Research On Hyperspectral Monitoring Of Sugar Beet Growth Parameters Under Different Nitrogen Levels

Sugar beet planting in Inner Mongolia area is a special economic crop,which has an important effect on the development of the Inner Mongolia area yield of sugar industry. Simple, rapid, non-destructive way to get beet growth conditions, a reasonable guide fertilization, promote beet yield of polysaccharides, is needed in production of sugar beet cultivation technology. In this study, by setting different nitrogen gradient (nitrogen levels, N0-N6; Nitrogen was 0,15.32,76,108,163,217 kg·hm-2), using hyperspectral technology to detect the sugar beet growth parameters. In this thesis, through the ASD Qualityspec spectrometer measured canopy reflectance, explore sugar beet spectral response characteristics at different nitrogen levels and different growth stages. On this basis, analysis of canopy spectral sensitive areas, screening different spectralcharacteristic parameters and regression method, construction of sugar beet in the whole growth period of aboveground biomass, nitrogen content, SPAD value of Hyperspectral Estimation model.The main work and research results are as follows:1.According to the field measurement of sugar beet canopy spectral data, the spectral reflectance curves of different growth periods and under different nitrogen levels and basically the same trend. Canopy spectral reflectance under different nitrogen levels, near infrared reflectance increases with the increase of nitrogen application, the reflectance of visible light decreased with the increase of nitrogen in sugar beet; different growth period, the canopy spectral reflectance increases with the increase in the early stage of canopy leaves, the latter is due to the decline of leaf reflectance decreases.2.Using four kinds of pretreatment method (smoothing, derivative, multiplicative scatter correction, standard normal transform) on canopy reflectance with biomass, nitrogen content, SPAD value of the correlation analysis, showed that derivative treatment can extract more sensitive band information associated with the growth parameters, multiple scattering correction and standard normal transformation have a certain improvement on the correlation coefficient, smoothing compared with other pretreatment methods were poor.3.Using three kinds of regression method (PLSR combination of spectral preprocessing and PCR, vegetation index, band depth analysis combined SMLR) establish beet aboveground biomass, nitrogen content, SPAD value estimation model, and by root mean square error and relative error of model effects to evaluation. The results showed that: the optimal band depth information combined with multivariate linear regression in theestimation of biomass have the ideal result (RMSE=128.34gm-2,RE=21.6%); using BDR combined SMLR at the nitrogen content estimates get better results using (RMSE= 2.30g kg-1, RE= 18.8%);BDR combined with SMLR to establish SPAD value estimation model accuracy is the highest (RMSE=2.54, RE=4.5%).