| Hyperspectral remote sensing is one of the important tools of precision agriculture.It is also an important way to monitor crop growth and nutrition status real-timely,rapidly and accurately without destruction.Agronomy parameters like Chlorophyll content and leaf area index are effective indicators to measure photosynthetic capacity,growth and development stage,nutrition and physical conditions and the degree of environmental stresses in wheat.Therefore,the response of the wheat canopy spectral reflectance to agronomy parameters is also the powerful tools to predict growth stage,monitor photosynthesis and nutrition condition and assess grain yield in wheat.Taking the wheat variety Pubing151 as the experimental material,wheat canopy spectral reflectance and agronomic parameters were measured respectively in different stage of growth and development based on the field trials with different N application rate and seeding rate.The relationship between agronomic parameters and spectral reflectance,its first derivative spectra and the existing spectrum indexes were analyzed.And quantitative monitoring models for wheat agronomy parameters were established.wheat grain protein content and yield spectral prediction models were established based on correlation analysis between the grain protein content and yield and wheat canopy chlorophyll content and leaf area index in different growth period.Expected results would provide the technical basis for reasonable management of the amount of nitrogen fertilizer and sowing rate,provide theoretical basic and key technology for non-destructive monitoring and accurate management of wheat growth characteristics.The main results were summarized as follows:1.Changing characteristics of wheat canopy spectral reflectance in different growth period were analysed under different N application rate and seeding rate.The results showed that canopy spectral reflectance fell with nitrogen treatment in visible band and under 129kg·hm-2(D1)and 202 kg·hm-2(D2)sowing rate,they reached the minimum values under 120kg·hm-2(N2)and 180 kg·hm-2(N3)application rate respectively,which showed an opposite trend in near infrared wave band.In visible band,the canopy spectral reflectance value underD2 sowing rate was lower than that under D1 sowing rate,which was opposite in the near infrared band.The "red shift" phenomenon was produced with appropriate N application rate and increased sowing rate.The changing trend of wheat canopy spectral reflectance in different growth period was almost same,reaching the maximum value at filling stage in visible band and at the heading stage in near infrared wave band.There was "red shift" phenomenon from jointing stage to heading stage,while there was "blue shift" phenomenon from heading stage to the flowering period.And red edge area reached the maximum and minimum value in the heading stage and filling stage respectively.These spectral information would provide basis for further utilization of wheat canopy spectral reflectance to monitor wheat growth condition.2.The correlation between the chlorophyll a,chlorophyll b and chlorophyll a+b and the original spectral reflectance,the first derivative spectral was discussed on the basis of the changing characteristics of wheat chlorophyll content under different sowing rate and nitrogen levels in different growth period.The result indicated that the spectral reflectance and canopy chla+b content showed a negative correlation in the visible band and showed a positive correlation in near infrared wave band.And the correlation coefficient reached the maximum value near the wavelength of 702 nm,which was 0.73.The correlation of Chl a content to canopy spectral reflectance was similar to chl a+b content,and correlation of chl b to canopy spectral reflectance was below chl a and chl a+b.The quantitative relationships of chlorophyll a,chlorophyll b and chlorophyll a+b with existed spectrum parameters were further analyzed.This showed that there was close relationship between the spectrum parameters RVI(729,755)and GREEN-NDVI and chlorophyll content,and the fitting effect of polynomial regression equation was better than the linear regression equation.By inspection with independent data,the estimate effects of RVI(729,755)and GREEN-NDVI to Chl b and Chl a+b were better,R2 were 0.5022 and 0.5305,0.5022 and 0.5305 respectively,while the estimate effects of RVI(729,755)to Chl a is poorer,R2 was 0.2918.Therefore,wheat canopy chlorophyll content monitoring model based on GREE-NDVI was better with comprehensive consideration of modeling performance of different spectral parameters.3.The correlation between the LAI and the original spectral reflectance,the first derivative spectral was discussed on the basis of the changing characteristics of LAI under different sowing rate and nitrogen levels in different growth period.The result indicated the correlation coefficient between the spectral reflectance and canopy LAI reached the maximum value near the wavelength of 670 nm and 1010 nm,which were-0.89 and 0.90.With 10 kinds of spectral indexes that had better correlation with LAI as independentvariables,LAI as the dependent variable,the linear model,polynomial model,index model and power function model were established respectively.This showed that there was closed relationship between the spectrum parameters NDVI 705,NDCI,OSAVI,NDVI(810,560)and LAI.By inspection with independent data,the estimate effects of NDVI 705 to LAI were best,r,R2,RMSE,RE,da,dap were 0.74,0.548,0.354,5.86 %,0.097,1.61 %respectively.The predictive value was close to the measured value.4.The correlation of mature grain protein content to wheat canopy chlorophyll content in different growth period was analyzed and the fitting model was set up.By inspection with independent data,it showed that the chlorophyll content of filling stage can be used to better predict the grain protein content in mature,R2 reached 0.635.On this basis,the regression analysis between the spectral parameters that could better predict wheat canopy chlorophyll content and grain protein content was carried out.The results showed that the polynomial models with NVI and PSSRb as independent variable were better and the fitting precision were 0.3278 and 0.3014 respectively.But inspection effect of PSSRb was better than NVI by examination.r,R2,RMSE,RE,da,dap were 0.833,0.693,0.152,1.11 %,0.108 and 1.11 %.Therefore,the monitoring model of wheat grain protein content with PSSRb as variable in mature period was best.5.The correlation of grain yield to wheat canopy LAI in different growth period was analyzed and the fitting model was set up.By inspection with independent data,it showed that the LAI of flowering stage can be used to better predict the grain yield,R2 reached 0.492.On this basis,the regression analysis between the spectral parameters that could better predict wheat canopy LAI and grain yield was carried out.The results showed that there were closed relationship between Spectral parameters NVI,PSSRb,OSAVI and MSAVI and grain yield,and fitting precision was 0.5405,0.6244,0.5926 and 0.5405 respectively..But inspection effects of PSSRb and MSAVI were better than NVI and OSAVI by examination.r,R2,RMSE,RE,da,dap were 0.833,0.693,0.152,1.11 %,0.108 and 1.11 %.Therefore,the monitoring model of wheat grain yield with NVI and OSAVI as variables in mature period was better. |
PDF Full Text Request