Study On Identification Method Of Total Nitrogen Content In Degraded Grassland Soil Based On Hyperspectral

In recent years,due to the continuous interference of factors such as human development and climate change,more than half of the natural grasslands in my country have been degraded to varying degrees,and grassland degradation has become one of the most serious ecological and environmental problems.Grassland degradation usually manifests as grassland vegetation degradation and grassland soil degradation.Although soil degradation is slower than vegetation degradation,soil degradation has a far greater impact on the function of the entire grassland ecosystem than vegetation degradation.At present,soil nutrients are used as a monitoring item for evaluating the degree of grassland degradation in my country.Traditional soil nutrient monitoring mainly uses laboratory chemical measurement.Although the accuracy is high,it is time-consuming and labor-intensive,and it is difficult to meet the needs of obtaining large-scale soil nutrient information.In order to monitor the total nitrogen content of the degraded grassland soil,this thesis selected the grassland soil of Inner Mongolia Autonomous Region as the research object.The Gaia Sky-mini hyperspectrometer was used to collect the hyperspectral data of the soil samples under the condition of natural light,and the total nitrogen content of the soil was determined by chemical analysis.test.The model relationship between soil total nitrogen content and soil spectrum in degraded grassland was studied by combining various spectral data preprocessing and feature band screening methods.The main research conclusions are as follows:(1)Different preprocessing methods have different effects on the accuracy of the model.Through comparative analysis,the partial least squares regression model using the first-order differential preprocessing method has the best modeling effect,the coefficient of determination of the calibration set is 0.835,and the RMSE of the calibration set is 0.090,the coefficient of determination of the prediction set is 0.836,the RMSE of the prediction set is 0.181,and the RPD is 1.592,which can be applied to the identification of grassland degraded soils.(2)After the modeling results of different feature band screening methods,it can be seen that the partial least squares regression model after the CARS feature band screening method has the best effect.The coefficient is 0.859,the prediction set RMSE is0.157,and the RPD is 1.978,which can be applied to the identification of grassland degraded soil.(3)Judging from the modeling effects of comprehensive preprocessing and feature band screening,partial least squares regression is more suitable for establishing the soil total nitrogen model in the study area.The RMSE is lower than that of the SVM and RF models,and the RPD is greater than the other two models.The model,all indicators show that the use of PLSR to establish the soil content model in the study area has a good effect.In this study,using grassland soil hyperspectral data,a variety of preprocessing methods and feature screening methods were used to establish support vector machines,partial least squares regression,and random forest estimation models for total nitrogen content.Nutrient information provides effective theoretical basis and data support.