Prediction of Soil Lead Content Using Visible and Near-Infrared Spectroscopy

Abstract

Soil contamination by heavy metals has been an environmental problem. Visible and near infrared spectroscopy (VNIRS) is a promising alternative to predict soil contaminant elements. Generally, the prediction is performed using the entire VNIR region of 400 – 2400 nm. Based on absorption of lead (Pb) on soil spectrally active constituents, a spectral region of 400 – 1100 nm was used to predict Pb content in soil. A combination of genetic algorithm and partial least squares regression (GA-PLSR) was adopted to develop the prediction model. Compared with the prediction using the entire VNIR region of 400 – 2400 nm, the prediction accuracy was improved by using the spectral region of 400 – 1100 nm for different divisions of calibration and validation sets. The result indicates the spectral region of 400 – 1100 nm is effective to predict Pb content in soil, and the method has potential to be applied in field condition.