煤矸石充填复垦重构土壤重金属含量高光谱反演

发布时间：2018-02-12 08:09

本文关键词： 煤矸石充填复垦土壤重金属含量高光谱 SMLR PLSR ANN　出处：《光谱学与光谱分析》2017年12期 　论文类型：期刊论文

【摘要】：为研究煤矸石充填复垦土壤重金属含量快速有效的监测方法,以淮南创大生态园煤矸石充填复垦田间试验小区为研究区域,首先采用化学方法监测土壤(0~20cm)重金属(Cu,Cr,As)含量,然后采用ASD(analytical spectral devices)FiSpec4型高光谱仪测量土壤样品的反射光谱,提取光谱特征,并对光谱进行一阶微分变换、二阶微分变换及倒数对数变换;将变换后的各光谱特征参数与监测的土壤重金属含量进行相关性分析,并依据相关性分析结果选择显著相关的波段作为相关因子供建模使用。采用多元逐步回归(stepwise multiple liner regression,SMLR)分析、偏最小二乘回归(partial least squares regression,PLSR)及人工神经网络(artificial neural network,ANN)三种方法分别建立基于光谱反射率估算土壤重金属含量的预测模型,并采用回归模型进行精度评定,然后确定各重金属含量的最佳预测模型。实验结果表明,经过微分变换的光谱波段与土壤重金属含量达到了显著相关;重金属Cu和Cr的一阶微分光谱的人工神经网络模型为最佳预测模型,重金属元素As的二阶微分光谱的偏最小二乘回归模型为最佳预测模型。
[Abstract]:In order to study the rapid and effective monitoring method of heavy metal content in reclaimed soil with coal gangue filling, the chemical method was used to monitor the content of the heavy metal Cu (Cr) in the soil (20 cm) in Huainan Chuangda Ecological Park, taking the field experiment area of coal gangue filling reclamation as the research area. Then, the reflectance spectra of soil samples were measured by ASD(analytical spectral devices)FiSpec4 type high spectrometer, and the spectral characteristics were extracted, and the spectrum was transformed by first-order differential transformation, second-order differential transformation and reciprocal logarithmic transformation. The correlation analysis was carried out between the transformed spectral characteristic parameters and the monitored soil heavy metal content, and the significant correlation bands were selected as the correlation factors according to the results of the correlation analysis. The stepwise multiple liner regression analysis was used for modeling. Three methods, partial least squares regression (PLSR) and artificial neural Network (Ann), were used to establish prediction models for estimating soil heavy metal content based on spectral reflectance, and the precision was evaluated by regression model. The experimental results show that the spectral band of differential transformation has significant correlation with the content of heavy metals in soil. The artificial neural network model for the first order differential spectra of heavy metals Cu and Cr is the best prediction model, and the partial least square regression model for the second order differential spectra of heavy metal elements as is the best prediction model.
【作者单位】：安徽理工大学测绘学院;中国矿业大学测绘科学与技术博士后流动站;
【基金】：国家自然科学基金项目(41472323) 安徽省国土资源科技项目(2012-k-23)资助
【分类号】：O657.3;X833

【相似文献】