Use of Hyperspectral Imaging for Pollution Detection

Spread the knowledge
  •  
  •  
  •  
  •  

Human vision undeniably aids us in navigating the world. Unfortunately, it only partially assists us in identifying the chemicals that constitute it. Our vision may be sufficient for us to watch the changing colors of tree leaves as winter comes, but we are unfortunately not able to watch the greenhouse gases spewing from cars and planes into the atmosphere. In the modern world, with its pollution, biocides, greenhouse gases, and other synthetic or over-abundant chemicals, it has become necessary for humans to know the content of the environment around them if they are to survive. In this blog post, I will discuss hyperspectral imaging and how it can be used to image the chemical composition of things in the natural and human-made worlds.

The molecular structure of methane: the black atom is carbon and the grey atoms are hydrogen. Methane warms the atmosphere more than carbon dioxide, the most common greenhouse gas, but it has a shorter atmospheric lifetime. From Wikipedia: Methane page August 2020.

Each molecule has a distinctive absorption ‘fingerprint’ which can be used to identify it. When light passes by a particular kind of molecule, the light of several particular wavelengths will be absorbed. If a scientist monitors the spectrum of light shining on a molecule and the spectrum of the light after it passes through the molecule, the scientist can determine the identity of the particular molecule. There are two reasons that humans cannot clearly observe these spectra. The first is that the human eye only observes three different primary colors: red, green, and blue. Many more observations are required to resolve the absorption peaks. Second, humans observe light only in the visible range. Many molecular absorption signatures are much more prominent in the infrared range.

Figure 3
Absorption spectrum of methane in the near-infrared range from Nelson, Plyler, and Benedict 1948 [1]. Note that the sharp absorption peaks appear as minima in spectrum.

Hyperspectral imaging is a technology that measures light at many wavelengths, typically in more than 100 bands. In contrast to commercially-available red-green-blue digital cameras, hyperspectral cameras can spatially-resolve the unique spectral signatures of different molecules. Their spectral sensitivity has made them ideal for identifying the gas plumes.

One of the most prominent successes of hyperspectral imaging has been identifying methane. In 2014, scientists funded jointly by the US National Aeronautics and Space Agency (NASA) and the European Space Agency (ESA) operated the COMEX (CO2 and Methane Experiment) mission. In this mission, the airborne AVIRIS hyperspectral imager recorded the southern California landscape, while ground-based inspectors visited locations such as landfills and fossil-fuel production sites in order to verify the locations of methane plumes [2]. The mission was able to identify several methane sources, and a few years later, the techniques developed in the mission were used to survey methane emitters across the state of California [3]. Techniques like this may help politicians and civil servants evaluate whether the communities they represent are meeting their emissions goals to protect the environment and public health.

At least one company has been formed to use hyperspectral imaging to search for methane. GHGSat was founded in 2016 in Quebec, Canada to build satellites to monitor methane emissions [4]. The plan is for GHGSat to sell their remotely-sensed hyperspectral data to companies for monitoring their pipelines, factories, refineries, etc. I hope that this business model works because it would be thrilling to see environmental research become self-funding.  The company had at least one notable public success, when, using the data from its GHGSat-D satellite, it identified a large methane leak in Turkmenistan, which was subsequently mitigated [5]. The development of GHGSat as a company will certainly be interesting to watch. If it proves the value of hyperpectral gas monitoring, other companies will be likely to follow. Perhaps even ground-based measurements will be economically feasible.

Beyond methane, gases such as carbon dioxide, carbon monoxide, and sulfur hexafluoride can also be detected with hyperspectral imaging [6,7]. The latter two gases, in addition to being potent greenhouse gases, can be dangerous to humans. Perhaps there is a possibility for more extensive environmental monitoring with hyperspectral imaging? The development of GHGSat as a company will certainly be interesting to watch. If it proves the value of hyperpectral gas monitoring, other companies will be likely to follow. Perhaps even ground-based measurements will be economically feasible.

The chief economic constraint on hyperspectral imaging, however, is the cost of the detectors themselves. Hyperspectral imagers that work in the visible range can use the same silicon photodetectors as normal digital cameras. One of the scientists I work with has even released a do-it-yourself diagram for 3-D printing a hand-held hyperspectral camera [8]. However, the absorption spectra of many molecules are more unique, and thus easier to detect, in the infrared regime relative to the visible regime. Semiconductor detectors for short-wave infrared radiation include Indium Gallium Arsenide and Mercury Cadmium Telluride, both of which are much more expensive than silicon and the latter of which must be cooled. One manufacturer has recently started offering a cheaper, quantum-dot based detector [9], and several members of my old lab did some work on extending the spectral range over which silicon is sensitive [10]. Some scientists have also developed a hyperspectral camera that relies on microbolometers rather than semiconductors for detection (unlike the detectors mentioned above) [11].

It is my hope that the cost of small, portable hyperspectral imagers will soon be reduced enough to put them in the hands of small businesses and citizen scientists. In the hands of scientists, these imagers will allow communities and other interested members of society to know what chemicals are present in the world around them and will better allow them to navigate and advocate for necessary change in the modern world.

References

[1] R. Nelson et al. “Absorption Spectra of Methane in the Near Infrared” 1948

[2] D. R. Thompson et al. “Real-time remote detection and measurement for airborne imaging spectroscopy: a case study with methane” 2015

[3] R. M. Duren et al. “California’s methane super-emitters” 2018; see also https://arstechnica.com/science/2019/11/californias-methane-super-emitters/

[4] www.ghgsat.com

[5] D. J. Varon “Satellite Discovery of Anomalously Large Methane PointSources From Oil/Gas Production” 2019, see also https://physicsworld.com/a/from-methane-emissions-to-space-weather-satellite-based-observations-forge-ahead/

[6] S. Sabbah et al. “Remote sensing of gases by hyperspectral imaging: system performance and measurements” 2012

[7] M. A. Gagnon et al., “Standoff Midwave Infrared Hyperspectral Imaging of Ship Plumes” 2016

[8] CAD Model: https://www.tinkercad.com/things/lq6HDDaggKV;  F. Sigernes et al. “Do it yourself hyperspectral imager for handheld to airborne operations” 2018; https://www.photonicsviews.com/hyperspectral-imagers-for-drones/

[9] www.swirvisionsystems.com

[10] L. Krayer et al. “Near-IR imaging based on hot carrier generation in nanometer-scale optical coatings” 2018; L. Krayer et al. “Optoelectronic Devices on Index-near-Zero Substrates” 2019; J. Kim et al. “Interfacial Defect-Mediated Near-Infrared Silicon Photodetection with Metal Oxides” 2019

[11] S. Sugawara et al. “Wide-field mid-infrared hyperspectral imaging of adhesives using a bolometer camera” 2017; J. Kilgus et al. “Application of a Novel Low-Cost Hyperspectral Imaging Setup Operating in the Mid-Infrared Region”; https://www.specim.fi/wp-content/uploads/2020/03/Spectral-Cameras-LWIR_ver1-19.pdf


Spread the knowledge
  •  
  •  
  •  
  •