Predição de Cr, Cu, Hg, Ni, Pb, Zn em solos de manguezal poluído no nordeste do Brasil por meio da espectroscopia de reflectância no infravermelho próximo

Abstract

Infrared reflectance spectroscopy has shown potential for use in environmental studies. It can assist in monitoring and preventing contamination in different environments. However, its application in mangroves is scarce, thus proving to be a viable alternative for monitoring these ecosystems vulnerable to environmental contamination. In this sense, this study aimed to evaluate the potential of near-infrared spectroscopy in predicting heavy metals in mangrove soils in the estuary of the Botafogo-PE River. For this, composite samples were collected in the 0-5 cm depth, obtaining 61 samples, in which spectral readings were taken in the near-infrared range (NIR, 1000-2500 nm). Pre-processing was applied to the data to improve the accuracy of the models, and partial least squares (PLS) regression was used to build the prediction models for the following attributes: clay content, SOM content, pH, Eh, Cr, Cu, Hg, Ni, Pb, and Zn concentrations. The performance of the models was evaluated using the root mean square error (RMSE), adjusted coefficient of determination (R²adj), bias, Ratio of Performance to InterQuartiledistance (RPIQ), and Lin's correlation coefficient of agreement (CCC) for the validation set. The best results were obtained after applying the following preprocessing: Savitzky-Golay (SG) and Multiplicative Scatter Correction (MSC). The predictive models that presented the best performances were: Cr (R2adj = 0.82; RMSE = 6.78; CCC = 0.85; bias = -0.15; RPIQ - 1.4) when the SG preprocessing was used, and Pb (R2adj = 0.85; RMSE = 2.35; CCC = 0.85; bias = -0.3; RPIQ = 1.44) when the MSC was applied. The variables pH and Eh showed the lowest performances for both preprocessing. The results provide evidence that near infrared spectroscopy can be used efficiently to predict the studied metals, mainly Cr and Pb, which showed the best results, presenting itself as a technique to complement traditional analyses.