Rapid measurement of polycyclic aromatic hydrocarbon contamination in soils by visible and near-infrared spectroscopy

Polycyclic aromatic hydrocarbons (PAHs) are widely distributed organic pollutants. At petroleum contaminated sites, PAHs are often the key risk drivers because of their carcinogenicity. Assessing the risk of PAH at contaminated sites by conventional soil sampling, solvent extraction and gas chromatography–mass spectrometry (GC–MS) analysis is expensive and time-consuming. Employing a rapid and cheap measurement technique for PAH would be beneficial to risk assessment by eliminating costs and time associated with the conventional method. The literature has shown that visible and near infrared (vis-NIR) spectroscopy is a rapid and cheap technique for acquiring information about key soil properties. In this study, models based on vis-NIR spectroscopy (350–2500 nm) were developed to predict and map PAH in contaminated soils for the ultimate aim of informing risk assessment and/or remediation. The reference chemical analytical method used was GC–MS while the multivariate analytical technique used for model development was partial least squares (PLS) regression analysis with full cross-validation. A total of 150 soil samples from the UK were used for the laboratory-scale study while 137 samples were used for the near-onsite adaptive trials at three oil spill sites in Ogoniland, Niger Delta province of Nigeria. Both laboratory- and field-scale results showed that soil diffuse reflectance decreased with increasing PAH concentration. Hydrocarbon absorption features observed around 1647 nm in the first overtone region of the NIR spectrum showed a positive link to PAH. Laboratory-scale study showed that both individual and combined effects of oil concentration, and moisture and clay contents on soil spectral characteristics and calibration models were significant (p<0.05). For the field-scale study, inverse distance weighting soil maps of PAH developed with chemically-measured and vis-NIR-predicted data were comparable with a fair to good agreement between them (Kappa coefficient = 0.19–0.56). Hazard assessment of the oil spill sites using both measurement methods showed that the impact of the contamination varied distinctly across the management zones. The type of action required for site-specific risk assessment and/or remediation also varied among the different zones. This result shows promise that vis-NIR can be a good screening tool for petroleum release sites.