Diagnosis Of Nitrogen Nutrition In Rice Based On UAV Remote Sensing

Nitrogen(N)is one of the most important nutrients in crop production in the world.Nitrogen is also the most nutritious element in most crops.For rice growth and development,nitrogen is an indispensable one of the important nutrients.Reasonable application of nitrogen fertilizer can not only improve the quality and yield of rice,but also alleviate the pressure from the resources and environment brought by agricultural production.According to the requirement of pr ecision agriculture,it is necessary to obtain the information of farmland features quickly and accurately.In this paper,we use UAV technology to m onitor the nutrient status of canopy nitrogen and chlorophyll in rice in real time and to further guide the fine management of nitrogen in the field,so as to reduce the production cost,improve the utilization rate of nitrogen fertilizer and reduce the environmental pollution.This paper was based on two varieties of rice field nitrogen gradient test(N0-N5).The canopy image of ri ce was obtained by using fixed-wing unmanned aerial vehicle(UAV)as the basic research object.From the visible remote sensing image and multispectral image to extract the im age features corresponding to the ground sampling points.The relationship between the characteristic values of the image and the ground measured values were analyzed.The spectral response characteristics of rice canopy were explored,and the spectral parameters of rice canopy spectral response were identified.The inversion model of rice canopy physiological parameters was cons tructed.To explore the feasi bility of monitoring nitrogen nutrition status of rice in large scale field by using UAV images.The main conclusions of this study are as follows:Because the rice vari eties had a gre at influence on the e igenvalues of the canine digital images of aerial photography,the digital images taken by UAVs were not sui table for the classification and i nversion of c anopy nitrogen in this field.In contrast,airborne multispectral images are more suitable for classification and inversion of canopy nitrogen in field rice.From the recognition results of canopy nitrogen of s ingle cultivar rice with different nitrogen application rates,the nitrogen gradient was th e highest in the overall classification of severe nitrogen deficiency(N0)and severe nitrogen application(N5),and the recognition rate was more than 95%.And rice varieties and classification methods have little effect on the recognition accuracy of N0 and N5.For the multi-species rice nitrogen level identification in Daejeon,the green normalized vegetation index(GNDVI)i mage obtained by the multi-spectral image is the highest overall recognition accuracy of N0 and N5,reaching 93.83%.The identification of the four nitrogen grades delimited by N0-N5,which the total nitrogen deficiency,micro-nitrogen deficiency,micro-nitrogen application and sev ere nitrogen application were poor.And the highest was only 57.47%,which indicated that the slight change of nitrogen in rice plants showed no obvious nitrogen classification in rice canopy images.The results showed that the fitting accuracy of spectral index of canopy and rice physiological parameters were higher.For the two rice cultivars,the correlation between GNDVI and SPAD was the highest(R2 was 0.9478 and 0.8587,respectively).The results showed that it was feasible to use GNDVI image to carry out the inversion monitoring of rice physiological parameters.Considering the influence of ric e varieties,in order to increase the universality of the inversion model,the regression model of SPAD and spectral index of the whole rice leaves was obtained after the data of the north and south experimental areas were m erged,and the optimal inversion model for nitrogen content was obtained.So as to provide reference for real-time,rapid and non-destructive monitoring of rice physiological parameters by using canopy image characteristics.