棉花黄萎病病情严重度的连续统去除估测法

Continuum removal method for cotton verticillium wilt severity monitoring with hyperspectral data

该文利用地面高光谱数据遥感监测棉花黄萎病病情严重度。通过分析黄萎病不同病情严重度下棉花单叶光谱变化特征,筛选对病情严重度识别的敏感波段,利用连续统去除法对光谱反射率进行处理并构建基于光谱特征吸收参量的病情严重度估测模型。红光光谱(650～700nm)是识别棉花单叶黄萎病病情严重度的最佳波段,且随受害棉叶病情严重度增加,红光光谱吸收波段位置向长波方向移动,吸收波段深度及特征吸收峰面积减小,二者与病情严重度呈极显著线性相关关系。以红光波段特征吸收峰右半端面积为自变量建立的线性模型具有最大决定系数,最小均方根误差,为棉花单叶黄萎病病情严重度估测的最佳模型。研究结果表明,利用高光谱遥感技术能够有效地监测棉花黄萎病病情严重度,为进一步深入研究棉花黄萎病的遥感监测机理提供理论依据。

The objective of this study was to estimate verticllium wilt severity levels of cotton leaf using ground hyperspectral data.According to the analysis of spectrum features of cotton leaf at different severity levels,the bands sensitive to verticllium wilt severity level were selected and the spectrum reflect was processed and the inversion model of severity level based on absorption parameters of spectrum features was established using a method of continuum removal.Results showed that the band ranging from 650 nm to 700 nm was the most sensitive region for distinguishing cotton leaf verticllium wilt severity.With increasing of cotton verticillium wilt severity levels,the red depth and absorption area decreased,and absorption position of the red spectrum moved to the long wavelength band.Especially,the severity level correlated very significantly with the band depth and absorption area of red spectrum.Consequently,the linear inversion model based on the right half area of the red absorption peak had the maximum determination coefficient and minimum root mean square error,which was taken in the model as the independent variable.The model was considered as the optimum one to estimate the severity levels of cotton verticillium wilt.Study confirmed that using hyperspectral remote sensing data to explore cotton disease information was an important and effective application,and it provides a good theoretical basis for further studying monitoring mechanism of cotton verticillium wilt.