An Exploratory Study on the Effect of Petroleum Hydrocarbon on Soils Using Hyperspectral Longwave Infrared Imagery

Abstract

Manmade crude oil contamination, which has negative impacts on the environment and human health, can be found in various ecosystems all over the globe. Hyperspectral remote sensing (HRS) is an efficient tool to investigate this crude oil contamination where its electromagnetic spectrum is analyzed. This exploratory study used an innovative HRS imagery sensor to study the effect of petroleum hydrocarbon (PHC), found in crude oil, on the spectrum of soils across the longwave infrared (LWIR 8–12 μm) spectral region. This contrasts with previous studies that focused on shortwave and midwave infrared (SWIR 1–2.5 and MWIR 3–8 μm, respectively) regions. An outdoor HRS image of three different types of soils, contaminated with 11 PHC concentrations, was processed and analyzed. Since PHC is spectrally featureless in the LWIR region, the analysis focused on the spectral alteration of the dominant minerals in the soils. Good evaluation metrics of R2 > 0.83 and a root-mean-squared-error (RMSE) between 1.06 and 1.33 wt % showed that the PHC level can be predicted with relatively good accuracy, even without direct spectral features of crude oil PHC, using an airborne LWIR camera in field conditions. This study can be used as a proof of concept for future airborne remote sensing of PHC-contaminated soils.

Quantification of Hydrocarbon Abundance in Soils Using Deep Learning with Dropout and Hyperspectral Data

“…However, it is noticed that this data is more complex as the individual models as it contains four different soil types. In the work of the authors of [16], an individual model per soil type was recommended as the generic models degraded the responses. We believe that with further tuning of the hyperparameters, improved results could have been achieved.…”

“…In order to simulate a more realistic scenario, and following a similar approach presented in the work by the authors of [16], noise was added to the data to simulate continuous spectra values instead of discrete and also to evaluate the noise rejection of the models. Here, the datasets were corrupted with Random Gaussian noise with signal-to-noise ratio (SNR) ranging from 10 to 40 dB.…”

“…Each pixel responses are captured by 88 spectral bands in the spectral range of 8 to 12 µm with spectral resolution of 0.25 cm −1 . The experimental protocol, data capturing, and preprocessing of these datasets are fully described in the work by the authors of [16].…”

“…Therefore, spectral information obtained in the NIR-SWIR range is excellent for both the quantitative and qualitative analysis of HCs in soils [13]. Recent works have also successfully demonstrated the use of Longwave Infrared (LWIR) for Petroleum HC detection [16].Different methods have been used to analyze reflectance spectroscopy data to detect HCs in soils; the authors of [5] used regression analysis and spectral preprocessing to generate statistical models to identify different HC products mixed with a mineral substrate. The authors of [15] used Diffuse Reflectance Infrared Fourier-Transform (DRIFT) spectroscopy which is a hand held spectrometer for the prediction of total petroleum hydrocarbons in contaminated soils.…”

“…Therefore, spectral information obtained in the NIR-SWIR range is excellent for both the quantitative and qualitative analysis of HCs in soils [13]. Recent works have also successfully demonstrated the use of Longwave Infrared (LWIR) for Petroleum HC detection [16].…”