基于PLSR的陕北土壤盐分高光谱反演  

Hyperspectral remote sensing inversion of soil salinity in north Shaanxi based on PLSR

作　　者：李晓明[1,2] 王曙光[1,2] 韩霁昌[1,2] 

机构地区：[1]陕西省土地工程建设集团,西安710075 [2]国土资源部退化及未利用土地整治工程重点实验室,西安710075

出　　处：《国土资源遥感》2014年第3期113-116,共4页Remote Sensing for Land & Resources

基　　金：陕西省自然科学基础研究计划(编号:2012JQ5015)

摘　　要：选取陕北盐渍土为研究对象,通过采集高光谱数据及土壤样品测定,研究土壤盐分含量与反射率之间相关性,遴选盐分特征波段,利用常规回归分析及偏最小二乘回归分析建立土壤盐分的定量反演模型,并利用检验样点进行对比分析和精度检验。研究结果表明,482 nm,1 365 nm,1 384 nm,2 202 nm及2 353 nm为土壤盐分含量的特征波段,利用高光谱数据进行盐分定量反演具有良好的精度;精度检验结果表明,通过Matlab进行偏最小二乘回归计算的反演模型,实测值与预测值相关性更好,精度较高。The salinized soil in northern Shaanxi Province was chosen as the study object. The hyperspectral data were collected and the soil samples were analyzed. First, the correlation between the soil salinity and the reflectance were analyzed, and the characteristic bands were fitted. The usual regression and partial least squares regression （ PLSR） analysis was used to study the inversion model of soil salinity, and some testing samples were used to compare the accuracies. The results show that 482 nm, 1 365 nm, 1 384 nm, 2 202 nm and 2 353 nm are five characteristic wavelengths, and the precision of inversion is satisfactory. The result of precision test indicates that the inversion model with PLSR calculated by Matlab is fairly good, and the correlation between the measured value and the predicted value is better.

关 键 词：偏最小二乘回归( PLSR) 土壤盐分 高光谱反演 

分 类 号：S156.4[农业科学—土壤学] TP79[农业科学—农业基础科学]