Visible and near infrared reflectance spectroscopy for measuring soil heavy metal content as a quick method

Conventional methods for investigating soil heavy metal contamination based on raster sampling and chemical analysis are time consuming and relatively expensive. Reflectance spectroscopy within the visible-near-infrared （VNIR） region is known for being rapid, convenient, and accurate. Due to low abundance, heavy metals in soils cannot exhibit their inherently spectral features. The objective of this study were （1） to examine the possibility of reflectance spectra as a rapid method to assess Ni, Cr, Cu, Hg, Zn, As, Pb, and Cd in soils, and （2） to explore the mechanism by which to predict spectrally featureless heavy metals. Two research areas located in the south （120 samples） and north （61 samples） of Nanjing City were researched. Reflectance spectra of the samples were recorded on a Perkin-Elmer Lambda 900 spectrophotometer. Partial least-square regression （PLSR） approach was used to develop calibration equations between the spectral data and measured values for heavy metals. The prediction results of the two areas were consistent. Except for Cd all the other elements could be successfully predicted using the reflectance spectra. The prediction accuracy for Ni, Cr, and Cu was highest, while that for Hg and Cd was lowest. In order to further explore the physicochemical mechanism behind the relations between reflectance spectra and heavy metals, one more principal component analysis （PCA） was done for nine elements （eight heavy metals and Fe）. The results of PCA for the both areas were also consistent. Loading plot of factors 1 and 2 from the results of PCA showed that Ni, Cr and Cu, the prediction accuracy of which was highest, were clustered together with Fe. For the other elements that were less correlated with Fe, their prediction accuracy was also lower than that of the three elements. Fe is spectrally active and can exhibit its absorption features. Therefore, the inter-correlation between heavy metals and spectrally active total Fe is the major mechanism by which to predict spectrally featureless trace metals. This study for the two areas showed the potential of reflectance spectra to predict microelements in soils.

Visible and near infrared reflectance spectroscopy for measuring soil heavy metal content as a quick method