摘要
叶类蔬菜的品质和产量与其净光合速率息息相关,秋冬季节受颗粒物污染的影响,温室大棚内叶菜的光合作用受到制约,对生理信息的准确预测不利。以采收期生菜为试验对象,研究其在颗粒物污染生长环境下,基于高光谱技术建立并对比生菜净光合速率的反演模型的最优方法。获得生菜的净光合速率和高光谱数据;选取比值植被指数(RVI)、差值植被指数(DI)、归一化植被指数(NDVI)、可视化气压阻抗指数(VARI)、再归一化植被指数(RDVI)、垂直植被指数(PVI)和植被衰减指数(PSRI)的原始和一阶变换共14个光谱植被指数;用相关矩阵法优选出最优波长位置,而非选择既有的固定波长,计算得到最优植被指数,与光谱位置变量(红边幅值,Dr)及光谱面积变量,即红边面积(SDr)、红边面积与蓝边面积的比值(SDr/SDb)、红边面积与黄边面积的比值(SDr/SDy)共同作为光谱特征变量,建立颗粒物污染环境下生菜净光合速率的反演模型。采用多项式拟合和多元散射校正(MSC)、标准正态变量变换(SNV)、偏最小二乘(PLS)和主成分回归(PCR)光谱建模组合方法进行反演预测结果比较。研究结果表明:最优波长位置的PSRI(515,499)和DI(515,499)与净光合速率的相关性最大,可以反映颗粒物污染下生菜的更多生理信息,采用光谱预处理建模组合方法按精度从高到低排序为:SG+MSC+PCR>SG+SNV+PLS>SG+SNV+PCR>SG+MSC+PLS,其中,SG+MSC+PCR组合建模方法的精度最高,决定系数为R_(c)=0.9011,R_(p)=0.9458,基于最优光谱植被指数建模效果最佳;光谱面积变量(SDr/SDb)的拟合精度最高(R^(2)=0.9365),可以实现生菜净光合速率的可靠预测,是基于光谱位置变量和面积变量建立生菜净光合速率的最优方法。该工作对颗粒物污染环境下利用高光谱技术进行植物的生理信息反演研究具有一定参考价值。
The quality and yield of leafy vegetables are closely related to the net photosynthetic rate.Affected by particulate matter pollutionespecially in autumn and winter,the photosynthesis of leafy vegetables in greenhouses is restricted,which is unfavorable for accurate prediction of physiological information.Under the growth environment of particle pollution,The lettuce during the harvest period was the test object.The optimal method for establishing the inversion model of lettuce net photosynthetic rate based on hyperspectral technology was studied.The net photosynthetic rate and hyperspectral data of lettuce were Obtained.The Ratio Vegetation Index(RI),Difference Vegetation Index(DI),Normalized Difference Vegetation Index(NDVI),visualization Visualize Atmospherically Resistive Vegetation Index(VARI),Renormalized Difference Vegetation Index(RDVI),Perpendicular Vegetation Index(PVI)and Vegetation Attenuation Index(PSRI),were selected.The original and first-order transformations of the 14 spectral vegetation indices were studied.The correlation matrix method was used to optimize the optimal vegetation index,which was related to the spectral position variable(Red edge magnitude,Dr)and spectral area variables,namely red edge area(SDr),the ratio of red edge area to blue edge area(SDr/SDb),the ratio of red edge area to yellow edge area(SDr/SDy)as spectral feature variables.The inversion model of the net photosynthetic rate about lettuce was established by using a combination of polynomial fitting and multivariate scatter correction(MSC),standard normal variable transformation(SNV),partial least squares(PLS)and principal component regression(PCR).The results showed that PSRI(515,499)and DI(515,499)at the optimal wavelength position had the greatest correlation with net photosynthetic rate,which could reflect more physiological information of lettuce under particulate pollution.The combined modeling method of SG+MSC+PCR had the highest accuracy,the coefficient of determination R_(c) was 0.9011,and the R_(p) was 0.9458.The modeling effect based on the optimal spectral vegetation index was the best.The spectral area variable SDr/SDb had the highest fitting accuracy(R^(2)=0.9365),which could reliably predict the net photosynthetic rate of lettuce.It was the optimal method to establish the net photosynthetic rate of lettuce based on the spectral position and area variables.This work could provide a certain reference value for the inversion of plant physiological information using hyperspectral technology under a particulate pollution environment.
作者
孔丽娟
隋媛媛
刘爽
陈丽梅
周丽娜
刘春慧
姜玲
李松
于海业
KONG Li-juan;SUI Yuan-yuan;LIU Shuang;CHEN Li-mei;ZHOU Li-na;LIU Chun-hui;JIANG Ling;LI Song;YU Hai-ye(College of Engineering and Technology,Jilin Agricultural University,Changchun 130022,China;College of Biological and Agricultural Engineering,Jilin University,Changchun 130000,China;College of Horticulture,Jilin Agricultural University,Changchun 130022,China)
出处
《光谱学与光谱分析》
SCIE
EI
CAS
CSCD
北大核心
2024年第4期1128-1135,共8页
Spectroscopy and Spectral Analysis
基金
国家自然科学基金青年科学基金项目(32001418)
吉林省重点研发项目(20200402015NC)资助
关键词
颗粒物
最优光谱指数
特征变量
净光合速率
建模反演
Particulate matter
Optimal spectral index
Characteristic variable
Net photosynthetic rate
Modeling inversion
