基于无人机多光谱遥感的干旱胁迫下玉米冠层SPAD值监测  

Monitoring of maize canopy SPAD value under drought stress based on UAV multi-spectral remote sensing

作　　者：李诗瑶 丛士翔 王融融 余海龙[1] 黄菊莹[3] LI Shiyao;CONG Shixiang;WANG Rongrong;YU Hailong;HUANG Juying(College of Geography and Planning,Ningxia University,Yinchuan 750021,Ningxia,China;College of Mechanical and Electronic Engineering,Northwest A&F University,Yangling 712100,Shaanxi,China;College of Ecology and Environment,Ningxia University,Yinchuan 750021,Ningxia,China)

机构地区：[1]宁夏大学地理科学与规划学院,宁夏银川750021 [2]西北农林科技大学机械与电子工程学院,陕西杨凌712100 [3]宁夏大学生态环境学院,宁夏银川750021

出　　处：《干旱区地理》2023年第7期1121-1132,共12页Arid Land Geography

基　　金：宁夏重点研发项目(2019BEG03029)资助。

摘　　要：精准动态监测干旱胁迫下的玉米冠层叶绿素相对含量对提高我国玉米旱灾预警水平、实现田间精准灌溉具有重要的指示作用。以无人机多光谱影像为数据源,选用多种具有明确物理意义且与玉米冠层叶绿素相对含量SPAD(Soil and plant analyzer development)值相关性强的植被指数,利用多元逐步回归、支持向量机、BP神经网络(Back propagation neural network,BPNN)建立玉米冠层SPAD值的遥感监测模型并验证,选取最优估测模型提取各生育期不同程度干旱胁迫下的玉米冠层SPAD值,分析不同生育期玉米冠层SPAD值变化,探究不同干旱胁迫程度对玉米冠层SPAD值的影响。结果表明:不同生育期的玉米冠层叶绿素敏感的植被指数不同,且各生育期估测能力最优的反演模型不同。对比3种建模方法,BPNN模型的建模结果及验证结果最优,说明其估测能力及稳定性表现最好,可以作为基于无人机多光谱的玉米冠层SPAD值建模的优选方法。此外,干旱胁迫会降低玉米冠层SPAD值遥感监测模型的估测精度,对苗期的影响最为显著。轻度干旱对玉米冠层SPAD值影响不显著,可见玉米对干旱胁迫具有一定的适应性和抗逆性。因此,基于植被指数的BPNN模型可以更好地估算SPAD值,为基于无人机遥感的SPAD值监测提供一种新途径,为干旱胁迫下夏玉米冠层SPAD值的无损监测以及田间水分精准管理提供参考。The accurate and dynamic monitoring of the relative content of chlorophyll in corn canopy under drought stress can improve the early warning level of corn drought and help realize precise irrigation.In this study,multispectral images captured by unmanned aerial vehicles(UAVs)were used as data sources,and various vegetation indices with clear physical significance and strong correlation with soil and plant analyzer development(SPAD)values of the chlorophyll relative content in corn canopy were selected.Multiple stepwise regression,support vector machine,the back propagation neural network(BPNN),and a remote sensing monitoring model of the corn canopy SPAD value were established and verified.The optimal estimation model was selected to extract corn canopy SPAD values under various degrees of drought stress in different growth periods.Furthermore,the changes in the corn canopy SPAD value in different growth periods were analyzed to investigate the effects of various degrees of drought stress on the corn canopy SPAD value.The results revealed that the chlorophyll-sensitive vegetation index of corn canopy was different at different growth stages.Furthermore,the inversion models with the best estimation ability varied at different growth stages.Comparing the three modeling methods revealed that the modeling results and verification results of the BPNN model were the best,which indicated that the BPNN model exhibited the best estimation ability and stability performance and can be used as the optimal method for the modeling of the SPAD value of corn canopy based on UAV multispectrum.Furthermore,drought stress can reduce the estimation accuracy of the SPAD value of corn canopy by using a remote sensing monitoring model.This reduction in accuracy affects the seedling stage.Mild drought did not significantly affect the SPAD value of corn canopy,which indicated that corn exhibited a certain adaptability and resistance to drought stress.Therefore,the BPNN model based on the vegetation index can estimate the SPAD value and ca

关 键 词：玉米 无人机 多光谱遥感 叶绿素相对含量 

分 类 号：U445.4[建筑科学—桥梁与隧道工程]