【摘要】 【目的】探索基于光谱指数建立土壤盐分离子含量的估算模型,为研究区土壤盐分离子快速遥感监测提供技术参考。【方法】以博斯腾湖湖滨绿洲为研究区,分析土壤表层盐分离子Cl^-、Ca²⁺、HCO^-₃、Mg²⁺、Na⁺+K⁺、SO₄^2-以及EC、TS含量变化特征;针对不同光谱指数,筛选特征波段,采用偏最小二乘回归建立高光谱估算模型。【结果】（1）土壤表层盐分离子的原始光谱反射率与土壤表层盐分离子Cl^-、EC、Na⁺+K⁺、Ca²⁺含量呈极显著相关（P<0.01）;（2）基于光谱指数SI和BSI的相关系数图具有相似性,RSI、DSI、NDSI的光谱矩阵系数图具有相似性,RSI、DSI、NDSI的建模效果优于SI和BSI,土壤表层盐分离子的特征波段集中在近红外波段,最佳估算波段在1562～2354 nm;（3）土壤表层Na⁺+K⁺盐分离子含量的高光谱估算模型Y=0.67+73.56x₁₈₀₉-75.75x₁₉₄₂,建模集R²=0.96,RMSE=0.38,验证集R²=0.82,RMSE=0.64。【结论】通过光谱指数筛选特征波段且利用偏最小二乘回归方程构建的土壤盐分离子模型发现,含量较高的离子估算效果较好,尤其是Na⁺+K⁺、Cl^-、SO₄^2-含量。更多还原

【Abstract】 【Objective】The present paper aimed to explore the establishment of an estimation model of soil salt soil content based on the spectral index, and a technical reference for rapid remote sensing monitoring of soil salt soil in the study area was provided.【Method】Taking the lakeside oasis on the west bank of Bostan lake as the study area, the content variation characteristics of the surface soil salt ions Cl^-, Ca²⁺, HCO^-₃, Mg²⁺, Na⁺+K⁺, SO₄^2- as well as EC and TS were analyzed. According to different spectral indices, the characteristic bands were screened, and the hyperspectral estimation model was established by partial least square regression.【Result】（i） The original spectral reflectance of soil surface salt ions was significantly correlated with the contents of soil surface salt ions(Cl^-,EC,Na⁺+K⁺,Ca²⁺)（P<0.01）.（ii） The correlation coefficient maps based on the spectral index SI and BSI were similar, and the spectral matrix coefficient diagrams of RSI, DSI, and NDSI were similar. The modeling effect of RSI, DSI, and NDSI was better than that of SI and BSI. The characteristic bands of the surface soil salt ions were concentrated in the near-infrared band, and the best estimated band ranged from 1562 to 2354 nm.（iii）The hyperspectral estimation model of the surface soil Na⁺+K⁺ salt ion content was Y=0.67+ 73.56x₁₈₀₉-75.75x₁₉₄₂, the modeling set R²=0.96, RMSE=0.38, the verification set R²=0.82, RMSE=0.64. 【Conclusion】The soil salt ion model, which was based on the partial least square regression equation and selected by spectral index, showed that the ion with higher content has better estimation effect, especially the content of Na⁺+K⁺,Cl^-,SO₄^2-.更多还原