基于高光谱遥感的小麦黄花叶病害等级监测研究  

Monitoring Yellow Mosaic Leaf Disease Grade of Winter Wheat Based on Hyperspectral Remote Sensing

作　　者：刘梦冉 张海艳 齐双丽 孙炳剑[4] 段剑钊 贺利 郭天财[1] 冯伟[1] LIU Mengran;ZHANG Haiyan;QI Shuangli;SUN Bingjian;DUAN Jianzhao;HE Li;GUO Tiancai;FENG Wei(Agronomy College,Henan Agriculture University,Zhengzhou,Henan 450046,China;College of Mechanical&Engineering,Henan Agriculture University,Zhengzhou,Henan 450002,China;Luohe Academy of Agricultural Science,Luohe.Henan 462000,China;Plant Protection College,Henan Agriculture University,Zhengzhou,Henan 450002,China)

机构地区：[1]河南农业大学农学院,河南郑州450046 [2]河南农业大学机电工程学院,河南郑州450002 [3]漯河市农科院,河南漯河462000 [4]河南农业大学植物保护学院,河南郑州450002

出　　处：《麦类作物学报》2022年第7期872-882,共11页Journal of Triticeae Crops

基　　金：中国博士后科学基金资助项目(2021M701109);河南省科技攻关项目(192102110118)。

摘　　要：为了构建小麦黄花叶病的遥感监测技术,在小麦返青期、拔节前期和拔节后期测定了不同黄花叶病等级下的冠层反射率,并同步调查与病害等级相关的小麦株高、含水量、氮含量、色素含量等农学参数,筛选出适宜监测小麦黄花叶病的植被指数,并构建病害等级监测模型。结果表明,小麦黄花叶病的反射光谱敏感波段在返青期和拔节前期集中于560~720nm范围,而拔节后期则集中于800~900nm区域。随病害等级的增加,光谱反射率在可见光波段逐渐增加,而在近红外波段区域降低。植被指数与病害等级相关性在不同生育时期间存在显著差异,整体上以拔节前期最好,决定系数(r^(2))为0.72~0.82,而拔节后期模型精度急剧下降(r^(2)=0.26~0.72)。在植被指数中,整体上以表征色素变化的mND705模型预测精度最好,r^(2)和RMSE分别为0.59~0.68和0.79~0.98。采用偏最小二乘回归(PLSR)建立黄花叶病害分级模型,三个时期的模型精度均高于植被指数模型,且整体上以返青期和拔节期前期估算效果较好,模型验证r^(2)为0.93~0.97,RMSE为0.24~0.32。因此,利用PLSR模型可以准确评价返青至拔节期前期小麦黄花叶病害等级。In order to construct the remote sensing monitoring technology for wheat yellow mosaic disease,the canopy reflectance under different yellow mosaic disease grades was measured at the re-greening stage,early jointing stage and late jointing stage of wheat.The agronomic parameters were investigated:plant height,water content,nitrogen content and pigment content related to the disease grade,to screen the vegetation index suitable for monitoring wheat yellow mosaic disease and to construct the disease grade monitoring model.The results showed that the sensitive band of reflection spectrum of wheat yellow mosaic disease was concentrated in the band of 560-720 nm at the re-greening stage and early jointing stage,while it was concentrated in the band of 800-900 nm at the late jointing stage.With the increase of disease grade,the spectral reflectance in visible band increased gradually,but that in near-infrared band decreased.There are significant differences between vegetation index and disease grade at different growth stages.On the whole,the early jointing stage was the best,and the determination coefficient(r^(2))was 0.72-0.82,while the accuracy of the model decreased sharply at the late jointing stage(r^(2)=0.26-0.72).For the vegetation index,the prediction accuracy of mND705 model,representing the change of pigment,was the best(r^(2)=0.59-0.68 and RMSE=0.79-0.98).The classification model of yellow mosaic disease was established by partial least squares regression(PLSR).The accuracy of the model at the three stages was higher than that of the vegetation index model.On the whole,the estimation effect in the early stage of re-greening stage and jointing stage was good(r^(2)=0.93-0.97 and RMSE=0.24-0.32).Therefore,PLSR model can be used to accurately evaluate the grade of wheat yellow mosaic disease from re-greening to early jointing stage.

关 键 词：小麦 黄花叶病 病害等级 植被指数 高光谱监测 

分 类 号：S512.1[农业科学—作物学] S314