土壤铜污染光谱特征波段分析、界限浓度划分及含量预测  

A Spectral Characteristic Band Analysis, Threshold Concentration Division, and Content Prediction of Cu Pollution in Soil

作　　者：付萍杰 杨可明 FU Pingjie;YANG Keming(State Key Laboratory of Coal Resources and Safe Mining/China University of Mining&Technology(Beijing),Beijing 100083,China)

机构地区：[1]中国矿业大学(北京)/煤炭资源与安全开采国家重点实验室,北京100083

出　　处：《生态环境学报》2018年第7期1276-1284,共9页Ecology and Environmental Sciences

基　　金：煤炭资源与安全开采国家重点实验室2017年开放基金课题(SKLCRSM17KFA09;SKLCRSM17KFB04);国家自然科学基金项目(41271436)

摘　　要：土壤重金属污染高光谱预测的大部分研究集中在重金属浓度的预测上,但是预测模型精度不高,或精度较高而均方根误差较大。利用土壤Cu污染在538~834 nm波段区间的光谱数据检测土壤Cu污染的界限浓度及光谱特征波段区间,并根据界限浓度建立土壤Cu浓度的预测模型。先对土壤Cu污染光谱数据进行一阶导数处理,利用一阶导数距离检测土壤Cu污染的界限浓度及光谱特征波段;再利用经验模态分解(Empirical Mode Decomposition,EMD)和多重分形理论预测土壤Cu污染浓度,旨在为土壤重金属污染的高光谱研究提供依据,精确土壤重金属污染光谱的研究范围,为利用界限浓度建立土壤重金属浓度预测模型提供新思路。结果表明,538~834 nm波段区间土壤Cu污染的界限浓度为120~160μg?g^(-1),根据一阶导数距离最大值和多重分形谱参量将土壤Cu污染划分为60~120μg?g^(-1)和160~740μg?g^(-1)两个浓度区间,其预测土壤Cu污染的特征波段分别为730~760 nm和540~590 nm;根据划分的界限浓度,利用土壤Cu污染光谱多重分形谱参量、538~834 nm波段区间反射率最大值、一阶导数最大值、及实测土壤含水率、实测土壤有机质含量等参数,分别建立20~120μg?g^(-1)和160~740μg?g^(-1)两个浓度区间土壤Cu含量MLR预测模型,R2在0.99以上,RMSE最大为47.3。因此,多重分形谱参量Δα值和一阶导数距离最大值可作为土壤Cu污染界限浓度划分和光谱特征波段提取的依据,其界限浓度为120~160μg?g^(-1),光谱特征波段区间为730~760 nm和540~590 nm;该文构建的MLR模型可有效预测土壤Cu浓度。Although great achievements have been made in the hyperspectral prediction of heavy metal soil pollution,the general prediction model parameters have included the actual measured content,and the spectral absorption features of various substances,but the prediction models have displayed low accuracy.Therefore,in this study,the spectral data from the 538~834 nm band of soil Cu would be used to detect the boundary concentration and spectral characteristic bands of soil Cu pollution,and prediction models of soil Cu concentration based on the limit concentration would be established.Firstly,the first derivative of the soil Cu pollution spectral data was used to detect the boundary concentration and spectral characteristic bands of soil Cu pollution using the first derivative distance;then the soil Cu pollution concentration was predicted by empirical mode decomposition(EMD)and multifractal theory.This study was to provide a basis for the hyperspectral study of soil heavy metal pollution,reduce the scope of research on soil heavy metal pollution spectra,provide a new way to establish a prediction model of soil heavy metal concentration using threshold concentration.This study showed that the threshold concentration of the Cu pollution in the band interval 538 to 834 nm was 120~160μg?g-1.In accordance with the maximum value of the first-order derivative distance and multi-fractal spectrum parameters,the Cu pollution of the soil was divided into two concentration intervals:60~120μg·g^-1 and 160~740μg·g^-1.Furthermore,the characteristic bands of the Cu pollution of the soil were predicted to be from 730 to 760 nm,and 540 to 590 nm.Then,in accordance with the division of the threshold concentration,a soil Cu content MLR prediction model was established for the concentration intervals of 20~120μg·g^-1,and 160~740μg·g^-1,respectively.In accordance with the multi-fractal spectrum parameters of the soil Cu pollution spectrum;maximum reflectivity in the band interval from 538 to 834 nm;maximum value of the first-order

关 键 词：土壤铜污染 光谱特征波段 界限浓度 土壤铜含量预测 

分 类 号：X122[环境科学与工程—环境科学]