region were the sensitive region which were reflected and absorbed by the pigment in disease of Chinese fir; pigment content had the highest correlation with the first order differential spectra in red edge （695-754 nm）, and the correlation coefficient of single band first order differential spectrum in 741 nm was the largest. And it also showed that the accuracy of Chla＋b, Chla and Chlb contents was the highest estimated by power function model which used the difference vegetation index DVI [FD587,FD741] as the variable, the relative errors were less than 15%, the RMS error was in the range of 0.093 to 0.241. This study showed that significant difference of the spectral reflectance under the different degrees disease stress of Chinese fir anthracnose, and the hyper-speetral information could be used for estimating pigment contents in canopy of Chinese fir under the disease stress and the estimation precision was high.