Study On Wheat And Maize Growth Information Base On Hyperspectral Imaging Technology

In this study,theory and method of hyperspectral remote sensing was used to investigate the spectral characteristic of winter wheat and summer maize,and analyze the correlation between agronomic parameter and parameters of spectral characteristic.Then establish hyperspectral monitoring model,to provide theoretical basis and technical support for Precision agriculture and monitoring in HuangHuaiHai region.Five nitrogen treatments were set up N0(0 kg/hm~2)?N1(150hm~2)N2(300hm~2)N3(450hm~2)N4(600hm~2).we got spectral reflectance and grayscale image through infrared imaging spectrometer SOC710 P,then computed the vegetation index and grayscale calculated values and initiated the correlation analysis,then fitted the models of LAI?dry matter on the ground?nitrogen content of leaves and got the feasibility of grayscale calculated values to monitor the growth of wheat and maize.(1)The monitoring model of wheat LAI:First,initiated correlation analysis between spectral parameters and LAI and fitting the monitoring model,than using the measured value and simulation value to compute root mean square variance RMSE(%)and relative error RE to test models.GND was the best parameter to fitting the monitoring model.Here comes the model y=1.885-15.68X+22.58x~2 coefficient of determination was 0.81 RMSE(%)=1.01 RE=19.69%.(2)The monitoring model of maize LAI:First,we initiated correlation analysis of spectral parameters and LAI and fitting the monitoring model,the models based on single spectral parameter had lager RESE and RE,the accuracy of models was terrible,so we used least squares method fitting the models based on grayscale calculated values and the model was y=0.97GD(550 700)+0.603GND(550 700)+4.81GSA(550 700)coefficient of determination was 0.82 RMSE(%)=1.2 RE=11.2%.The monitoring model of dry matter weight on the ground of wheat :First we initiated correlation analysis of spectral parameters and LAI further more fitted the monitoring model,than using RMSE(%)and RE to test the model.GND have stronger correlation coefficient,and higher coefficient,Finally,we got the model based on GND.y=-0.236+2.63x+2.53x~2 coefficient of determination was 0.78 RMSE(%)=0.5 RE=25.6%.The monitoring model of dry matter weight on the ground of maize :First,we initiated correlation analysis of spectral parameters and dry matter weight on the ground and fitting the monitoring model,the models based on single spectral parameter had lager RESE(%)and RE,the accuracy of models was too terrible,so we used least squares method fitting the models based on grayscale calculated values and the model was y=57.07GR(660 760)+31.16GND(660 760)+41.27GGR(660 760)RMSE(%)=0.65 RE=46.2%(3)The monitoring model of wheat leave nitrogen content:First,we initiated correlation analysis of spectral parameters and leave nitrogen content further more fitting the monitoring model.Here came the result,compared with other spectral parameters,GND had stronger correlation,larger determine of the coefficient and lower deviation.So we chose the model based on GND(550 680)y=1/(0.211×0.12x)RMSE(%)=5.7 RE=26.6%.