Assessing the detection limit of petroleum hydrocarbon in soils using hyperspectral remote-sensing

Abstract

One of the most important characteristics of any quantitative method is its detection limit, which is the lowest detectable amount of analyte in a sample in a given method. The detection limit helps to know to which extent a method is applicable and valid. The detection of petroleum hydrocarbon contamination in bare soils is important, because of its vast distribution and its negative effects on humans and the environment. Hyperspectral remote sensing is an acceptable, cost-effective and spatially comprehensive tool, which has detection possibilities of petroleum hydrocarbon. However, previous studies that engaged with this topic did not directly consider the detection limit of the method and no detection limit has been systematically explored. In this study, an outdoor experiment was executed in which three types of soil were contaminated with 14 levels of pure crude oil and measured in outdoor conditions with a hyperspectral camera (950–2500 nm) at three distances from the camera (4, 8 and 12 m). For each soil type, and for each distance, the detection limit was systematically calculated. The results show that the detection limit is a dynamic range affected by the spatial domain and the soil type. Moreover, a significant (p < 0.001) difference between the average detection limit (1.32, 2.09 and 2.90 wt%) of the three soil types was found, suggesting that the detection limit is not only related to the PHC wt% but also depends on the soil type. The results presented in this study can help to improve our understanding of the limitations of hyperspectral remote sensing in direct detection of petroleum hydrocarbon and can be projected to other hyperspectral domains and soils.