Inversion Of Nitrogen Content In Apple Trees Canopy Based On GF-1 Satellite Image

Nitrogen is one of the indispensable nutrients in apple growth process.It's an important material basis for fruit tree growth.The traditional method of measuring plant nitrogen content is time-consuming,laborious and real-time,which can't meet the needs of large-scale,rapid and real-time monitoring.Satellite remote sensing has been widely used in crop growth and nutrition diagnosis because of its low cost,easy access and handling,and has become a research hotspot.In this study,the apple orchard in Qixia City,Yantai,Shandong Province was taken as the research area.The nitrogen content of Apple canopy in the study area was retrieved by using GF-1 satellite remote sensing image and combining with the nitrogen content data of apple canopy leaves.The GF-1 multispectral image was used to compare and analyze the ground measured spectral data.The blue,green,red and near infrared bands of the GF-1 image were calculated.The vegetation index was extracted by band operation,which was compared with the nitrogen content in the canopy of apple trees.Through correlation analysis,the nitrogen sensitive vegetation index of apple canopy was screened.The inversion model of nitrogen content in apple canopy was established by regression model of single variable and multivariable factors.The best inversion model of nitrogen content in apple canopy was screened out.Using GF-1 full-color and multispectral fusion images,the apple orchard in the study area was extracted.Combined with the best inversion model of nitrogen content,the apple orchard in the study area was used.The canopy nitrogen content was spatially inverted.The main results are as follows:(1)Relevant vegetation indices of nitrogen content in apple canopy were extracted and screened.Using the pre-processed GF-1 multispectral images and considering the spectral characteristics of vegetation.Twelve vegetation indices were extracted based on the blue,green,red and near infrared bands of GF-1 images.Correlation analysis showed that the correlation vegetation index was selected,which were RVI,NDVI,EVI,VARI,NPCI and NRI.(2)Established and tested the inversion model of apple tree canopy nitrogen content.Established and tested a single variable based inversion model of apple tree canopy nitrogen content.Using RVI,NDVI,EVI,VARI,NPCI and NRI as independent variables and nitrogen content as dependent variables,a single variable nitrogen content model was established.The results showed that the quadratic polynomial model with NPCI as independent variable had R~2 of 0.59 and RMSE of 1.18.The model had the highest accuracy and stability,and was based on the best single variable nitrogen content model.The model is:N_C=20.16*NPCI~2-29.54*NPCI+29.73,which can be used to estimate nitrogen content.Established and tested a multivariate based inversion model of apple tree canopy nitrogen content.Using RVI,NDVI,EVI,VARI,NPCI,NRI as independent variables and nitrogen content as the dependent variable.A linear regression model was established for the canopy nitrogen content of apple trees:MLR,MSR and PLSR.The nonlinear regression models:SVM and RF.In linear regression model,MSR model has higher R~2 and lower RMSE,which are 0.69 and 1.07.The accuracy and stability of the model are the highest.The model was:N_C=35.74-41.98*NPCI-10.78*NDVI.In the nonlinear regression model,the R~2 of the SVM model test are higher and the RMSE is 0.61 and 1.33.Therefore,the nonlinear optimal inversion model based on multivariate is SVM model.(3)The optimum model for inversion of nitrogen content in Apple canopy was determined.The established univariate apple canopy nitrogen content inversion model and multivariate apple nitrogen content inversion model were compared and analyzed.The best inversion model of the canopy nitrogen content of apple trees was obtained.The model is:N_C=35.74–41.978*NPCI-10.78*NDVI.(4)The extraction of apple orchards and spatial inversion of nitrogen content in Apple canopy were completed in the study area.The full-color and multispectral images of GF-1 were fused by NNDiffuse Pan Sharpenning.The apple orchard information of the fusion image of GF-1 was extracted by using NDVI,elevation and slope information of the study area and maximum likelihood method.Based on the established optimal model of apple tree canopy nitrogen content,the spatial inversion of apple canopy nitrogen content was carried out.The remote sensing spatial inversion map of apple canopy nitrogen content in the study area was obtained.The nitrogen content in apple canopy was retrieved by using domestic GF-1 satellite images,which provided theoretical basis and technical support for rapid monitoring of nutrient elements in large area of regional fruit trees.