矿区植物重金属污染的高光谱特性及其反演模型

植物体内重金属的大量聚集将造成植被叶面反射光谱在特定波段内的特征性变化。以陕西潼关矿区葎草、玉米、苦菜、梧桐典型植被为研究对象,利用ASD Field Spec 4光谱仪采集植物光谱曲线,采用导数光谱、连续统去除法等信息提取方法,分析植物红边位置、红边斜率、植被指数等光谱特征参数与植物样本重金属相关性。选择蓝谷、红边位置等相关性较高的敏感波段,构建矿区两种葎草和梧桐的镉(Cd)、铅(Pb)质量比与其光谱参数定量反演模型。结果表明,葎草、梧桐的Cd、Pb与一阶微分光谱特征相关程度更好,多数符合二次多项式函数关系,然而葎草的Cd符合反函数关系,均方根误差在0.03~0.07。室内试验验证两种植物叶绿素含量与Cd含量负相关,在Pb低浓度时植物的叶绿素含量减少得不明显,随Pb浓度升高,叶绿素含量降低明显。因而植物高光谱特征间接反映植物重金属的含量。

Hyper-spectral features of heavy metal pollutants in vegetables and their inversion model in the mining areas

The paper is aimed at making an exploration and optical analysis of the correlation between the spectral characteristic features (the red edge position,the red edge slope,the vegetation index and some other related spectral parameters) and the content rate of the heavy metal pollution in the sampling plant.This is because it has already become one of the most advanced methods for identifying and determining the content rates of such heavy metal pollutants on the condition that their content rate is lower for its features at the much more spectral bands with the high resolution and noticeable succession and completion.The necessity for the given research comes from the great amount of heavy metal content accumulation in the plants that may account for the vegetation foliage reflectance spectra at the specific band.For example,as compared with the traditional chemical monitoring methods,the hyper-spectral data processing method has been gradually recognized as one of the most advanced methods for identifying and determining the content rates of such heavy metal pollutants.Thus,for the research purpose,we have collected 4 kinds of field samples,with the spectral curves of Humulus,corn,sow thistle and buttonwood via the ASD Field Spec 4 spectrometer in a mining area of Tongguan County,Shaanxi.In proceeding with the given research,we have managed to extract efficient and highly accurate information of the original spectral reflectance changing data by means of the first derivative spectrum (FD) and the Continuum Removal (CR) at first.And,then,we have done an analysis immediately as to the correlation of the spectral features (e.g.the red edge position,the red edge slope,the vegetation index and the other spectral related parameters)and the heavy metal content rates in the samples.As the result of analysis,we have established 2 quantitative inversion correlation models of Cd and Pb content and the spectral parameters in the regard to Humulus and buttonwood based on the high sensitive spectrum bands,i.e.the blue valley and the red edge position.The results of our analysis indicate that the correlation between the Pb-Cd content rate and the first difference proves quite rational,which should belong to the quadratic polynomial functions properly whereas in the inversion function to Humulus,the rootmean square error (RMSE) should be equal to 0.03-0.07.On the other hand,there should exist a negative correlation between Chlorophyll and the content of Cd in accordance with the experimental results.Besides,it would be possible for Chlorophyll to get reduced obviously at the lower content rate of Pb.And,consequently,the hyper-spectral features of the plant can be said a kind of indirect reflection of the heavy metal content situation which may have a broad potential application in predicting and forecasting of its content and the pollution involved.Therefore,the discovery of this paper can be said to provide a scientific proposal for the vegetation restoration and ecological environment protection in the mining areas.