Detection of hydrocarbons in clay soils: A laboratory experiment using spectroscopy in the mid- and thermal infrared

https://doi.org/10.1016/j.jag.2012.11.001Get rights and content

Abstract

Remote sensing has been used for direct and indirect detection of hydrocarbons. Most studies so far focused on indirect detection in vegetated areas. We investigated in this research the possibility of detecting hydrocarbons in bare soil through spectral analysis of laboratory samples in the short wave and thermal infrared regions. Soil/oil mixtures were spectrally measured in the laboratory. Analysis of spectra showed development of hydrocarbon absorption features as soils became progressively more contaminated. The future application of these results airborne seems to be a challenge as present and future sensors only cover the diagnostic regions to a limited extent.

Highlights

► We show a laboratory experiment on clay soils with hydrocarbon pollution in different quantities. ► Absorption features related to hydrocarbon presence are strongly developed at low quantities. ► Statistical analysis with PLSR and SMLR indicates that it is possible to develop predictive models. ► Present and upcoming sensors only cover part of the diagnostic features.

Introduction

Remote sensing in the near- and shortwave infrared can be used for direct and indirect detection of hydrocarbons. Previous research largely focused on direct spectral detection of hydrocarbons (e.g. Lammoglia and Filho, 2011, Bihong et al., 2007, van der Werff et al., 2006, Winkelmann, 2005, Kühn et al., 2004, Hörig et al., 2001, Malley et al., 1999). The presence of hydrocarbons in a soil can also lead to chemical and mineralogical alterations. Bacterial oxidation of hydrocarbons can establish anomalous redox zones that favour the development of a diverse array of chemical changes (Schumacher and Abrams, 1996). An overview of hydrocarbon-induced alterations, and consequences for remote sensing is given by van der Meer et al. (2002), Schumacher and Abrams (1996), and Cloutis (1989).

The aforementioned studies focused on natural hydrocarbon seepage and/or controlled experiments. Recent work, related to pipeline leakage, in the laboratory (Noomen et al., 2006, Smith et al., 2004), field (van der Meijde et al., 2009) and airborne (van der Werff et al., 2008), showed that indirect detection of hydrocarbons through analysis of vegetation reflectance is possible. These studies, however, focused only on areas with relatively dense vegetation cover, leaving areas with little or no vegetation unaccounted for.

In this research, we investigate the possibility of detecting hydrocarbons in bare soil through spectral analysis of the short wave and thermal infrared wavelength regions. Based on the discriminative wavelengths found and available thermal infrared sensors, a feasibility study shows that the direct applicability of such techniques for airborne based detection and monitoring is currently limited.

Section snippets

Laboratory experiment

The laboratory experiment consisted of two stages. At first, 10.00 g soil samples was created from a mixture of air-dried clay soil and a hydrocarbon (Shell mineral engine oil). The hydrocarbons were added to the soil in increments of 10% weight, from 10% till 100% hydrocarbon content. In the second stage, the air-dried clay soil was moistened (8.50 g soil to 1.50 g water) and well mixed to make it comparable to field conditions. The sample was prepared with hydrocarbon concentrations that ranged

Trends in spectra

Analysis of both dry soil to oil mixtures (0–100% hydrocarbon contamination) as well as the moist soil to oil mixtures (0–10% hydrocarbon contamination) shows that there was development of hydrocarbon features and reduction in the soil features as soils became progressively more contaminated (see Fig. 1 for an example of the moist mixture). The most dominant and specific hydrocarbon features (Stuart, 2004) between 2850 and 2960 cm−1 were already visible from a hydrocarbon concentration of

Application for different hydrocarbons and soils

In the laboratory an oil hydrocarbon was used. In real field situations often other hydrocarbons will be present. It was not possible, unfortunately, to measure high concentrations of highly toxic substances, such as e.g. benzene condensates, without a fume hood. We believe, however, that the impact of different hydrocarbons is not necessarily large. An overview of hydrocarbon absorption features (Stuart, 2004) shows that hydrocarbon absorption features in the so-called Csingle bondH fingerprint region of

