农作物在受到重金属污染以后, 会破坏本身的组织细胞结构和叶绿素含量, 从而影响农作物的新陈代谢和健康状况。 人和动物如果食用了污染的农作物以后, 会有致命的伤害。 高光谱遥感目前被广泛应用于监测农作物受重金属污染的程度。 重金属污染下的农作物叶片的光谱变化很微小, 传统的监测方法和常规的光谱特征参数很难将光谱之间的微弱差异区别开, 目前高光谱遥感应用是研究的重点和难点。 通过设置不同浓度的Cu2+和Pb2+胁迫下玉米盆栽实验, 采集玉米叶片的光谱数据、 叶绿素的相对含量以及重金属Cu2+和Pb2+的相对含量。 提出了包络线去除(CR)、 光谱相关角(SCA)、 光谱信息散度(SID)以及正切函数(Tan)和兰氏距离(LD)相结合的LD-CR-SIDSCAtan模型, 将其与传统的光谱测度方法, 如光谱相关系数(SCC)、 光谱角(SA)、 光谱角正切(DSA)、 光谱信息散度-光谱相关角正切(SIDSAMtan)、 光谱信息散度-光谱梯度角正切(SIDSGAtan)和常规的光谱特征参数, 如红边最大值(MR)、 绿峰高度(GH)、 红边一阶微分包围面积(FAR)、 红边一阶微分曲线陡峭度(FCDR)、 蓝边(DB)、 红谷吸收深度(RD)相比较, 验证了该模型的优越性和可行性。 并且将LD-CR-SIDSCAtan模型应用于不同浓度下Cu2+和Pb2+胁迫的玉米叶片的整体波形和子波段的光谱差异信息的测度上。 结果表明, LD-CR-SIDSCAtan模型实现了重金属Cu2+和Pb2+污染的定性分析, 能够测度光谱相关系数达到0.99以上的相似光谱之间的差异信息, 波形差异信息与叶片测得的叶绿素相对含量和重金属Cu2+和Pb2+相对含量显著相关, 也分别找到了重金属Cu2+和Pb2+胁迫下的光谱响应波段。 在测度光谱数据的整个波段区间范围, 模型值为负值时的光谱差异要比模型值为正值更加明显; 在模型值为正值时, 如果数值越大, 光谱的差异性也越大。 因此, 随着重金属Cu2+和Pb2+浓度的增加, 光谱的差异增大, 意味着重金属Cu2+和Pb2+污染程度更为严重; 玉米植株受到重金属Cu2+胁迫污染, 在测度光谱数据的局部子波段区间范围时, “蓝边”、 “红边”、 “近谷”、 “近峰B”处对重金属Cu2+胁迫污染响应特别的敏感, 可以作为监测重金属Cu2+污染程度的有效波段; 当玉米植株受到重金属Pb2+胁迫污染时, 在“紫谷”、 “蓝边”、 “黄边”、 “红谷”、 “红边”、 “近峰A” 处对重金属Pb2+胁迫污染响应特别的敏感, 可以作为监测重金属Pb2+污染程度的有效波段。 最后通过LD-CR-SIDSCAtan模型的应用结果与玉米叶片中Cu2+和Pb2+含量进行线性拟合分析, 从而反演和预测了重金属Cu2+和Pb2+对玉米植株的污染程度。

When crops are contaminated with heavy metals, their tissue structure and chlorophyll content will be destroyed, which will affect the metabolism and health of crops. People and animals will have fatal injuries if they eat the contaminated crops. Hyperspectral remote sensing is now widely used to monitor the extent of crops affected by heavy metals, and in the heavy metal pollution, the spectral crop leaves are still very similar to those of the traditional monitoring methods and spectral characteristic parameters of routine, so it is difficult to distinguish between different spectral weak information, and application of hyperspectral remote sensing is the focus and difficulty of the study. The maize leaf spectral data, chlorophyll content and relative content of heavy metals Cu2+ and Pb2+ were collected by setting different concentrations of Cu2+ and Pb2+ stress. A LD-CR-SIDSCAtan model combined with the continuum removal (CR), spectral correlation angle (SCA), spectral information divergence (SID) and tangent function (Tan) and Langmuir distance (LD) is proposed in this study, and the traditional measure, such as spectral correlation coefficient (SCC), SA (spectral angle), tangent spectrum (DSA), spectral information divergence and spectral correlation tangent (SIDSAMtan), spectral information divergence and spectral gradient tangent (SIDSGAtan) and conventional spectral characteristic parameters, such as the maximum value of red edge (MR), green peak height (GH) and red edge area surrounded by a first order differential (FAR), red edge derivative curve steepness (FCDR), blue (DB), red band depth (RD) compared to verify the feasibility and superiority of the model. The LD-CR-SIDSCAtan model was applied to measure the spectral difference information about the overall waveform and the subband of maize leaves under Cu2+ and Pb2+ stress at different concentrations. The results show that the LD-CR-SIDSCAtan model realized the qualitative analysis of heavy metal Cu2+ and Pb2+ pollution, could measure the spectral correlation coefficient of more than 0.99 of the difference information between the similar spectral information, and waveform difference information was significantly related to the leaf chlorophyll content and the relative content of heavy metals Cu2+ and Pb2+ that measured, and also found the spectra response wave band under the stress of heavy metals Cu2+ and Pb2+. When the whole spectral range of spectral data is measured, the spectral difference is more obvious when the value of the model is negative. When the value of the model is positive, the larger the value of the model is, the larger the spectral difference will be. Therefore, with the increase of heavy metals Cu2+ and Pb2+ concentration, the difference of spectra increased, which means that the heavy metal Cu2+ and Pb2+ pollution degree is more serious; maize plants suffer from heavy metal pollution in Cu2+ stress when measuring the local subband range of spectral data, in the “blue” and “red edge”, “near the Valley”, “at the peak of B” were specially sensitive to heavy metal Cu2+ stress pollution response and can be used as an effective band of monitoring heavy metal pollution Cu2+; when the maize plants are under heavy metal pollution in Pb2+ stress, in the “Purple Valley”, “blue”, “yellow” and “red Valley”, “red edge”, “near at the peak of A” were specially sensitive to heavy metal Pb2+ stress pollution response and can be used as an effective band of monitoring heavy metal pollution Pb2+. Finally, through the linear fitting analysis of the application results of LD-CR-SIDSCAtan model and the content of Cu2+ and Pb2+ in maize leaves, the pollution degree of heavy metals Cu2+ and Pb2+ to maize plants was inversed and predicted.