Unmanned Aerial Vehicle (UAV) Based 'Image And Spectrum' Information Fusion For Nitrogen Diagnosis In Winter Wheat

Unmanned aerial vehicle(UAV)based hyperspectral remote sensing technology,with the advantages of flexibility,low cost,and short reture cycle,can obtain high spectral resolution and high spatial and temporal resolution remote sensing image,so it has become a hot topic and an important data source in precision agriculture.Nitrogen is one major nutrient element during plant growth and organ development,and it is of great significance for plant growth monitoring and yield estimation.Therefore,to improve nitrogen use efficiency and ensure crop yield and quality,it is necessary to understand and obtain nitrogen nutrition status of plants in the process of plant growth.Saturation problem exists in the inversion of agronomic parameters by spectral features.Hyperspectral remote sensing technology also provides abundant texture information.Texture features can provide spatial structure information of crops,effectively weaken the saturation problem,and provide support for the inversion of physical and chemical parameters.Therefore,in this study,field experiment was carried out at Xiaotangshan National Precision Agriculture Research Demonstration Base in Changping District,Beijing.Hyperspectral images of winter wheat were obtained based unmaned aerial vehicle at booting stage and anthesis stage,also field experiment data of winter wheat were obtained.A nitrogen diagnosis model based 'image and spectrum,information fusion was constructed by integrating spectral information and texture information of hyperspectral image,also agronomy knowledge model combined,hopefully,to provide scientific nitrogen diagnosis to winter wheat.Some detailed conclusions can be drawn as follow:(1)Spatial variability of biomass and plant nitrogen accumulation in winter wheat was analyzed.Results showed that the increase of nitrogen application could promote the accumulation of plant nitrogen and wheat biomass,whilst variability of nitrogen accumulation and biomass in winter wheat were reduced.However,excess nitrogen would not promote the accumulation of nitrogen accumulation and biomass,also the variability of nitrogen accumulation and biomass would not decreased.(2)Anti-saturation ability and the accuracy of biomass model based on the optimized spectral index was explored.Results showed that biomass model constructed from optimized indices by integrating spectral indices and texture feature(R2=0.798)yield more than that of the vegetation index based model(R2 = 0.619)and texture feature based model(R2 = 0.727)in anti-saturation ability and accuracy.Integrating spectral information and texture information could performed better in estimation of biomass in winter wheat.Remote sensing information has a good application prospect in the application of crop yield estimation and growth monitoring.(3)Nitrogen nutrition index(NNI)and nitrogen deficiency model were constructed by'image and spectrum' information fusion,and then winter wheat nitrogen concentration was studied.Results showed that increased nitrogen application would promote accumulation of nitrogen in wheat,but excess nitrogen application rate had no obvious effect on the increase of plant nitrogen concentration.The optimum nitrogen application rate is between 0.5 times of regular fertilizer level(N2)and regular fertilizer level(N3)for ZM175,and between regular fertilizer level(N3)and 1.5 times of regular fertilizer level(N4)for J9843.Also,much more nitrogen was necessary during anthesis stage to meet growth requirements.Therefore,spectral information and texture information fusion could provide a scientific nitrogen diagnose in crop,which is of great importance for crop precision research.