Conclusion

We have investigated the possibility of detecting hydrocarbons in bare soil through spectral analysis of the short wave and thermal infrared regions. For this purpose a laboratory experiment was set up. Hydrocarbon polluted soil samples were analysed for diagnostic wavenumbers that can predict the amount of hydrocarbons in the soil. A two-step approach modelling using PLSR and SMLR resulted in a model that had a goodness of fit around 0.95 (Adjusted R2) and is based on spectral features that

Acknowledgements

The authors wish to thank two reviewers for their constructive comments. We thank Boudewijn de Smeth, head of the geochemical laboratory at UT-ITC, for his support during the measurements. This research was financially supported by the Nederlandse Aardolie Maatschappij BV.

References (39)

  • F. Bihong et al.

    Mapping hydrocarbon-induced mineralogical alteration in the northern Tian Shan using ASTER multispectral data

    Terra Nova

    (2007)
  • E. Cloutis

    Spectral reflectance properties of hydrocarbons – remote-sensing implications

    Science

    (1989)
  • M. Crawley

    Statistical Computing: An Introduction to Data Analysis Using S-Plus

    (2006)
  • B. Hörig et al.

    Hymap hyperspectral remote sensing to detect hydrocarbons

    International Journal of Remote Sensing

    (2001)
  • J.A. Hackwell et al.

    LWIR/MWIR imaging hyperspectral sensor for airborne and ground-based remote sensing

    SPIE

    (1996)
  • J.L. Hall et al.

    Space-based mineral and gas identification using a high-performance thermal infrared imaging spectrometer

  • J.L. Hall et al.

    First flights of a new airborne thermal infrared imaging spectrometer with high area coverage

  • J.L. Hall et al.

    Mineral and gas identification using a high-performance thermal infrared imaging spectrometer.

  • C. Hecker et al.

    Thermal infrared spectrometer for earth science remote sensing applications – instrument modifications and measurement procedures

    Sensors

    (2011)

    • Cited by (24)

    • Detection of petroleum hydrocarbons in continental areas using airborne hyperspectral thermal infrared data (SEBASS)

      2021, Remote Sensing of Environment
      Citation Excerpt :

      Applications of TIR instruments for the detection of oiled samples are limited (van der Meijde et al. 2013), due to weak spectral features in this range when compared to NIR-SWIR-MWIR ranges (Correa Pabón and Souza Filho, 2019). However, previous experiments conducted in the laboratory (van der Meijde et al. 2013) and in the field (Pelta and Ben-Dor 2019) indicate that close-range TIR hyperspectral measurements can detect contaminated soils through variations in the spectral features of minerals induced by the addition of PHC. Here, the applicability of TIR data for the detection of soils contaminated with PHCs is investigated.

    • Hydrocarbon micro-seepage detection from airborne hyper-spectral images by plant stress spectra based on the PROSPECT model

      2019, International Journal of Applied Earth Observation and Geoinformation
      Citation Excerpt :

      Subterranean hydrocarbon gases may be traced to their source by locating natural hydrocarbon seepage (Smith et al., 2004a), gas vents that escape from underlying gas-bearing rock due to partial failure or temporary breach of the top seal, resulting in numerous changes in the rocks and soils through which they pass (Meijde et al., 2013; Asadzadeh and de Souza, 2017).

    • Crude oil spectral signatures and empirical models to derive API gravity

      2019, Fuel
      Citation Excerpt :

      Some of crude oil absorption features prevail even if they are mixed with soil or water. The main absorption features that remain in the mixture with soil are found near 1200 nm, 1725 nm, 1760 nm, 2310 nm, 2350 nm, 3390 nm, 3424 nm and, 3507 nm [90–95]. These absorptions are observed even with a low level of crude oil in the soil (1% of contamination) [36].

