基于高光谱的景电灌区盐渍化土壤水、盐估算研究

Estimation of Salt and Water Content in Jingdian Irrigation Area Based on Hyperspectral of Salinized Soils

高光谱定量遥感技术是大尺度盐渍化调查的理想手段之一,但现有研究大多为含盐量单因子的分析,野外水盐同时存在的状态下光谱响应特征尚未明确。本文以甘肃景电灌区盐渍化土壤为研究对象,以土壤含水率、电导率和野外实测光谱数据为信息源进行盐渍土壤的水分、盐分光谱响应特征提取。利用相关系数法筛选敏感波段,并通过多元线性回归法建立估算模型。结果表明:(1)水盐同时存在状态下,12种光谱变换与含水率的相关系数大多高于电导率,水分对盐渍土光谱的影响作用更强烈;(2)导数光谱可以增强光谱曲线的细微变化,本次研究中R′、SR′、logR′和CR′与土壤含水率、电导率均具有较高的相关性;(3)基于特征波段建立的回归模型中,通过反射率一阶导数R′建立的线性回归模型的稳定性和预测精度较好,光谱反射率对于盐渍土含水率、电导率均有一定的指示作用,可以用于土壤盐渍化程度的估测。

Hyperspectral quantitative remote sensing is one of the ideal methods for large-scale salinization survey.However,most of the existing researches are single-factor analysis of salt content.The spectral response characteristics of soil water and salt in the field are not clear.In this paper,the salinized soil in Jingdian Irrigation District of Gansu Province was taken as the research object.The water content,conductivity and field measured spectral data were used as information sources to extract the water and salt spectral response characteristics of saline soil.We use the correlation coefficient method screened sensitive bands for conductivity and water content,and established an estimation model by multiple linear regression.The results showed that:(1)The correlation coefficient between 12 kinds of spectral and water content is mostly higher than that of conductivity,under the condition that water and salt exist simultaneously.For saline soil,the effect of water content on soil spectrum is more prominent;(2)The derivative spectrum can enhance the subtle changes of the spectral curve.In this study,R',SR',LORG'and CR'have high correlation with soil moisture content and conductivity.(3)In the regression model based on the characteristic band,the linear regression model established by the first derivative(R')has stronger stability and higher prediction accuracy,and the spectral reflectance is different.The spectral reflectance can reflect the change of water content and conductivity of saline soil,and can be used to estimate the degree of soil salinization.