The Land Salinization Remote Sensing Inversion Of Zhalong Wetland Based On The Support Vector Regression Algorithm

Due to the influence of natural and human factors,the land salinization in Zalong Wetland Nature Reserve is becoming more and more serious,which affects the regional ecological environment stability and sustainable development.Wetland as an important ecosystem,for the migration of migratory birds,soil and water conservation has an important role,therefore,for the wetland salinization research is even more urgent.Remote sensing provides effective technical means for soil salinity monitoring,and it is important to dig out the salinity information of salinity soil in remote sensing image,which is of great significance to effectively control land salinization,rational exploitation and utilization of saline soil and maintain ecological sustainable development.Zhalong wetland is located in the western part of Heilongjiang Province,the Songnen Plain,Wuyuer River and Shuangyang River Basin(referred to as Wu Shuang basin)downstream of the valley.In this paper,Landsat8 remote sensing images were used to obtain the spectral data of bare saline soil in Wuzao watershed.The soil sampling points were selected,the soil electrical conductivity and p H were measured,and the soil samples were analyzed by chemical analysis data.Firstly,the spectral spectrum of saline-alkaline soil was calculated by the correlation analysis of saline-alkaline soil spectrum,the correlation between saline-alkali reflectance and salinity,and the spectral diagnosis index.The spectrum of susceptibility to salinity soil was obtained: band 5,band 6,band 7;(SVR)model,the spectral reflectance of the 5,6,7 bands of remote sensing images and their transformation are also based on the support vector machine regression(SVR)model.The spectral reflectance of the remote sensing images is analyzed by geostatistical analysis.The results show that the SVR model can be used as the independent variable of the SVR model,and the soil salinity is taken as the dependent variable of the SVR model.The salinity of the saline soil is retrieved and the results of the inversion are compared with the field measurements.And the accuracy of remote sensing inversion of land salinization was improved.Finally,the present model was used to invert the classification of saline-alkaline land in Zhalong wetland in 2016.In this paper,we try to use the support vector machine regression model to study the salinity of saline soil in Zhalong wetland,and hope to make a useful exploration for quantitative estimation of soil salinity.The research results can be used as reference materials in the fields of land survey and environmental protection,and provide relevant decision-making support to the relevant departments in the prevention and control of salinization in Zhalong wetland.