Diffuse reflectance spectroscopy for monitoring potentially toxic elements in the agricultural soils of Changjiang River Delta, China

The correlations between six toxic elements (Cd, Cr, Pb, Cu, Hg and As) and soil diffuse reflectance spectra within the visible and near-infrared (Vis/NIR, 400–2500 nm) and the mid-infrared (MIR, 4000–400 cm− 1) regions were studied to assess their binding forms and predict their concentrations. Based on the correlation patterns with spectra reflectance, the six potentially toxic elements studied could be categorised into two groups. The first group (Cr, Cu and As) have stronger negative correlation coefficients with the spectral bands attributed to the absorption features of iron oxides, clays and organic matter, suggesting they are strongly bound to these soil constituents. The second group (Cd, Pb and Hg) only display a significant correlation with the spectral region related to organic matter, indicating that binding with organic matter is important for these metals. Univariate models were built using spectral bands with highest correlation coefficients for these elements. In contrast, partial least-square (PLS) regression models were also constructed for avoiding colinearity. Overall, univariate and PLS models for the Vis/NIR spectra showed better prediction capabilities than the corresponding MIR spectra, and are feasible to predict the toxic metals levels in agricultural soils.