土壤铅含量光谱反演模型研究

Study on spectral inversion model of soil lead content

使用ASD FieldSpec 4(350~2 500nm)地物光谱仪对采自陕西省扶风县、杨凌县、武功县的44个土壤样品进行光谱反射率测定。结合NOR、MSC、SNV预处理后分别进行一阶微分、二阶微分、反射率倒数对数变换,采用偏最小二乘回归法建立Pb元素的高光谱反演模型。化学分析结果表明:该研究区域存在较为严重的Pb污染现象,Pb含量接近临界值。模型结果表明:反射光谱经过NOR、MSC、SNV变换后显著提高了反射光谱的信噪比,结合微分变换有助于提高土壤中的重金属元素信息,使用相关性较高的组合波段能显著提高模型的稳定性和预测能力;采用PLSR法建立的Pb元素光谱最优模型的建模精度分别为0.618 2;不同元素采用不同的处理方法建立的最优估算模型稳定性较好、精度较高,能够实现该区域重金属元素Pb含量的快速预测。

The spectral reflectance of 44 soil samples collected from the country of Fufeng,Yangling,and Wugong of Shaanxi Province was measured by the ASD FieldSpec 4(50~2 500 nm).After pretreatment with NOR,MSC and SNV,the first order derivative,the second order difference and the reciprocal of logarithm of reflectance were used respectively,and the hyperspectral inversion model of Pb element was established by partial least squares regression.Chemical analysis showed that there was a serious phenomenon of Pb pollution in the study area,and Pb content approached the critical value.The results of the model show that:The reflectance spectra of the reflectance spectra significantly increase the signal-to-noise ratio of the reflectance spectrum after being transformed by NOR,MSC and SNV.Combined with the differential transformation,it helps to improve the information of heavy metal elements in soils.Significantly improve model stability and predictive power.The modeling accuracy of the optimal model of Pb element spectrum established by PLSR method is 0.618 2,respectively.The optimal estimation models established by different treatment methods of different elements have better stability and higher precision,and can quickly predict the Pb content of heavy metals in this area.