基于无人机多光谱遥感的水稻冠层SPAD值反演

Inversion of Rice Canopy SPAD Value Based on UAV Multispectral Remote Sensing

为研究水稻叶片叶绿素相对含量(SPAD)在3种水分处理和5种施氮处理下的变化规律,探讨无人机多光谱遥感技术反演水稻SPAD的可行性,本研究利用大疆精灵4多光谱无人机,采集了水稻拔节孕穗期、抽穗开花期和乳熟期的冠层多光谱遥感影像,并同步测定水稻SPAD值,基于25个光谱变量(5个波段反射率和20个植被指数),采用多元线性逐步回归、岭回归和套索回归3种方法构建了水稻SPAD的反演模型。结果表明:水稻3个生育期的SPAD最佳反演模型均是采用套索回归方法构建的,其中乳熟期建立的SPAD最佳反演模型在3个生育期中的反演精度最高,决定系数为0.782,均方根误差为1.2177,相对误差为6.6113%。因此,该研究可对水稻叶片SPAD进行遥感监测,并为水稻精准灌溉和施肥提供科学依据和数据支撑。

To study the changes of chlorophyll relative content(SPAD)in rice leaves under three water treatments and five nitrogen treat⁃ments,and to explore the feasibility of multispectral remote sensing technology for unmanned aerial vehicle(UAV)to retrieve rice SPAD.In this study,DJI Phantom 4 multispectral UAV was used to collect multispectral remote sensing images of rice canopy at jointing and booting stage,heading and flowering stage and milk ripening stage,and simultaneously measure the SPAD value of rice.Based on twenty-five spec⁃tral variables(five band reflectance and twenty vegetation indexes),the retrieving model of rice SPAD at different stages is established by multiple linear stepwise regression(MLSR),ridge regression(RR)and lasso regression(LR).The results show that the best retrieving mod⁃els of rice SPAD at three growth stages are established by lasso regression,and the best retrieving model of SPAD established at milk ripen⁃ing stage has the highest inversion accuracy among the three growth stages,with coefficient of determination of 0.782,root mean square error of 1.2177 and relative error of 6.6113%.Therefore,this paper can monitor rice leaf SPAD by remote sensing,and provide a scientific basis and data support for rice precision irrigation and fertilization.