刚竹毒蛾胁迫下毛竹叶绿素与叶光谱特征的关系及其变化  

Relationship Between Chlorophyll and Leaf Spectral Characteristics and Their Changes Under the Stress of Phyllostachys Praecox

作　　者：胡新宇 许章华[1,2,3,5,6] 黄旭影 张艺伟 陈秋霞 王琳 刘辉[4] 刘智才 HU Xin-yu;XU Zhang-hua;HUANG Xu-ying;ZHANG Yi-wei;CHEN Qiu-xia;WANG Lin;LIU Hui;LIU Zhi-cai(College of Environment and Safety Engineering,Fuzhou University,Fuzhou 350108,China;Fujian Provincial Key Laboratory of Resource and Environment Monitoring&Sustainable Mangement and Utilization,Sanming University,Sanming 365004,China;Key Laboratory of Spatial Data Mining and Information Sharing,Ministry of Education,Fuzhou 350108,China;Fuzhou Zhonggu Haichuang Science and Technology Development Co.,Ltd.,Fuzhou 350108,China;School of Advanced Manufacturing,Fuzhou University,Quanzhou 362200,China;Postdoctoral Research Station of Information and Communication Engineering,Fuzhou University,Fuzhou 350108,China;School of Public Administration,Fujian Agriculture and Forestry University,Fuzhou 350002,China;International Institute for Earth System Science,Nanjing University,Nanjing 210023,China)

机构地区：[1]福州大学环境与安全工程学院,福建福州350108 [2]福建省资源环境监测与可持续经营利用重点实验室,福建三明365004 [3]空间数据挖掘与信息共享教育部重点实验室,福建福州350108 [4]福州中谷海创科技发展有限公司,福建福州350108 [5]福州大学先进制造学院,福建泉州362200 [6]福州大学信息与通信工程博士后科研流动站,福建福州350108 [7]福建农林大学公共管理学院,福建福州350002 [8]南京大学国际地球系统科学研究所,江苏南京210023

出　　处：《光谱学与光谱分析》2022年第9期2726-2739,共14页Spectroscopy and Spectral Analysis

基　　金：国家自然科学基金项目(42071300,41501361);福建省自然科学基金面上项目(2020J01504);福建省自然科学基金面上项目(2016J01188)资助;中国博士后面上基金一等项目(2018M630728);福建省资源环境监测与可持续经营利用重点实验室开放基金项目(ZD202102);3S技术与资源优化利用福建省高校重点实验室开放课题(fafugeo201901);晋江市福大科教园区发展中心科研项目(2019-JJFDKY-17);中国科学院战略性先导科技专项(XDA23100202)。

摘　　要：叶绿素是反映绿色植被健康状态的重要生理参数,虫害胁迫下叶绿素与叶光谱的变化机制较为复杂,深入剖析二者关系对于虫害检测有重要意义。以福建省南平市顺昌县为试验区,测定不同受害情景下毛竹叶叶绿素含量(SPAD)与叶光谱,采用Pearson相关法筛选叶光谱特征指标,建立叶SPAD的多元线性回归、岭回归、随机森林与XGBoost估测模型。通过比较光谱特征指标筛选结果及模型估测效果,分析刚竹毒蛾胁迫下毛竹叶绿素与叶光谱特征的关系及其变化。结果表明:(1)随着虫害程度上升,毛竹叶SPAD呈下降趋势;(2)较之于未受害状态,刚竹毒蛾胁迫下毛竹叶光谱特征发生明显变化,“绿峰”和“红谷”趋于消失,“红边”斜率减小,近红外波长反射率降低;(3)基于全样本拟合叶SPAD的最优光谱特征指标为VOG_(2),R_(515)/R_(570),CI_(red),PRI与NDVI_(705),最佳估测模型为多元线性回归模型(R^(2)=0.7537,RMSE=3.0150);(4)基于不同受害程度样本拟合毛竹叶SPAD,最优光谱特征指标分别为健康:CI_(red),VOG_(2),ARVI,R_(515)/R_(570),DVI;轻度:RENDVI,RERVI,REDVI;中度:RENDVI,RERVI,REDVI;重度:VOG_(2),CI_(red),NDVI_(705),PRI;小年:PRI,NDVI_(705),VOG_(1),CI_(red)。最佳估测模型为多元线性回归模型,模型精度分别为健康(R^(2)=0.8823;RMSE=1.6388);轻度(R^(2)=0.1802;RMSE=3.3354);中度(R^(2)=0.3604;RMSE=3.8867);重度(R^(2)=0.4677;RMSE=2.6018);小年(R^(2)=0.7324;RMSE=2.3754)。由此发现,随着虫害等级上升,毛竹叶光谱特征指标也随之改变,关系模型估测精度呈现先急剧下降后缓慢抬升的态势,模型对健康与小年叶SPAD估测效果较好,对轻—中—重度危害叶SPAD估测效果较差;当毛竹叶SPAD与叶光谱特征的关系趋向紊乱时,预示可能有刚竹毒蛾危害发生。Chlorophyll is an important physiological parameter reflecting the health status of green vegetation.The change mechanism of chlorophyll and leaf spectrum under pest stress is complex.It is of great significance to analyze the relationship between chlorophyll and leaf spectrum in depth for pest detection.Taking Shunchang County,Nanping City,Fujian Province as the experimental area,the leaf SPAD and leaf spectrum of Phyllostachys pubescens under different damage scenarios were measured.Pearson correlation method was used to screen the leaf spectrum characteristic indexes,and multiple linear regression,ridge regression,random forest and XGBoost estimation models of leaf SPAD were established.By comparing the screening results of spectral characteristics and the estimation effect of the model,the relationship between chlorophyll and leaf spectral characteristics of Phyllostachys pubescens under the stress of Pantana phyllostachysae was analyzed.The results showed that:(1)SPAD of Phyllostachys pubescens leaves showed a downward trend with the increase of insect pests;(2)Compared with the undamaged state,the spectral characteristics of Phyllostachys pubescens leaves changed obviously under the stress of Pantana phyllostachysae,and the“green peak”and“red valley”tended to disappear,the slope of“red edge”decreased,and the reflectance of near infrared wavelength decreased.(3)The best spectral characteristics of leaf SPAD based on full sample fitting are VOG_(2),R_(515)/R_(570),CI_(red),PRI and NDVI_(705),and the best estimation model is multiple linear regression model(R^(2)=0.7537,RMSE=3.0150).(4)SPAD of Phyllostachys pubescens leaves was fitted based on samples with different damage degrees.The optimal spectral characteristic indexes were health:CI_(red),VOG_(2),ARVI,R_(515)/R_(570),DVI;mild hazard:RENDVI,RERVI and REDVI;moderate hazard:RENDVI,RERVI and REDVI;severe hazard:VOG_(2),CI_(red),NDVI_(705);off year:PRI,NDVI_(705),VOG_(1),CI_(red).The best estimation model is the multiple linear regression model,an

关 键 词：虫害胁迫 叶绿素含量 叶光谱特征 相关分析 机器学习 

分 类 号：O433.4[机械工程—光学工程]