The high risk of environmental contamination posed by the use of petroleum hydrocarbons (HCs) necessitates technologies for detecting leaks. The effectiveness of sensors operating in near infrared (NIR) and short wave (SWIR) have already been demonstrated for this purpose. However, the applicability of thermal data (TIR) remains largely unexplored. This study aims to evaluate the applicability of TIR data for the detection and monitoring of soils contaminated with HCs, by tracking changes in the emissivity of samples in the post-contamination period. A controlled laboratory experiment was conducted in which samples of different mineral substrates were mixed with oils of different APIs° and at different concentrations. The samples were measured using a FTIR spectroradiometer over six months. Emissivity features characteristic of each substrate and HCs were selected and parameterized to analyze their geometric variations. PCA and GLMM were the statistical models used to quantify the variations of these features as a function of i) contamination time; ii) temperature; iii) API°; and iv) oil concentration. Based on the findings, it was demonstrated that TIR is useful for detecting contamination by HCs. However, the patterns are different for each type of substrate. It was identified that the depth of the oil feature is directly proportional to the increase in HC concentration in mixtures with clayey substrates and inversely proportional to sandy substrates. As for API°, in sandy and dolomite soils, the oil feature depth is greater in mixtures with lighter oils and, in the case of clay soils, is greater in mixtures with heavier oils. As for temporal analysis, the oil feature depth in mixtures with dolomitic substrate increases over time. On the other hand, variations in sandy and dolomite soil features showed that the soil feature depth increases as the weeks pass. In relation to temperature, the characteristic of oil and soil feature depth decreases with increasing temperature in all mixtures. Despite the optimistic results, the influence of all these variables and the presence of water, along with the variation of soil and HC features, make the quantification of oil contamination a challenge in the TIR range.
