Study On Nitrogen Nutrition Monitoring Of Rice Based On Hyperspectrum

Rice is one of the main food crops in China.Increasing rice yield can affect the living standards of the people and provide food security.Nitrogen fertilizer is one of the important factors affecting crop production.The use of nitrogen fertilizer has a great impact on crop yield and quality.In order to increase rice yield,nitrogen fertilizer is often applied excessively,causing serious damage to the environment,so rational fertilization has become necessary.Scientific fertilization is an important and important way to solve this problem,which is beneficial to increase the utilization rate of nitrogen fertilizer,increase grain production,and reduce environmental pollution.Hyperspectral remote sensing technology has the characteristics of fast,accurate,real-time and non-destructive when acquiring crop information.Therefore,in agricultural production,hyperspectral technology can be used to monitor the nitrogen nutrition status of rice to guide fertilization,which is of great significance for the development of precision agriculture.In this paper,the high field test base of Jiangxi Irrigation Test Center Station was used as the research area to determine various agronomic parameters and corresponding canopy spectral data of rice.The spectral response characteristics of rice under different nitrogen application levels were studied.Based on the spectral data of rice,single-band,“three-sided”parameters and vegetation parameters with high correlations between leaf nitrogen and leaf SPAD values were extracted to construct related diagnostic models.The main findings are as follows:1.Through the analysis of the spectral characteristics of rice canopy,it was found that there were regularity in the change of reflectivity:the reflectivity in visible region was small,and the reflectivity in near infrared region was large;With the increase of nitrogen application level,the spectral reflectance of canopy decreases in visible region and increases in the near infrared region.2.After deriving the spectral reflectance of rice canopy,the"three-sided"parameter can be extracted.The red edge amplitude Dr and the red edge area SDr increase with the increase of nitrogen application level;the red edge position?r increases with the increase of nitrogen application rate,that is,the red edge position?r of rice varies with nitrogen content.The increase of"red shift"occurred;the green peak position?g decreased with the increase of nitrogen application rate,so the green peak position?g of rice occurred"blue shift"with the increase of nitrogen application rate.3.The changes of nitrogen content in rice leaves under different nitrogen levels were studied.It was found that with the growth and development process,the nitrogen content of rice leaves decreased continuously after tillering stage;in the same period,it increased with the increase of nitrogen application rate.4.Analysis of spectral reflectance of rice canopy and nitrogen content in leaves showed that the correlation coefficient at band 1397nm was the largest,which was 0.994.Based on the single-band construction model,the formula is y=1.3594e^0.0343x and R²=0.9995.The canopy spectrum of rice was deduced,and the"three-sided"parameters were extracted.The model of the yellow-edge area SDr with high correlation with the nitrogen content of the leaves was selected.The formula was y=0.0006x²-0.0241x+2.2794,R²=0.9984.On the basis of predecessors'vegetation index,five kinds of indices mSR₇₀₅,VOG2,VOG3,R-M and MTCI with high correlation with leaf nitrogen content were selected to construct the model.The R-M coefficient?R²?was the largest,and the formula was y=1.695e^0..0197x,R²=0.9721.5.According to the analysis of rice canopy spectral reflectance and leaf SPAD value,the correlation is the highest at 1464nm,and the single-band model is constructed based on1464nm.The formula is y=27.894e^0.0248x and R²=0.9302.The model is constructed by selecting the green peak position?g with high correlation with the SPAD value of the blade.The formula is y=-594.5n?x?+3787.6,R²=0.924.Five kinds of indices R'665,DVI?R'665,R'460?,NDVI?R'665,R'460?,PRI,NDVI?R'665,R'460?with high correlation with leaf SPAD value were selected.Among them,NDVI?R'665,R'460?has the largest correlation coefficient,and the formula is y=-4.5261x+47.005,R²=0.9916.