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

      2019, Science of the Total Environment
      Citation Excerpt :

      Liquid and gaseous PHCs can be remotely sensed and detected directly on the surface (Brekke and Solberg, 2005; Cloutis, 1989; Hörig et al., 2001; Salem et al., 2005) or indirectly by measuring their effect on local rocks, soils and vegetation (Almeida-Filho, 2002; Noomen et al., 2012; Sanches et al., 2013b; Van der Werff et al., 2008). PHC detection methods include the study of their reflected spectra in the visible and near-infrared (VNIR) and shortwave infrared (SWIR) wavelengths (350–2500 nm) (Andreoli et al., 2007; Lammoglia and Souza Filho, 2012; Schwartz et al., 2012), and emitted spectra in the thermal infrared (7000–14,000 nm) (TIR) region (Emengini et al., 2013; Lyder et al., 2010; Van der Meijde et al., 2013). Infrared reflectance spectroscopy and imaging spectroscopy played an important role in prospecting and exploration of new PHC reservoirs (Ellis et al., 2001; Huang et al., 2010; Lammoglia and Souza Filho, 2011; Van Der Meer et al., 2002; Xu et al., 2007; Yatabe and Fabbri, 1986).

    • Spectroscopic characterization of red latosols contaminated by petroleum-hydrocarbon and empirical model to estimate pollutant content and type

      2016, Remote Sensing of Environment
      Citation Excerpt :

      These spectral features have been used by several authors to directly detect soils contaminated with PHCs (Chakraborty et al., 2014; Hörig, Kühn, Oschütz, & Lehmann, 2001; Okparanma & Mouazen, 2013b). RS and IS data and techniques have been used extensively in petroleum exploration (Chung, Choi, & Ku, 1999; Ellis, Davis, & Zamudio, 2001; Hörig et al., 2001; Kühn, Oppermann, & Hörig, 2004; Lammoglia & Souza Filho, 2011; Li et al., 2010; Lyder, Feng, Rivard, Gallie, & Cloutis, 2010; Noomen et al., 2008; Rivard et al., 2010; Smith et al., 2004; Souza Filho et al., 2008; Van Der Meer, Van Dijk, Van Der Werff, & Yang, 2002; Xu, Ni, Jiang, Jiang, & Chi, 2007) and environmental monitoring (Ayotamuno, Okparanma, Nweneka, Ogaji, & Probert, 2007; Kokaly et al., 2013; Lammoglia & Souza Filho, 2015; Noomen, Hakkarainen, van der Meijde, & van der Werff, 2015; Okparanma, Ayotamuno, Davis, & Allagoa, 2011; Okparanma & Mouazen, 2013a; Salem, Kafatos, El-Ghazawi, Gomez, & Yang, 2005; Van der Meer, Van der Meijde, Kooistra, Van der Werff, & Noomen, 2006; Van der Meijde et al., 2013; Winkelmann, 2005). However, specific studies to identify leaks and spills are rare in the literature.

    • Evaluating the feasibility of multitemporal hyperspectral remote sensing for monitoring bioremediation

      2015, International Journal of Applied Earth Observation and Geoinformation
      Citation Excerpt :

      As this study only focuses on vegetated areas, both hyperspectral datasets were divided into non-vegetated and vegetated subsets, using an NDVI (Normalized Difference Vegetation Index; Rouse et al., 1974) threshold of 0.4. Areas that have NDVI < 0.4 require other techniques to detect hydrocarbon pollution (e.g., Kühn et al., 2004; Van der Meijde et al., 2013). After masking all areas with NDVI < 0.4, 17 locations where groundwater monitoring was done in both years were left in the areas with NDVI > 0.4.

    View all citing articles on Scopus
    1

    Presently at: Research and Development Group, South African National Space Agency (SANSA) – Earth Observation Division, P.O. Box 484, Silverton, Pretoria 0127, South Africa.

    View full text
    Copyright © 2012 Elsevier B.V. All rights reserved.