Spectral Characteristics And Spatial Distribution Of Saline Soil In Tuoketuo County

Taking Tuoketuo County of Hohhot City as the research area,this thesis selects the saline soil under three land use modes of typical woodland,grassland and wasteland as the research object,explores the spectral characteristics and differences of salinity by using the ground object spectrometer,and carries out the electrical conductivity inversion of saline soil through different modeling methods.On this basis,combined with the measured soil conductivity value and the inversion value of the optimal inversion model,Kriging interpolation was carried out to study the spatial distribution pattern of soil conductivity in the three sample plots.On the other hand,taking the non saline soil collected in the study area as the research object,through the indoor salt control experiment,the spectral characteristics and differences of soil salinity with different salt types and salt contents were explored Modeling inversion is carried out at the same time.The main conclusions are as follows.The results show that:1.The soil conductivity curves of the three plots in the field were consistent with the law that the reflectivity increased with the increase of the conductivity value,and the spectral curves of the wasteland were significantly higher than those of the woodland and grassland due to the high salt content.On the whole,each curve showed a rapid rise at first,then slowed down,and finally showed a downward trend;The results show that the EC inversion model of three sample plots has the following characteristics:machine learning model(BP neural network)>partial least squares model(PLSR)>multiple linear regression model(MLSR);Compared with the determination coefficient and root mean square error of each model,BP neural network is more suitable for model inversion of soil conductivity.2.In the indoor control experiment,different kinds and contents of salt have different effects on the spectral curve of soil.The spectral reflectance of soil treated with Na₂CO₃,Na₂SO₄and Na HCO₃salt solutions increases with the increase of soil salt content(SSC),and the soil reflectance treated with Na Cl solution decreases with the increase of SSC,and the reflectance of each soil sample is higher than that of the untreated soil sample;The correlation between SSC and the original spectral data of four soil samples was as follows:Na Cl>Na₂SO₄>Na HCO₃>Na₂CO₃;It is verified that PLSR model is more suitable for soil salt inversion.3.There are some differences in soil electrical conductivity under different land use patterns in the study area,and the electrical conductivity values of all kinds of land are higher,and the size relationship is as follows:wasteland>grassland>woodland.The sample plot is alkaline soil.The results show that the optimal models of semivariance function for woodland,grassland and wasteland are exponential model,exponential model and Gaussian model respectively,and the coefficient of determination ranges from 0.673 to0.896,and the sum of squares of residuals is small.Kriging interpolation map showed that the spatial distribution of soil electrical conductivity in woodland,grassland and wasteland was not balanced,with arc and patch distribution.The predicted values of soil conductivity obtained by BP model based on Soil Spectrum basically conform to the distribution characteristics of soil conductivity measured in the sample plots,which can be used to retrieve the content of soil conductivity with high accuracy.