引用本文:李福根,张保辉※,段玉林.利用植被指数非相似性监测水稻病虫害方法研究[J].中国农业信息,2020,32(1):46-63
【打印本页】   【HTML】   【下载PDF全文】   查看/发表评论  【EndNote】   【RefMan】   【BibTex】
←前一篇|后一篇→ 过刊浏览    高级检索
本文已被:浏览 41次   下载 27 本文二维码信息
码上扫一扫!
分享到: 微信 更多
利用植被指数非相似性监测水稻病虫害方法研究
李福根1,2,张保辉※1,段玉林1
1.中国农业科学院农业资源与农业区划研究所 / 农业部农业遥感重点实验室,北京 100081;2.中国科学院空天信息创新研究院 遥感科学国家重点实验室,北京 100101
摘要:
【目的】提出植被指数非相似性的计算方法以及利用植被指数非相似性监测水稻病虫 害。【方法】首先将具有空间特性的植被指数影像假设成为具有概率统计特性的信息量,并 基于信息理论和 SID 模型推导出两个不同区域相同大小影像的植被指数非相似性(VID)计 算方法;然后根据 VID 计算方法,计算出无水稻病虫害的参考区域与有水稻病虫害的试验 区域的 10 种植被指数 VID;之后利用计算出的 10 种植被指数 VID 与实测水稻病虫害等级 数据进行回归分析,判断 VID 与水稻病虫害等级数据的相关性;最后选择相关程度较高的 几种植被指数 VID 进行 K-fold 交叉验证,判断植被指数非相似性监测水稻病虫害的精度。 【结果】10 种植被指数 VID 与地面实测水稻病虫害等级数据进行回归分析后,R 2 的范围 在 0.63~0.95 之间。3 种相关程度较高的植被指数 VID 与地面实测水稻病虫害等级数据进 行交叉验证后,R 2 的范围在 0.91~0.97 之间,RMSE 在 0.16~0.24 之间,广义监测精度在 97.62%~100% 之间。【结论】植被指数非相似性与水稻病虫害等级数据具有较强的相关性, 利用 3 种相关程度较高的植被指数 VID 监测水稻病虫害等级具有很高的精度。
关键词:  遥感;Planet 卫星数据;植被指数非相似性;水稻病虫害;回归分析;K-fold 交叉 验证
DOI:10.12105/j.issn.1672-0423.20200106
分类号:
基金项目:中国农业科学院基本科研专项“天空地大数据驱动的水稻病虫害智能诊断预警系统”(Y2019XK24-02)
Detection and discrimination of pests and diseasesin rice using vegetation index divergence
Li Fugen1,2, Zhang Baohui ※1, Duan Yulin1
1.Institute of Agricultural Resources and Regional Planning,Chinese Academy of Agricultural Sciences/Key Laboratory of Agricultural Remote Sensing,Ministry of Agriculture and Rural Affairs,Beijing 100081,China;2.State Key Laboratory of Remote Sensing Science,Aerospace Information Research Institute,Chinese Academy of Sciences,Beijing 100101,China
Abstract:
[Purpose]The paper aims to propose a method to calculate the vegetation index divergence(VID);and to detect and discriminate pests and diseases affections in rice using such a VID.[Method]Firstly,the VID is calculated by analyzing satellite images’ spatial characteristics based on the information-theory and Spectral Information Divergence model. Then the method calculated the VID between the same size images in two different regions. Then,according to the method of VID calculation,the VID of 10 vegetation indexes between the reference area without rice diseases and test area with pests and diseases in rice are calculated. Subsequently,the VID of 10 vegetation indexes against the measured data of pests and diseases in rice are used for a regression analysis to explore the correlation between the VID and field measured data. Finally,the VID of several vegetation indexes in high correlation with field measured data were selected for a K-fold cross validation to validate the accuracy of detection and discrimination of pests and diseases in rice using VID.[Result]The results of regression analysis between the VID of 10 vegetation indexes and the field measured data of pests and diseases in rice produced the R2 ranging from 0.63 to 0.95,and the results of K-fold cross validation between the VID of three vegetation indexes with goodness correlation and field measured data produced the R2 ranging from 0.91 to 0.97,with the RMSE ranging from 0.16 to 0.24 and the general detection accuracy ranging from 97.62% to 100%.[Conclusion]There is a strong correlation between the VID and the field measured data of pests and diseases in rice. Using the VID of three vegetation indexes with goodness correlation to detect and discriminate pests and diseases affections in rice will be in highest accuracy.
Key words:  remote sensing;PlanetScope;vegetation index divergence;pests and diseases in rice;regression analysis;K-fold cross validation