Hyperspectral Predictive Model Of SPAD Value Of Different Early Indica Rice And Autumn Maize Varieties

In this paper,in order to find the spectral characteristic diversity of two kinds of crops and the best estimate model of SPAD value,using different varieties of early indica rice and autumn corn as research object,using the ASD Field Spec 3 spectrometer features spectral data and plus chlorophyll meter SPAD-502 acquisition of chlorophyll relative value,and analyzing the correlation between SPAD value of rice and maize and its first derivative spectrum and spectral parameters.Also using sensitive wavelength and maximum correlation spectrum parameters of two kinds of different crop varieties,as the independent variable.The SPAD value estimation models of early indica rice and autumn maize varieties were constructed,and the following results were obtained:（1）The overall variation trend of the raw spectral reflectance of the 12 early indica rice varieties are the same,all of which showed lower reflectivity in the visible band than in the near-infrared band.At the same time,all the first derivative spectra show two distinct peaks in the range of visible green light and in the near-infrared band,and two distinct troughs in the near-infrared band.Moreover,the spectral reflectance of each species showed a trend of decreasing with the progression of the growth period.（2）Selecting 6 sensitive wavelength based on the first derivative spectrum and 5spectrum index with the highest correlation coefficient as the independent variables,and selecting SPAD value as the dependent variable,built different models to predict the SPAD value of early indica rice varieties,and the decision of the calculation model for the coefficient（R²）,root mean square error（RMSE）and relative error（RE）,and the best estimating models for different varieties of SPAD value are as followed:Zhuliangyou689,Y=-2357.791*DSI(R’₁₉₀₇,R’₅₁₃)+36.902;Lingliangyou 265,Y=-1982.599*DSI(R’₁₉₀₅,R’₅₂₄)+40.344;9liangyou 179,Y=7711.955*R’₅₀₉+34.033;Zhuliangyou 158,Y=-17980.797*R’₆₁₈+35.401;Zhongzao 35（CK）,Y=-1432.906*DSI(R’₁₉₇₅,R’₇₀₅)+34.578;Lingliangyou 747,Y=-6862.259*DSI(R’₂₂₃₆,R’₆₈₀)+37.461;9liangyou165,Y=-14179.845*R’₅₈₅+34.400;Zhongzu 120,Y=6686.750*DSI(R’₁₉₇₇,R’₅₇₅)+29.897;Zhongzao 59,Y=-5842.944*DSI(R’₁₉₁₇,R’₅₁₃)+36.222;Zhongliangyou 474,Y=7541.843*DSI(R’₁₉₅₆,R’₅₈₃)+34.289;Jialiangyou 500,Y=6904.326*DSI(R’₁₉₇₀,R’₅₈₅)+34.815;Zhongzao 64,Y=6904.326*DSI(R’₁₉₇₀,R’₅₈₅)+34.815。（3）The variation law of the raw spectral reflectance of different autumn maize varieties is generally the same,all of which show low reflectivity in the visible band and high reflectivity in the near-infrared waveband,and the wave peaks are close to each other in the visible waveband.The first derivative spectral curves of different varieties have bimodal characteristics in the waveband of 400nm-1000nm.（4）The best estimation models for SPAD values of each variety were obtained by considering RMSE and RE of 8 autumn maize cultivar modeling sets and validation sets:Shuangxingyu No.2,y=-19533118R’²₆₇₁+48050.806R’₆₇₁+46.482;Zhengda999,y=14888708R’²₆₇₆+15133.378R’₆₇₆+41.232;Xiangnongyu No.27,y=86746.221DSI(R’₆₂₈,R’₆₇₃)+18.798;Xingyu818,y=136.176NDSI(R’₆₆₇,R’₇₃₉)+190.901;Linao No.1 B₃,y=31211.09DSI(R’₅₄₉,R’₆₇₁)+54.037;Xiangnongyu No.32,y=26016.52DSI(R’₅₅₃,R’₆₇₇)+54.037;Luoyu No.1,y=530.47.443DSI(R’₅₃₉,R’₆₅₄)+91.222;Tianyu335,y=-12646105R’²₆₇₇+52181.083R’₆₇₇+29.068.Among them,except shuangxingyu no.2,zhengda 999 and tiantian jade 335,which showed the best polynomial model,the other varieties all showed the best linear model with one variable.