Distinction And Inversion Of Copper And Lead Concentration In Soil In View Of Frequency Domain Spectroscopy

Effective identification of heavy metal pollution information from soil spectral data and quantitative analysis of heavy metal pollution and its pollution degree are major requirements for soil pollution monitoring and prevention.The monitoring of copper(Cu)and lead(Pb)pollution in five typical heavy metal elements of soil pollution is an urgent problem to be solved.Visible-short-wave infrared(VINR-SWIR)and X-ray fluorescence(XRF)spectra belong to electromagnetic spectrum signals.Summing up the application results,inversion of Cu and Pb concentrations in soil could only be achieved by spectral domain transformation.There is less in-depth mining of local and detailed information on the spectrum of Cu and Pb contaminated soils from the frequency domain.After the spectra of soil Cu and Pb pollution are transformed by frequency domain,the anomaly information could be highlighted by the high frequency or low frequency characteristics of the spectral line.According to this,the local anomalous variation of soil Cu and Pb pollution spectra could be identified and analyzed.In this study,the VNIR-SWIR spectra of soil Cu and Pb contaminated experimental samples and the VNIR-SWIR and XRF spectra of the mining area were collected.The frequency domain methods were used to explore the weak difference between the spectra of soil samples,the frequency domain parameters of VNIR-SWIR spectra were used to establish the inversion models of copper and lead concentration in soil.It aimed to provide a theoretical basis and a new perspective for the frequency domain VNIR-SWIR and XRF spectral study of Cu and Pb concentration in soil,to specify the research scope of the spectra of soil with different Cu and Pb concentrations,to provide new ideas for establishing models for determining Cu and Pb concentration limits,classifying pollution categories and detecting Cu and Pb excess by using the spectral frequency domain method,and to promote the application of spectral technology in monitoring Cu and Pb concentration in soil.Firstly,the smooth pseudo-Wigner distribution(SPWVD)of the spectrum was applied exploratory,and a spectral second-order differential Gabor(SD-Gabor)expansion method was proposed to analysis and explore the frequency domain transform of VNIR-SWIR spectra of laboratory soil samples with different CuSO4·5H2O or Pb(NO3)2 additions.Results showed that SPWVD of soil samples with different concentrations of Cu or Pb showed weak difference in spectral range of 2,100-2,500 nm,it could be inferred that the Cu concentration threshold of soil was between 121 and 492.3?g·g-1,and the Pb concentration threshold of soil was 568-747.33 ?g·g-1.In the spectral range of 800?2500 nm,the weak difference of transformation spectra of soil that uncontaminated and contaminated by CuSO4·5H2O or Pb(NO3)2 could be distinguished used the SD-Gabor expansion method of the spectrum.The Cu concentration threshold in soil estimated by this method was 159-215 ?g·g-1,and the Pb concentration threshold in soil was 568-747.33 ?g·g-1.Whether the pollution of Cu and Pb could be distinguished from the soil remains to be verified.Then,based on the results obtained by laboratory research methods,the spectra of soils with copper and lead concentrations significantly lower than those of laboratory designed samples were studied.In spectral range of 800-2,500 nm,VNIR-SWIR spectral SD-Gabor expansion method based on frequency domain could identify 75%of Pb concentration exceeded the standard of in-situ sampled soil.The method of harmonic decomposition reconstruction could be successfully applied to soil XRF spectral denoising.By compared with empirical mode decomposition(EMD)method and wavelet reconstruction denoising method,it was found that soil XRF spectrum denoising effect was best when the number of harmonic decomposition was 400.SPWVD of soil denoising XRF spectrum could identify Pb concentration in the in-situ sampled soil exceeded the standard,in all in-situ sampled soil,75%of samples with excessive Pb concentrations could be identified,and 79.17%of samples with Pb concentration not exceeded the standard could be identified.However,neither method nor spectrum could identify the soil Cu concentration exceeded the standard in the region.Finally,the HHT parameters(the peak of the spectral Hilbert spectrum(Ea),EMD energy entropy(Em))were extracted by Hilbert-Huang Transform(HHT)analysis of VNIR-SWIR spectra of soil samples with different Cu and Pb concentrations.And integrated spectral domain parameters(first derivative(FD),continuum removal(CR),multiplicative scatter correction(MSC),absorbance conversion(log(1/R))),soil water content and SOM concentration,used the mathematical analysis method of Boxplot,cluster analysis and partial least squares regression(PLSR)to establish the BC-PLSR inversion model of Cu and Pb concentration in soil.The results showed the R2 of predicted values and the measured values of the optimal inversion model of Cu concentration were 0.9922,0.9528 and 0.9227,respectively.The predicted values of the optimal inversion model of Pb concentration and the measured values of R2 were 0.8578,0.6980 and 0.8702,respectively.The comparative analysis verified the effectiveness of HHT parameter inversion and the superiority and universality of the BC-PLSR inversion model.