Remote Sensing Modelling And Application For Nitrogen Nutrition Index Of Winter Wheat

With the development of remote sensing,hyperspectral remote sensing has unparalleled advantages at acquiring crop information.Nitrogen plays an important role in the growth of the crops because enough nitrogen nutrition guarantees for high-yield.So the peasants often apply excessive nitrogen fertilizer to ensure the harvest.However,the luxury nitrogen did reduce the economic benefits and pollute the environment,seriously.Therefore,it is important to detect the crop nitrogen nutrition,accurately,and apply appropriate supplement nutrient to ensure the crop's growth.Many studies showed that nitrogen nutrition index(NNI)could be used to detect crop nitrogen nutrition accurately.The empirical model inversing the NNI with the spectral indices directly(NNId)lacks of agronomy mechanism.And the indirect NNI(NNIid),the above ground biomass(AGB)and plant nitrogen concentration(PNC)was estimated first,has many inversion steps which leads to cumulative error.This dissertation is committed to propose a better way to reduce the cumulative error with the remote sensing data.Grain yield and nitrogen requirement can be estimated based on the results of this method.In this study,the objective of this work is to predict the NNI by coupling the remote sensing data and the steps to construct the agronomy NNI model,and analysis the spectral indices relation to AGB(n=15)and PNC(n=15)constructed by ASD hyperspectral data.Agronomy parameters,AGB and PNC,were submitted by corresponding spectral index to construct the remote sensing critical nitrogen dilution curve(Nc_RS),and the remote sensing NNI(NNIRS)can be acquired by the ratio of the corresponding PNC index and NC_RS.Then NNIRS and measured NNI were compared to select the optimal combination of the spectral indices.At last,grain yield and nitrogen requirement model based on NNIRS were developed and different years' data were used to calibrate and validate.The conclusions were described as follows:1)the combination of the mSR705 and WI performed best to calculate NNIRS,where an NC_RS equation can describe as NC_RS=3.59WI-4.66(R2=0.75).The NNIRS based on the above NC_RS was in best agreement with the observed NNI,with R2 value of 0.73 and root mean square error(RMSE)of 0.128.Estimated NNI accuracy of using NNIRS was better than NNIid method(R2=0.66,RMSE=0.216)and NNId model(R2=0.66,RMSE=0.132),the results indicate that NNIRS model proposed in this study could diagnose the nitrogen nutrition of winter wheat,and the hyperspectral image data verified it.2)the results showed that the model based on NNIRS data can be used to predict the grain yield of winter wheat.Among the single stage models,the model of anthesis stage had the highest R2 value,R2=0.70,and the booting stage model was stable and had a better prediction ability with an R2 value of 0.60,RMSE value of 1157 kg ha-1.The yield estimation model with the whole stages had an R2 value of 0.64 and RMSE value of 1377kg ha-1.3)The requirement of nitrogen varied significantly with the change of nitrogen application.The proposed nitrogen requirement models based on NNIRS data could be used for guiding fertilization precisely except the model based on the jointing stage data.