基于野外实测高光谱数据的盐碱土盐分含量估测研究

Estimation of Salt Content in Saline Alkali Soil Based on Field-derived Spectra of Salinized Soils

利用高光谱野外实测数据,实现对黄河三角洲盐碱土盐分含量SSC快速、准确定量反演,对于进一步治理盐渍土具有至关重要的意义。通过SVC HR-768i型光谱仪野外实测了黄河三角洲地区盐碱土高光谱数据,在对野外实测高光谱数据进行一阶微分和连续统去除处理的基础上,获取了与SSC相关性较高的光谱波段;分别运用偏最小二乘回归PLSR和多元逐步线性回归SMLR方法构建了SSC的高光谱定量反演模型,建模及分析结果表明:基于一阶微分获取的响应SSC敏感波段,采用PLSR方法得到的定量反演模型具有较好的稳定性和预测精度。本研究采用的方法和取得研究结果对野外实际环境中,进行大面积盐碱土SSC的快速准确定量估测提供了科学依据,具有重要的应用价值。

The quick and accurate quantitative inversion of soil salt content （SSC） of saline alkali soil in the Yellow River Deha by using hyperspectral field data is of vital significance for the further treatment of salinized soil. In this study, continuum removed reflectance （CR） and first derivative spectral （FD） were extracted based on the hyperspectral field data of saline alkali soil in Yellow River Delta fetched through optical spectrometer SVC HR-768i to obtain spectral bands with higher correlation to SSC. Partial least squares regression （PLSR） and stepwise multiple linear regression （SMLR） were respectively used to construct hyperspectral quantitative inversion models of SSC. The modeling and analysis results indicate that the PLSR model of SSC sensitive bands based on the first derivative spectral is better in both stability and prediction accuracy. The proposed method and the results are of great practical value since they can provide scientific reference for the quick and accurate estimation of SSC of large area saline alkali soil in field environment.