Study On Estimate Model Of Corn Grow Information Based On Hyperspectral Data

“Grains are the life of peoples and the treasure of Countries.” Food production is a strategic industry that can be responsible for the social stability and harmony. To realize the strategic industry, the precision of agriculture can be taken advantage, for example, we can gain the information of the growth of crops using Hyperspectral Remote Sensing(HRS) technology. In this research, we detected the change of chlorophyll content and water content in the leaves of corn planted in the field through HRS technology. Furthermore, we analyzed the correlation between these parameters and the spectral reflectivity, then we selected the appropriate spectrum characteristic parameter variables and extracted the variables sensitive to agronomy parameter wavelengths. We build the single variable linear and nonlinear regression equation about agronomy parameter through analyzing the correlation between these parameters and agronomy parameter. At the same time, we evaluated the accuracy of the high determination coefficient model. Then we selected the optimal quantitative modle for inversion of the corn growth information. The main results are as follows:(1) The characteristics of corn canopy spectral reflectance curve were in conformity with that of most of the green crops spectral reflectance curve. General performance: there were 2 plant spectral reflectance lower area within the scope of the blue light band range and red light band. Besides, there was a reflection peak near 550 nm range of the green plant spectrum with a higher reflectivity. The reflectivity of the spectrum of green plant suddenly increased near 710 nm range and became stable after arriving at near infrared wave band area, which formed a spectral reflectance high platform within the scope of the near infrared wave band area. There were significant differences in the characteristics of corn spectral reflectance curve in different growth period. There were decreasing trends of reflectance from seedling stage to jointing stage and to grouting spinning stage in the visible wavelength range, while there were increasing trends of reflectance from grouting spinning stage to mature period. Moreover, corn spectral reflectance values showed a trend of rising from seedling stage to jointing stage and to grouting spinning stage within the scope of the near infrared wave band area. Spectral reflectance in a declining trend from grouting spinning stage to mature period.(2) In corn different growth period, spectral curve edge(red) showed big differences in 680-760 nm spectral range. The location of the spectral reflectance hypotenuse(red edge) shift to the right in 680-760 nm spectral range, from seedling stage to jointing stage and to grouting spinning stage. After grouting spinning stage, the location of the spectral reflectance hypotenuse(red edge) shift to the left in 680-760 nm spectral range. Moreover, with the growth period of corn, the red edge amplitude of corn showed the same change features with the red edge position, followed with first increasing then decreasing.(3) Under the condition of different chlorophyll content, there are big differences between corn spectral reflectance curve values. The higher the value of chlorophyll content, the lower the reflectivity of corn in the range of visible light, with the different changes in near infrared wavelengths area. Moreover, corn canopy spectral amplitude of red edge had the same changes with the chlorophyll content. Based on the sensitive bands of the original canopy, sensitive bands of the first derivative spectra and different vegetation index to construct the cornchlorophyll regression model. At the seedling stage based on the model of the effect of sensitive bands of the first derivative spectra of corn chlorophyll regression were the best, the establishment of equation y=-3413x2-3595x+55.027 decision coefficient is the highest, and R2 =0.88, RMES=1.89, RE%=2.94.(4) Under the condition of different leaf water content, corn sensitive wavelengths and spectral signature can produce significant changes, which were expressed in a opposite trend of the corn canopy spectral reflectance curve with the change of leaves water content, thus the less the value of leaves water content, the greater the reflectance curves, besides, the influence of the change of leaves water content on the corn canopy spectral reflectance curve was more significant in near infrared wave band area. Based on the sensitive bands, spectral indices constructed water rate regression model. In the mature period based on water in leaves of corn(Rg-Rr) /(Rg+Rr) rate regression model is the best, established power equation y= 0.2132x-0.775 decided coefficient of the highest, R2=0.58, RMES=0.0220, RE%= 2.42.(5) Under different biomass condition, remarkable changes have taken place for the original spectrum and the derivative spectral characteristics of corn in different growth period, which were expressed in a same trend of the value of the corn canopy spectral reflectance curve with the change of corn biomass, thus the greater the value of corn biomass, the greater the reflectance curves. Based on the sensitive bands of the original canopy, sensitive bands of the first derivative spectra and different vegetation index to construct the corn biomass regression model. At the seedling stage based on the model of the effect of sensitive bands of the first derivative spectra of corn biomass regression were the best, the establishment of regression equation y = 108x2-307x-305 decision coefficient is the highest, and R2 = 0.67, RMES=8.75, RE%=5.17.