摘要
文章以恩施地区土壤重金属As、Cd、Cr、Pb、Cu、Zn为研究对象,利用ASD地物光谱仪采集土壤光谱数据,在光谱预处理和相关分析的基础上,采用偏最小二乘回归(PLSR)方法,构建土壤重金属含量与光谱的反演预测模型。结果表明,平滑处理、一阶导数、去趋势、归一化处理散射校正(SNV)4种光谱处理方法组合使用可以有效地提高模型反演精度;通过对PLSR模型RMSEC、RMSECV、RMSEP、R_(Cal)^(2)、R_(CV)^(2)、R_(Pred)^(2)、6种参数分析发现,As、Pb、Cu相比Cd、Cr、Zn在PLSR模型中的测误差值较小且模型的相关性较高,说明模型可以预测As、Pb、Cu 3种元素,该研究为后期采用高光谱遥感影像数据快速监测恩施地区土壤重金属污染情况提供了技术参考。
Taking As,Cd,Cr,Pb,Cu and Zn as the research objects,the soil spectral data were collected by ASD surface object spectrometer.Based on spectral preprocessing and correlation analysis,partial least squares regression method was used to construct the inversion and prediction model of soil heavy metal content and spectrum.The results show that the combination of smoothing processing,first derivative,detrend and normalized scattering correction can effectively improve the inversion accuracy of the model.Based on PLSR model,analysis of six parameters inclduing RMSEC,RMSECV,RMSEP,found that compared with Cd,Cr and Zn,the measurement error of As,Pb,Cu is smaller and the correlation is higher,showing that the model can predict the three elements of As,Pb,Cu.This study would provide a technical reference for using hyperspectral remote sensing image data to rapidly monitor soil heavy metal pollution in Enshi,Hubei Province.
作者
方臣
匡华
周小娟
陈曦
万翔
刘烨青
FANG Chen;KUANG Hua;ZHOU Xiaojuan;CHEN Xi;WAN Xiang;LIU Yeqing(Hubei Geological Survey Academy,Wuhan 430034,China)
出处
《环境科学与技术》
CAS
CSCD
北大核心
2021年第9期154-159,共6页
Environmental Science & Technology
基金
湖北省地质局科技项目(KJ2018-9,KJ2020-25,KJ2021-5)。
