The use of reflectance visible–NIR spectroscopy to predict seasonal change of trace metals in suspended solids of Changjiang River

• First derivative transformation highlights the peaks of spectral of suspended solid.
• Spectra HPCA differentiates samples of main channel and tributaries seasonally.
• PLS models showed better prediction for heavy metals in suspended solids.

Diffuse reflectance visible and near infrared spectroscopy (Vis/NIR) was used to analyze the tracemetal content (As, Cd, Co, Cr, Cu, Hg, Ni, Pb and Zn) of suspended solids collected from the Changjiang River during different seasons and to compare concentrations between the main river channel and its tributaries. First derivative (FD) transformation highlighted peaks in the reflectance spectra, including a weak overlapping peak in the visible region, which coincided with peaks in trace metal concentration. Principal component analysis (PCA) showed the main peaks at around 430 nm, 500 nm, 530 nm and 650 nm to be from iron oxides in visible region, and around 1414 nm, 1914 nm and 2210 nm to be from water, hydroxyl and clay minerals. In addition, hierarchical principal component analysis (HPCA) showed that the spectral peaks for the tributary samples increased between 800 and 2250 nm during the flood season, when compared to the dry season. Partial-least square (PLS) regression was also used to build models for evaluating trace metal concentration in the suspended solids, whereby the optimal spectra regions, factors and regression parameters were determined by comparing different pre-processing procedures. These results indicate the potential of providing a rapid non-destructive method for the efficient suspended solids analysis and the acquisition of suspended solids information using diffuse reflectance Vis/NIR spectroscopy.