土壤重金属铬含量高光谱灰色关联度估测

Hyperspectral Grey Relational Estimation of Heavy Metal Chromium Content in Soil

快速、准确地测定土壤重金属含量,对防治土壤重金属污染、改善土壤环境和保障食品安全有着重要意义。以山东省烟台市采集的70个土壤样本为基础,首先分析土壤重金属铬含量的分组光谱特性;然后利用6种变换方法对土壤光谱反射率进行变换,根据极大相关性原则选取反演因子;最后利用灰色关联度模型初步估测铬含量,并对估测结果进行修正,采用决定系数和平均相对误差评价模型的有效性。结果表明,土壤光谱反射率随铬含量的升高而降低,二者呈负相关性;利用灰色关联度模式识别方法对重金属铬含量进行估测后的决定系数为R^2=0.656,平均相对误差为16.590%,而利用灰色关联度修正模型对估测值进行修正后,决定系数为R^2=0.912,平均相对误差为6.632%。研究表明,利用灰色关联度修正模型定量估侧土壤重金属铬含量有效。

Rapid and accurate determining heavy metal content in soil is of great significance in preventing heavy metal pollution, improving soil environment and ensuring food safety. Based on the data of 70 soil samples in Yantai city of Shandong Province, firstly, the spectral characteristics of the heavy metal chromium content in soil are analyzed. Then six transformations of soil spectral reflectance are transformed and the inversion factor is selected according to the principle of maximum correlation. Finally, the grey correlation model is used to estimate the chromium content, and the estimation results are revised. At the same time, the stability and effectiveness of the model are verified. The results show that the spectral reflectance decreases with increasing chromium content, showing negative correlation. The determination coefficient of heavy metal chromium content is 0.656 after using the grey relational method and the average relative error is 16.590%. While the coefficient of determination is 0.912 using the grey relational correction model to correct the estimated value and the average relative error is 6.632%. The study shows that it is feasible to use the grey relational correction model to quantitatively estimate the heavy metal content of soil.