作物重金属铜污染的HHT边际谱特征与污染预测模型

为了解读作物受重金属污染的光谱响应与光谱特征，以不同浓度梯度硫酸铜（CuSO₄·5H₂O）胁迫土壤的盆栽玉米培养胁迫实验为研究对象，依据不同胁迫梯度下玉米叶片反射光谱以及叶片中Cu²⁺含量的测定数据，采用希尔伯特-黄变换（HHT）方法，研究了玉米叶片在不同Cu²⁺胁迫梯度下光谱的Hilbert边际谱变化特征与污染程度预测方法.通过构建玉米叶片光谱的边际谱熵（MSE），边际谱幅值（MSA）、边际谱陡坡斜率（MSSS）和边际谱包围面积（MSEA）等特征参量，分析叶片在不同Cu²⁺污染程度下的边际谱变化；同时基于边际谱特征参量值与叶片中Cu²⁺含量的相关性分析和逐步回归统计，提出了玉米叶片重金属污染的Cu²⁺含量预测指数模型.实验结果表明，不同Cu²⁺胁迫梯度下，玉米叶片光谱的边际谱为分布在100Hz频率以内的连续谱；MSE值表现出与叶片中Cu²⁺含量呈负相关的变化趋势，而MSA、MSSS和MSEA值都表现出与叶片中Cu²⁺含量呈正相关的变化趋势；由于MSEA值与叶片中Cu²⁺含量的相关性最好，可把MSEA作为监测作物重金属污染衡量或预测的最优指标；根据MSE、MSA、MSSS和MSEA值构建的Cu²⁺含量预计指数模型应用结果比较，证明MSEA指数模型具有最优的预测能力.

Characteristic changes of HHT marginal spectra and pollution predicting models on crop polluted by the heavy metal copper

In order to understand the spectral responses and spectral characteristics of crop polluted by heavy metal, the cultivated experiments of corns were implemented based on the pot soil stressed by different CuSO₄·5H₂O concentrations, and the Hilbert marginal spectral characteristics and the predicting pollution degree method were researched by the Hilbert-Huang transform (HHT) according to the measured data of corn leaves' reflectance spectra and the Cu²⁺ contents in leaves under the different stress gradients. The characteristic changes of marginal spectra were analyzed on different Cu²⁺ pollution levels of corn leaves by constructing some marginal spectral characteristic parameters such as marginal spectrum entropy (MSE), marginal spectral amplitude (MSA), slope of marginal spectral slope (MSSS) and marginal spectral enclosing area (MSEA), at the same time, some exponential models on predicting the Cu²⁺ contents in corn leaves were put forward based on the correlation analysis and stepwise regressing statistics on the marginal spectral characteristic parameter values and the Cu²⁺ contents in leaves because of the heavy metal pollution. The results show that Hilbert marginal spectra of corn leaf under different Cu²⁺ stress gradients was distributed within 100Hz frequency continuously, the MSE values showed a variation trend of negative correlation with Cu²⁺ contents in leaves, but the MSA, MSSS and MSEA values showed a variation trend of positive correlation with the Cu²⁺ contents. And the MSEA can be used as the best index to measure or predict heavy metal pollution in crops due to its best correlation between MSEA values and Cu²⁺ contents in leaves. The application results of the exponential models, which were built by the MSE、MSA、MSSS and MSEA values for predicting the Cu²⁺ contents in corn leaves, were compared and proved that the MSEA exponential model had the best predicting ability.