Inversion of Soil Water-Soluble Salt Ion Content in Saline-Alkali Soil Based on Sentinel-2 Satellite Imagery and Soil Variables

This study aims to explore the feasibility of estimating the content of water-soluble soil ions by combining multispectral remote sensing technology with soil physicochemical properties.The research area is situated in the saline soil regions of southern Xinjiang,where the concentrations of major water-soluble cations and anions (K⁺、Na⁺、Ca²⁺、Mg²⁺、HCO₃^-、Cl^-、SO₄^2-) were measured.Machine learning algorithms such as Random Forest （RF）,Gradient Boosting Regression （GBR）,and Extreme Gradient Boosting （XGBoost） were employed to construct soil ion content inversion models based on remote sensing spectral features and soil information.Additionally,the study compared the estimation accuracy of models incorporating soil variables with those that did not.Results indicate that when only multispectral remote sensing data were used as input variables,all three models could only differentiate between high and low levels of soil ion content,with limited ability to accurately estimate the concentrations of individual ions.Incorporation of soil variables into the models significantly enhanced estimation accuracy.Among the methods used,the RF model exhibited the highest prediction accuracy,followed by XGBoost,and GBR had the lowest accuracy.Regarding the estimation of specific ions,the concentrations of Mg²⁺,Ca²⁺,and Na⁺were predicted with relatively high precision and model performance was stable;SO₄^2-,Cl^-,and K⁺showed moderate performance with quantitative prediction capabilities;whereas HCO₃^-content estimation was only feasible to a certain extent with the GBR model.Optimal models varied for different ions,with the RF model providing the best inversion results for K⁺,Mg²⁺,and Cl^-;the XGBoost model excelling in the inversion of Ca²⁺,Na⁺,and SO₄^2-;and the GBR model performing well for HCO₃^-inversion.Notably,the optimal relative analysis errors for Mg²⁺,Ca²⁺,and Na⁺content estimation were 2.829,1.951,and 1.870,respectively,indicating that these models are highly reliable for estimating the concentrations of these ions.The findings of this study provide a scientific reference for the regional-scale estimation of major ion concentrations in soil salinity within arid regions.