Establishment Of SPAD Value Inversion Model Based On Hyperspectral Ramie Leaf

Hyperspectral remote sensing data can quickly,non-destructively and accurately monitor the growth,physiological and biochemical characteristics and quality of crops.It can detect subtle changes in the growth process of crop leaves.The spectral software can make the leaves exhibit their corresponding spectral response curves.Hyperspectral data can be used to invert various biochemical parameters of the blade.In the national economy,castor has always had a high economic status.At present,there are few studies on the hyperspectral properties of ramie at home and abroad.The chlorophyll content of ramie in actual growth is also affected by many factors,especially the relationship between reflectance and chlorophyll content.More monitoring methods need to be carried out and compared.By selecting the model with the highest simulation accuracy,it provides a more powerful support for the monitoring of chlorophyll content of crop physiological and biochemical parameters,and provides a theoretical basis for realizing the high yield and quality of ramie and the precise management of Matian.In this study,ramie was used as the research object,and the plant geosynthesis spectrometer and SPAD-502 chlorophyll instrument were used as technical support.The hyperspectral characteristics of the leaves at different upper and lower leaves were studied in depth,and the ramie was analyzed.The variation of leaf SPAD(Chlorophyll content)value and its hyperspectral reflection characteristics,the spectral characteristics of ramie leaf and the sensitive band corresponding to SPAD were obtained.The hyperspectral characteristic parameters of ramie leaf were extracted by using various data dimensionality reduction methods.Principal component analysis and sub-band based on correlation coefficient were used.Principal component analysis and vegetation index were used to extract the characteristic variables.The multiple linear regression method and partial least squares regression method were used to establish the SPAD value prediction model based on the leaf hyperspectral feature parameters,and the best solution was obtained.The main work and research results of the full text are as follows:(1)Statistics on the original spectral parameters and SPAD values of the castor leaves.It is found that the trend of the spectral reflectance curve of the upper,middle and lower leaves is basically the same,but there are some differences in some wavelengths/bands:the spectral reflectance difference of each leaf position is smaller in the range of 420nm~480nm,and the spectral curves overlap each other;The reflection peak appeared at 540 nm(green light band),and the spectral reflectance of the lower leaf position in the range of 510 nm to 630 nm was significantly higher than that of other leaf positions;the reflectance increased sharply from 680 nm(red valley band).(2)Correlation analysis between SPAD values of ramie leaves and original spectral reflectance and vegetation index.It is found that the maximum wavelength of the correlation coefficient in the original spectrum is 715nm(r=-0.813),and the sensitive band is in the range of 509nm-624nm and 693-733nm.In the correlation analysis between vegetation index and SPAD value,MCARI(improved chlorophyll absorption reflection)is found.The index and CARI(chlorophyll absorption ratio index)parameters have the best correlation with SPAD values,and the correlation coefficients are 0.858 and 0.838,respectively.(3)Using principal component analysis,sub-band principal component analysis based on correlation coefficient and vegetation index to reduce the dimensionality of ramie leaf hyperspectral data.The first five principal factors were extracted as variables in the full-band PCA analysis.In the sub-band principal component analysis based on correlation coefficient,the eigenvalues of the first three components are all greater than 1 and the eigenvalue accumulation rate has reached 99.398%.Therefore,the first three principal component factors are selected as variables to accurately reflect the information of the original data.The construction of the model.In the vegetation index method,the ratio vegetation index RVI,normalized vegetation index NDVI,leaf chlorophyll index LCI,conversion chlorophyll absorption ratio index TCARI,chlorophyll absorption ratio index CARI,improved chlorophyll absorption reflex index MCARI and improved red edge ratio were selected.The MSR,which is closely related to the chlorophyll content of vegetation,is used as a characteristic variable to construct a predictive model of physiological and biochemical parameters of ramie.(4)Applying the above-mentioned characteristic variables to multiple linear regression and PLSR partial least squares regression to establish an inversion model based on SPAD values of hyperspectral ramie leaves and compare and analyze them.The results show that the PLSR model based on the sub-band raw data has high precision,and the verification fitting coefficient is close to 1,R~2=0.6896,RMSE=2.576,RE=3.8%,and the root mean square error and relative error are the smallest.The best model for leaf SPAD values.