开都河流域下游绿洲盐渍化土壤高光谱特征

Spectral characteristics of salinized soil in the lower reaches of Kaidu river basin

土壤光谱反射特性是土壤遥感的物理基础。通过野外调查采样、土壤盐分实验分析与土壤高光谱数据采集,对土壤高光谱数据一阶和二阶导数微分变换处理,分析土壤样品的光谱特征,建立土壤光谱和土壤盐分含量间的相关关系,对研究区盐渍化土壤含盐量进行定量反演。研究结果表明：1）从土壤光谱反射率的形态特征来看,土壤的光谱反射率曲线总体上变化较为平缓,光谱特征形态较为相似,且基本平行。2）研究区土壤光谱反射率曲线的形状大致可由300-600nm、600-800nm、800-1000nm、1000-1400nm、1400-1900nm、1900-2100nm、2100-2500nm七个折线段和560nm、900nm、1400nm、1900nm、2200nm五个特征吸收点来控制。3）利用光谱反射率一阶导数微分的盐渍化土壤含盐量多元线性回归预测模型的预测效果均优于利用反射率原型和反射率二阶导数微分,其中氯化物-硫酸盐型RMSE=0.33,硫酸盐型RMSE=0.31,硫酸盐-氯化物型RMSE=0.22。

Reflectance spectra characteristics of soil was the physical basis of soil remote sensing,analysis of reflectance spectra characteristics of soil,the relationship between properties and the spectrum of soil were the important methods for quantitative remote sensing monitoring and extracting soil information. The correlation coefficients between soil salt content and different transform of reflectance spectra were analyzed,and the multiple linear regressions were used to build model for forecasting the soil salt content in study area. The results showed that： 1） Characteristics of the spectral curves were similar in shape,and the change of the curves was gentle. 2）The reflective curves of the reflectance spectra were controlled by seven fold lines（ 300 ~ 600 nm,300 ~ 800 nm,800 ~ 1000 nm,1000 ~ 1400 nm,1400 ~ 1400 nm,1900 ~ 2100 nm and 2100 ~ 2500nm） and five absorbing points（ 560 nm,900nm,1400 nm,1900nm and 2200nm）. 3） The first derivative spectral reflectance was used to build multivariate linear regression soil salt content prediction model,which forecast result was better than reflectance spectra prototype and the second derivative of differential reflectance spectra： for the type chloride-sulfate R2= 0. 91,RMSE = 0. 33; for the type sulfate R2= 0. 85,RMSE = 0. 31,for the type sulfate-chloride R2= 0. 90,RMSE = 0. 22.