Main content area

Reflectance and imaging spectroscopy applied to detection of petroleum hydrocarbon pollution in bare soils

Correa Pabón, Rosa Elvira, Souza Filho, Carlos Roberto de, Oliveira, Wilson José de
The Science of the total environment 2019 v.649 pp. 1224-1236
absorption, cost effectiveness, data collection, ecosystems, hazardous waste, human health, hyperspectral imagery, image analysis, models, monitoring, oils, petroleum, pipelines, polluted soils, reflectance, reflectance spectroscopy, remediation, risk, soil pollution, soil sampling, spectrometers, wavelengths
Accidental releases of hazardous waste related to the extraction, refining, and transport of oil and gas are inevitable. Petroleum facilities and intrinsic pipelines present environmental pollution risks, threatening both human health and ecosystems. Research has been undertaken to enhance the conventional methods for monitoring hazardous waste problems and to improve time-consuming and cost-effective ways for leak detection and remediation process. In this study, both diffuse and imaging (hyperspectral) reflectance spectroscopy are used for detection and characterization of petroleum hydrocarbon (PHC) contamination in latosols. Laboratory and field measurements of PHC-contaminated and PHC-free soils were collected from an oil facility using an ASD FieldSpec-3 high-resolution portable spectrometer (2150 channels) covering visible, near infrared and shortwave infrared wavelengths (VNIR-SWIR: 350–2500 nm). The hyperspectral image dataset was acquired with the ProSpecTIR-VS airborne sensor using 357 channels in the VNIR-SWIR range at 1 m of spatial resolution. Narrow intervals of reflectance spectra were analyzed to identify the primary mineral and PHC absorption bands in soil samples and to investigate the spectral match with airborne hyperspectral data. The Multiple Endmember Spectral Mixture Analysis (MESMA) method was employed in three hierarchical levels to classify the hyperspectral imagery. The classification product yielded from MESMA model at the fourth level was 98% accurate in discriminating contaminated soils. The results demonstrated the applicability of both diffuse reflectance and imaging (hyperspectral) spectroscopy to identify bare soils contaminated by PHC leaks and spills. These technologies can also provide useful information for remediation initiatives, thereby avoiding further problems with hazardous waste.