Inverting Soil Salinity By Multiple Information Fusion And Relative Equipment Technology

Soil salinization is a common phenomenon of soil degradation,soil salinization situation is very universal especially in arid and semi-arid areas in the west part of China,and severely disrupt local ecological environment.So how to effectively detect soil salinization is meaningful for the agriculture in the arid areas.There are lots of methods to detect soil salinization,among them,the remote sensing method has been developed rapidly in the field of soil salinization detection due to its advantages of rapid measurement,convenient portability and easy operation.However,most of the research based on remote sensing technique are preliminary qualitative discussions,and not considering the complex coupling relationship between soil salinity and water content,so it is impossible to achieve the desired results,and has not been applied to practical applications.This paper analyzes correlation mechanism between soil spectral information/permittivity and soil salinization,research the inversion effect of SVM under incomplete information conditions.The optical remote sensing method has many problems,such as high cost,insufficient mechanism research and external factors,so the degree of soil salinization cannot be quantitatively measured.This paper designs a realtime detection system of soil water and salt content based on soil dielectric properties,explores the relationship between soil water/salt content and amplitude/phase difference of induced signal,and establishes an empirical formula for detecting soil salinization and drought degree,in order to realize quantitative measurement of soil salinization.The research and conclusion of this paper are as follows:(1)In the experiment of inverting soil salinization by optical remote sensing,the influence components of soil spectral curve and the sensitive bands related to soil water and salt are analyzed.In view of the characteristics of nonlinear characteristics of soil spectral curves and inregularity,the traditional nonlinear fitting model BP neural network is improved,and the SVM regression method is constructed to verify the applicability of the inversion of soil salinization algorithm.The SVM model has a very harsh theoretical basis for Mathematics,the effect of the small sample experiment is quite outstanding,the fitting precision can be raised to 99.4119%,mean square error is 9.253%,the result of the prediction is very good.(2)In the experiment of designing real-time detection system for soil water and salt content,sine wave signal generator module,sensor module and signal amplification module are designed respectively.Through the oscilloscope simulation,the signal generator module can output the excitation signal and the frequency offset is within 5%.The probe and signal processing module can correctly handle the detection signal.The signal amplification detection module can detect the amplitude difference and phase difference of the polarization signal.(3)In the experiment of dielectric constant inversion,the correlation between soil permittivity and soil water/salt content is analyzed,the correlation between the amplitude difference and the degree of soil salinization is 0.536,the correlation between the amplitude difference and the soil drought degree is 0.68,all reached significant correlation;The correlation between the phase difference of induction signal and the soil salinization is 0.147,the correlation between the phase difference of induction signal and the soil drought degree is 0.976.Using multiple linear regression method,the mapping equation of soil drought degree and amplitude difference /phase difference ? is established,and the accuracy is 95.8%.By observing the detection data,the empirical equation of soil salt content and amplitude difference /phase difference ? is established.Under the requirement of soil salinization degree classification precision,the 75% prediction precision is realized,the requirements for the measurement of soil salinization degree is met.Comparison of two kinds of soil salinization detection methods,the advantage of optical remote sensing is high precision and wide range of detection,but it's computation is complex,applicability is not high,and is easily affected by external factors.The advantages of the dielectric constant inversion method are simple operation,no interference and quantitative measurement.,but the detection accuracy is not high enough.In the detection process of soil salinization,two methods can complement each other,the dielectric constant inversion method is used in the area with low requirement for soil salinization accuracy;If particular area need high precision detection,optical remote sensing detection matrix model can be customized to realize high precision measurement.