基于棉花黄萎病多“症状”特征的植被指数构建及病情遥感监测研究  

Study of Cotton Verticillium Wilt:Construction of a Vegetation Index Based on Multiple "Symptoms" Characteristics and Remote Sensing Monitoring

作　　者：王姣 李志沛 张立福 黄长平 WANG Jiao;LI Zhi-pei;ZHANG Li-fu;HUANG Chang-ping(Institute of Remote Sensing and Digital Earth,Chinese Academy of Sciences,Beijing 100101;University of Chinese Academy of Sciences,Beijing 100049;HyTech,(Beijing/Tianjin) Co.,Ltd,Tianjin 300384,China)

机构地区：[1]中国科学院遥感与数字地球研究所,北京100101 [2]中国科学院大学,北京100049 [3]中科遥感科技集团有限公司,天津300384

出　　处：《地理与地理信息科学》2019年第5期46-51,共6页Geography and Geo-Information Science

基　　金：兵团重大科技课题(2018AA00402);国家自然科学基金重点项目(418301108);中国科学院青年创新促进会项目(2017086);新疆生产建设兵团绿洲生态农业重点实验室开放课题(201801)

摘　　要：综合考虑棉花黄萎病多“症状”特征对黄萎病遥感精准监测及其抗性鉴定和防治工作具有重要意义。该文结合黄萎病胁迫下棉花冠层光谱响应的生理机制,基于Relief-F算法优选出对棉花黄萎病不同“症状”变化敏感的特征谱段(531nm、699nm、701nm、1404nm),构建了一种新的棉花黄萎病病情指数(Cotton Verticillium Wilt Index,CVWI),并建立了基于支持向量机(SVM)的黄萎病遥感监测模型。研究表明:与传统病害植被指数相比,CVWI综合考虑了黄萎病导致的棉花水分、叶绿素、叶黄素、红边等理化与生理参数变化,可更好指示黄萎病病情;基于CVWI的黄萎病监测模型精度高于传统表现最好的色素比值指数(Pigment Specific Simple Ratiochl-b,PSSRb),模型的精确率、召回率与F1值分别提高了19%、6%、13%。研究结果可为棉花黄萎病大面积遥感精准监测提供新的思路与方法。Full consideration of multiple "symptoms" characteristics (including chlorophyll,water content,photosynthesis,structure,et al.) is critical for the accurate diagnosis and monitoring of cotton Verticillium wilt (CVW).Those factors should not be used independently in actual research.However,most of the current disease vegetation indices are related to only one or few of "symptoms",which limited the detecting sensitivity to CVW.In this study,we aim to develop a new spectral index(cotton Verticillium wilt index,CVWI) for better sensitivity,which was derived from a liner combination of two single bands (701 nm,1 404 nm) and a normalized wavelength difference (531 nm,699 nm) according to the form of NDVI.The set of spectral features were extracted by using the Relief-F algorithm.The CVWI has comprehensively taken into account the changes of cotton moisture,chlorophyll,lutein and red-edge bands caused by CVW,and resulted in a higher relevance in the linear statistical regression ( R^2=0.904).Finally,the CVWI and PSSRb (pigment specific simple ratio chl-b) were used to construct the classification model based on support vector machine (SVM),and the CVWI got a better classification accuracy (accuracy rate,recall rate and F1 score increased by 19%,6%,13%,respectively).The results have important theoretical and practical significance for precise monitoring of the CVW in large areas using remote sensing technology.Future work on the effects of different growth periods and different stresses can further verify and improve the proposed CVWI.

关 键 词：棉花黄萎病 高光谱 病害植被指数 Relief-F算法 支持向量机 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置] S435.621[自动化与计算机技术—控制科学与工程]