Estimation Of Nitrogen Content In Wheat Leaves Based On Different Pretreatment Methods

Precision agriculture is an information-based agricultural management system.It uses sensors and modern advanced monitoring technologies to provide complete,accurate,and timely understanding of detailed land data and crop data,combined with accurate spatio-temporal statistical analysis,and timely and promptly Make the right decision.The mathematical model established by using hyperspectral data and different nutrients(nitrogen,phosphorus,potassium,etc.)can be used to estimate the supply status of crop nutrients.A large number of studies have shown that the precision of monitoring the nitrogen content of crops through remote sensing is superior to the accuracy of monitoring other nutrient elements of crops.Hyperspectral data preprocessing method has a very important position in the processing of hyperspectral data.Effective preprocessing methods can reduce or even eliminate hyperspectral irrelevant information(such as sample background,electrical noise,and stray light).The impact on data provides a more reliable and effective source of data for subsequent analysis based on hyperspectral data.As a very important food crop in China,it is particularly important to use hyperspectral technology to monitor its growth status and production status in a timely and accurate manner.It is also a very important part of the management of precision agriculture to achieve wheat growth.Nitrogen is a very important nutrient for wheat growth,yield,and quality of wheat.Increasing the availability of nitrogen can increase the yield of wheat crops in a relatively short period of time.At present,the country with the largest nitrogen fertilizer consumption in the world is China.Excessive application of nitrogen fertilizer not only affects the quality and yield of wheat,but also causes a series of environmental pollution problems.Therefore,formulating a scientific nitrogen application plan to increase the efficiency of fertilizer use is not only beneficial to wheat growth and yield increase,but also can protect the ecologicalenvironment,reduce environmental pollution,and then realize the common development of economic benefits,social benefits,and ecological benefits.In this study,the winter wheat at heading date was used as a research object.Field measurements and sampling experiments were used as a means to obtain canopy spectral data of winter wheat at the heading stage through field measurements.Nitrogen content of wheat leaves was obtained through chemical means in the laboratory,and wheat crowns were fully analyzed.The relationship between layer spectral data and nitrogen content of wheat leaf was screened out in the sensitive band.The classical least squares method was used to establish the nitrogen content estimation model for winter wheat at heading stage.The model was then verified and optimized.The main results of the study are:(1)After the data is pre-processed,the correlation with the leaf nitrogen content of wheat will be correspondingly increased,indicating that the pretreatment has a very important role in the processing and application of spectral data,and can reduce or even eliminate the hyperspectral irrelevant information.Can provide a more reliable and effective data source for subsequent analysis based on hyperspectral data;(2)After comparisons after various pretreatments,it was found that the model with second-order differential spectra was the most accurate,indicating that the second-order microdivision was the optimal spectral pretreatment;at the same time,multiple radiation corrections and wavelet transforms(bior1.3).The fitting coefficient of the inversion model established by the subsequent spectral data also reached 0.8 or more,the adaptability to the sample was relatively good,the prediction result was relatively accurate,and the model accuracy was relatively high;(3)Wavelet analysis As a kind of spectral data preprocessing method,it has multiple mother functions,and it is necessary to select an appropriate mother function for analysis based on the actual situation.In this study,the final selection was made by comparing various mother functions.The wavelet type is bior1.3;(4)Spectral smoothing and first-order derivative transformation As a general method of hyperspectral preprocessing,the model established in this study is not accurate,probablybecause the influence of background noise is too large,which affects the accurate use of data and reduces the The accuracy of the model.(5)After the preprocessing of the mean spectral centering and data standardization of the original spectral data,the accuracy of the inversion model will be greatly improved,and the increase range is almost the same,but it is not applicable to this study.The reason may be that the data In the acquisition process,other influences were too great,and the accuracy of the original data was reduced.This caused an error and interfered with the experiment.