Rapid On-Site Determination of the Total Petroleum Hydrocarbon Content of Soils by Handheld Fourier Transform Near-Infrared Spectroscopy: Development of a Global, Site- and Scanner- Independent Calibration Model

Abbreviations

TPH: Total Petroleum Hydrocarbon; NIR: Near-Infrared (Spectroscopy); MIR: Mid-Infrared (Spectroscopy); FT-NIR: Fourier Transform Near-Infrared; LW-PLS: Locally Weighted Partial Least Squares; GC-FID: Gas Chromatography–Flame Ionization Detection; RMSEC: Root Mean Square Error of Calibration; RMSECV: Root Mean Square Error of Cross Validation; RMSEP: Root Mean Square Error of Prediction; LVs: Latent Variables.

Synopsis

The described technique of handheld NIR spectroscopy allows the rapid, on-site determination of TPH in soil independent of the geographical location of the soil and the serial number of the specified NIR scanner with prediction errors < 1500 ppm.

Introduction

Despite extensive safety protocols, road transportation of hydrocarbon compounds can result in accidental spills, leading to terrestrial soil contamination [1, 2]. In such scenarios, a rapid, on-site method is critical to determine the extent of contamination, guide appropriate remediation strategies, and monitor their effectiveness [3]. Similarly, in remote exploration studies, rapid determination of the total petroleum hydrocarbon (TPH) content in soils is essential for evaluating the site’s environmental status and economic potential. However, traditional laboratory-based techniques—whether spectroscopic or chromatographic— are not suitable for these requirements due to several limitations.

Conventional chromatographic methods, while precise, are time-consuming, require elaborate sample preparation, depend on stationary lab infrastructure, and generate hazardous organic solvent waste. Likewise, even laboratory- based mid-infrared (MIR) and near-infrared (NIR) spectroscopy methods, which are capable of quantifying total carbon, nitrogen, moisture, and hydrocarbon contaminants [4, 5, 6, 7, 8, 9] without extraction, still necessitate the transport of soil samples to centralized facilities thereby introducing delays.

To overcome these constraints, feasibility studies have explored the use of portable MIR and NIR instruments for in-situ TPH quantification [1, 10, 11]. Compared to MIR spectroscopy, recent technological advances have significantly improved the practicality of handheld NIR spectroscopy. Lightweight (<500 g), cost-effective devices now offer optimized ergonomics and operational simplicity, making them especially attractive for field deployment [12]. Importantly, NIR spectroscopy is less susceptible to spectral interference from water than MIR, where hydrocarbon- specific ν(CH) stretching vibrations overlap strongly with broad ν(OH) water absorptions—an issue that complicates quantitative MIR-based TPH analysis in moist soil samples.

In the present study, we employ NeoSpectra MEMS- based FT-NIR instruments [13] equipped with a single-chip Michelson interferometer in combination with a rotating- dish sampling accessory. This configuration significantly facilitates sample presentation and helps average out soil heterogeneity. The spectrometer spans a broad wavelength range (1350–2500 nm; 7400–4000 cm-1), encompassing key hydrocarbon absorption bands such as the ν(CH2/ CH3) + δ(CH2/CH3) combination bands (4500–4000 cm-1), first overtones of 2ν(CH/CH2/CH3) (6200–5400 cm-1), and 2ν(CH2/CH3) + δ(CH2/CH3) combination bands (7500–7000 cm-1) [14]. The latter region was excluded from calibration due to interference from the first overtone of the 2ν(OH) water absorption.

To develop a robust, site-independent calibration model, the locally weighted partial least squares (LW-PLS) approach was applied, as described in previous studies [15, 16] and illustrated for different agronomic data sets by Lesnoff, et al. [17]. To minimize the impact of device-specific variations in hardware production on model performance, we integrated spectra from eight auxiliary scanners in addition to the master unit. These spectra were used for both calibration and validation, ensuring broader model applicability and prediction accuracy across different instruments.

Conclusions

In this work, the development of a site- and scanner-independent LW-PLS calibration model for the determination of Total Petroleum Hydrocarbons (TPH) in soil using a handheld NIR spectrometer has been successfully demonstrated. Site-independence was achieved by incorporating spectral data from samples collected at six geographically diverse sites across three continents into the calibration model. Furthermore, scanner-independence was ensured by replacing the conventional PLS algorithm with a locally weighted PLS (LW-PLS) approach and integrating spectra acquired from eight auxiliary scanners in addition to the master device.

This comprehensive strategy effectively addressed both sample variability and hardware-induced spectral variation, enabling the creation of a robust global calibration model. As a result, unknown soil samples can be accurately analyzed at any location, using different scanners, with an average absolute prediction error of less than 1500 ppm TPH.

From the perspective of international environmental regulations, this level of predictive accuracy is highly relevant. Regulatory thresholds for TPH in soil vary depending on jurisdiction, land use, and hydrocarbon fraction, but typical remediation or risk assessment trigger levels range from 100 to 3000 ppm. A prediction error of <1500 ppm is therefore well within an acceptable range for industrial applications and may also support preliminary screening in more sensitive residential or environmental settings.

These findings support the practical deployment of handheld NIR spectroscopy, enhanced by LW-PLS modeling, as a reliable and scalable tool for global soil contamination monitoring.

Funding

This research was funded by the Chevron Technology Center.

Conflict of Interest

The authors declare no competing financial interest.

Acknowledgements

The authors gratefully acknowledge the generous instrumental support from Si-Ware Systems, Heliopolis, Cairo, Egypt. The authors also acknowledge the Chevron Technology Center for providing the samples and giving permission to publish this article.