Estimation Of Winter Wheat Chlorophyll Content Based On Canopy Hyperspectral Data

Winter wheat is one of the important food crops in China,the chlorophyll content is the basic substance of winter wheat photosynthesis and material accumulation,the level of its content has a good indicator effect on the photosynthetic capacity,growth and nutrient status,and yield of winter wheat.The traditional method of chlorophyll content determination is time-consuming,laborious,and difficult to be popularized in a large area.In recent years,with the development of hyperspectral remote sensing technology in crop growth monitoring,it provides an effective way for fast and nondestructive estimation of crop chlorophyll content.This paper took winter wheat as the research object,Xiaotangshan National Precision Agriculture Research Demonstration Base in Beijing was selected as the research area,ASD hyperspectral spectrometer was used to obtain the canopy hyperspectral data of winter wheat at different growth stages,combined with the measured data of chlorophyll content,transformed hyperspectral data,obtained vegetation index,characteristic parameters,fractional differential spectrum,wavelet energy coefficient,and principal component,used univariate analysis,optimal subset regression,support vector machine,multiple linear regression,and other methods,construct a model for estimating the chlorophyll content of winter wheat in different growth stages.The conclusions of this paper are as follows:(1)In each stage of growth,the vegetation index with the strongest correlation with chlorophyll content was REP(Red Edge Position Index).In jointing stage and booting stage,flowering stage and filling stage,full growth stage,the characteristic parameters with the strongest correlation with chlorophyll content were SDb(blue edge area),λR(red edge position),and the ratio of SDR(red edge area)to SDb(blue edge area).In each stage of growth,the differential spectrum with the strongest correlation with chlorophyll content were all in integer-order.In jointing stage and booting stage,flowering stage and filling stage,full growth stage,the corresponding layers of the wavelet energy coefficient with the strongest correlation with chlorophyll content were5,1,7,respectively.(2)In jointing stage,the principal component-multiple linear regression model had the best accuracy in estimating chlorophyll content,the R~2,RMSE and n RMSE of modeling were 0.62,2.54μg/cm~2,6.96%,respectively,the R~2,RMSE,n RMSE and RPD of the validation model were 0.61,3.66μg/cm~2,9.71%,1.40,respectively;in booting stage,the wavelet energy coefficient-support vector machine model had the best accuracy in estimating chlorophyll content,the R~2,RMSE and n RMSE of modeling were 0.87,1.80μg/cm~2,4.46%,respectively,the R~2,RMSE,n RMSE and RPD of the validation model were 0.70,3.67μg/cm~2,8.74%,1.56,respectively;in flowering stage,the wavelet energy coefficient-optimal subset regression model had the best accuracy in estimating chlorophyll content,the R~2,RMSE and n RMSE of modeling were 0.78,2.12μg/cm~2,4.77%,respectively,the R~2,RMSE,n RMSE and RPD of the validation model were 0.75,4.44μg/cm~2,9.73%,1.58,respectively;in filling stage,the wavelet energy coefficient-support vector machine model had the best accuracy in estimating chlorophyll content,the R~2,RMSE and n RMSE of modeling were 0.74,6.68μg/cm~2,18.46%,respectively,the R~2,RMSE,n RMSE and RPD of the validation model were0.71,8.29μg/cm~2,24.14%,1.77,respectively;for the full growth stage,the wavelet energy coefficient-support vector machine model had the best accuracy in estimating chlorophyll content,the R~2,RMSE and n RMSE of modeling were 0.72,4.57μg/cm~2,11.70%,respectively,the R~2,RMSE,n RMSE and RPD of the validation model were0.73,5.35μg/cm~2,13.64%,1.73,respectively.