粮食主产区耕地土壤重金属高光谱综合反演模型

发布时间：2018-06-27 04:31

本文选题：耕地 + 土壤重金属　；参考：《农业机械学报》2017年03期

【摘要】：粮食主产区的耕地土壤重金属污染已成为影响粮食安全和人居环境安全的突出问题。高光谱遥感技术为快速高效获取土壤重金属含量提供了新的途径,也为土壤总金属污染的监测和防治提供了技术保障。以河南省粮食主产区新郑市为研究对象,共采集191个耕地土壤样品,利用Rank-KS法划分为144个建模集样本和47个验证集样本;在室内利用ASD Field Spec 3型地物光谱仪获取土壤高光谱数据,对耕地土壤样品在400~2 400 nm的光谱反射率与Cr、Cd、Zn、Cu、Pb 5种重金属元素进行相关性分析,筛选出5种重金属均通过P=0.01显著性检验的共用高光谱特征波段作为反演模型的自变量;采用基于OLS的固定影响变系数面板数据模型,对新郑市144个建模集样本的5种土壤重金属面板数据构建高光谱综合反演模型。结果表明:面板数据模型整体显著,拟合优度较高(R~2=0.993 7,F统计量为1 365.94)。模型精度检验Cu的相对分析误差为2.046,Pb的相对分析误差为3.432,都具有较好的预测精度;Cr、Cd、Zn的相对分析误差在1.4~1.8之间,具有一般的定量预测能力。面板数据模型通过一次建模综合反演多种土壤重金属,计算简便、速度快,可以用于新郑市耕地土壤重金属的高光谱快速监测。
[Abstract]:Heavy metal pollution in cultivated land in the main grain producing areas has become a prominent problem affecting food security and human settlement environment security. Hyperspectral remote sensing technology provides a new way to obtain soil heavy metal content quickly and efficiently, and also provides technical guarantee for monitoring and prevention of soil total metal pollution. A total of 191 cultivated land soil samples were collected from Xinzheng City, a major grain producing area of Henan Province, and were divided into 144 modeling and 47 validation samples by Rank-KS method. Soil hyperspectral data were obtained by ASD Field Spec 3 ground object Spectrometer in laboratory. The correlation between spectral reflectance of cultivated soil samples at 400 ~ 2 400 nm and 5 heavy metal elements, Cr ~ (2 +) CD ~ (2 +) Zn ~ (2 +) Cu ~ (2 +) Pb, was analyzed. Five kinds of heavy metals were selected as independent variables of the inversion model by using the common hyperspectral characteristic bands tested by P0. 01 significance test, and the fixed influence variable coefficient panel data model based on OLS was used. The hyperspectral comprehensive inversion model was constructed from 5 soil heavy metal panel data from 144 modeling samples in Xinzheng city. The results showed that the panel data model was significant as a whole and had a high goodness of fit (RX20.993 7F statistic was 1 365.94). The relative analysis error of Cu is 2.046% Pb and the relative analysis error of Cu is 3.432.It has good prediction accuracy and the relative analysis error of Cr ~ (2 +) CD ~ (2 +) Zn is between 1.4 ~ 1.8, and it has general quantitative prediction ability. Panel data model can be used for hyperspectral rapid monitoring of heavy metals in cultivated land in Xinzheng City.
【作者单位】：河南理工大学测绘与国土信息工程学院;河南科技学院生命科技学院;华北水利水电大学资源与环境学院;
【基金】：国土资源部公益性行业科研专项(201411022-2)
【分类号】：S151.93

【相似文献】