基于小波神经网络的矿区土壤铜含量反演研究

Remote Sensing Inversion on Copper Content in Mine Tailings Based on Wavelet Neural Network

研究了基于小波神经网络方法反演矿区尾矿土壤铜元素含量的可行性。以陕西金堆城矿区尾矿库为研究区,利用ASD光谱仪测量土壤光谱,通过实验室化学分析获取土壤样本铜元素含量;利用小波神经网络方法对矿区尾矿土壤的Cu含量进行建模研究,建立矿区尾矿土壤铜含量高光谱反演模型,并对模型进行了检验。研究发现:土壤铜含量反演模型预测铜元素含量的相关系数R2为0.7381,均方根误差RMSE为0.9788。研究结果为深入研究矿区尾矿土壤重金属含量遥感监测机理提供了理论依据,对利用高光谱遥感数据获取土壤重金属含量信息具有重要的应用价值。

A study on the feasibility of inversion method based on wavelet neural network to estimate the contents of copper in mining area was carried out.Taking the tailings dam in Shaanxi Jinduicheng mine as the study area,the tailing's spectral was measured by ASD spectrometer,and the copper element content in tailing samples was obtained by laboratory chemical analysis.By wavelet neural network method,the hyperspectral inversion model of copper content in mine tailings was built,and accuracy of the model was evaluated.The study found that between predicted copper content and the measured copper content,the correlation coefficient R2 was 0.7381,and the root mean square error was 0.9788.The results of the research provided a theory basis for remote sensing monitoring mechanism of heavy metals contents in mine tailings,and had important application value for obtaining heavy metal content in soil by hyperspectral remote sensing data.