The story Infrared camera shows the impact of vehicle emissions says:

In an effort to visually demonstrate the potentially fatal impacts of air pollution, FLIR Systems (Wilsonville, OR, USA; www.flir.com) has released a new video containing footage captured by FLIR infrared cameras that shows vehicle emissions and the resulting pollution issues.

[...] To accomplish this, FLIR used its GF320 infrared camera to capture video of the emissions. This camera features a 320 x 240 cooled InSb infrared detector with a spectral response of 3.2 to 3.4 μm. Designed specifically for gas leak detection and electrical inspection, the camera also embeds GPS data into the image, allowing workers to pinpoint the location of the leak or hot spot.

Wikipedia says InSb is a narrow-gap semiconductor with an energy band gap of 0.17 eV at 300 K and 0.23 eV at 80 K and a focal plane array can image light with a wavelengths in the 1 to 5 micron range. They are often cooled to reduce dark current for better sensitivity. In this case it sounds like they've added an interference filter to selectively pass 3.2 to 3.4 microns.

My question is about what it is they are really imaging here. It is true that gases such as CO2 and methane have strong IR bands, this is why they are important greenhouse gasses. Are they measuring a temperature difference or an emissivity difference or light scattering or absorption or...? How does this technique actually work?

enter image description here

above: Image from this article.

  • 2
    $\begingroup$ As far as I know, the relevant IR absorption of CO$_2$ is due to the bending motion at about 600 cm$^{-1}$ - and it does overlap with the heat radiation of the Earth. For methane, it's the peak at 1300 cm$^{-1}$, the motion can be seen here $\endgroup$
    – Szgoger
    Aug 8, 2022 at 13:47
  • 3
    $\begingroup$ Methane does have a strong absorbance in the 3-4 micron range: webbook.nist.gov/cgi/… $\endgroup$
    – Andrew
    Aug 8, 2022 at 14:53
  • 2
    $\begingroup$ This paper discusses methane detection by FLIR cameras in detail. It is indeed based on absorbance of the background IR in the 3.2-3.4 micron window: pubs.acs.org/doi/10.1021/acs.est.6b03906 $\endgroup$
    – Andrew
    Aug 8, 2022 at 15:26
  • 3
    $\begingroup$ I wouldn’t necessarily assume it’s the same temp as surroundings, but whatever the temp, the re-emission will be much broader spectrum than the absorbance, so there is still a significant decrease in observed transmission at the measured wavelengths. $\endgroup$
    – Andrew
    Aug 9, 2022 at 11:53
  • 1
    $\begingroup$ Slight correction - I read more of the paper, and they say that both the emissivity of the methane and the absorbance of background are important contributors to the signal depending on the situation. A hot methane plume is visible against a cold sky because of the plume's higher emissivity, whereas a cold plume is best observed against high background emissions because of the absorbance of the plume. At least that's how I'm reading it. But you should read the paper yourself rather than trust my quick interpretation. $\endgroup$
    – Andrew
    Aug 9, 2022 at 12:53

1 Answer 1


Short answer

The cameras you are asking about are called optical gas imaging (OGI) cameras, they are not thermographic cameras per se, but are a modified kind instead. However, both use the same principles to actually detect the infrared light, the key difference is the incident radiation is filtered to allow only the wavelengths that interact with the methane gas plume (or other analyte gas) to interact with the camera sensor. Thermographic cameras could be configured to take both kinds of images since the principle of their operation is essentially the same, but they are usually separate devices for practical reasons like cost, increasing image contrast, etc.. In the images you provided, you can see the solid objects because they emit, reflect, and absorb infrared light across the spectrum, so they can also be seen by the OGI cameras. However, other gases are not seen because the IR light they absorb is filtered out. Finally, a temperature difference must exist between the plume of gas being imaged and the background scenery because otherwise the amount of infrared radiation emitted will be equal to the amount absorbed by the gas, following Kirchoff's Law of thermal radiation.

Detailed Answer

Methane detection colour analogy

Cameras for visible light discriminate between colours by placing a filter before the sensor so that only one colour (i.e., red, green, or blue) is absorbed by the sensor at any given pixel. The sum of the RGB signal reproduces the subject's colour. Although we cannot see them with our eyes, infrared light has different "colours" that can be discriminated by a camera sensor with the appropriate absorption range, filters, etc.. Visible light has different colours depending on the energy (i.e., wavelength or frequency) of the light: red light has a longer wavelength (lower energy and frequency) than green, and blue, respectively. We can see an apple is red when the green and blue light in white backlight are absorbed by its skin while the red light is reflected into our eyes (or camera). In the same way, we can see the infrared "colour" of different objects by observing the infrared light they absorb. FTIR spectrum of CO2 Figure 1 FTIR spectrum of carbon dioxide from NIST. enter image description here Figure 2 FTIR spectrum of methane from NIST, notice the lack of overlap between this spectrum and that of carbon dioxide (above).

Each arrangement of atoms in a molecule with different nuclear charges (i.e., polar bonds) absorbs a specific wavelength of light because the vibrations are quantum mechanical in nature. These vibrations are characteristic to each molecule, so we can identify them by observing which wavelengths of light they absorb. See the above two Fourier-transformed infrared (FTIR) spectra of carbon dioxide and methane. This is the basis of spectroscopy, which is out of the scope of this answer. But, if we designing a sensor that can discriminate between certain wavelengths of infrared light, a camera can be made that can see how much of a particular wavelength (e.g., 3.5 μm for methane, 10.57 μm for sulfur hexafluoride) is being absorbed relative to the background. Couple this principle to a system that can tell where the light is coming from, i.e. add pixels, add some optics to focus the light, and you have a camera.

Optical gas imaging (OGI)

The objects in the field of view (FOV) in OGI are illuminated by the background infrared radiation emitted from the sun and all other warm objects. This is why OGI is also called a "passive" thermal imaging technique, in contrast to "active" techniques where the scene is illuminated with an artificial light source directed at the subject. OGI depicts the relative decrease (or increase) in the intensity $\Delta I(\lambda)$ at a certain wavelength $\lambda$ of this background radiation $I_B(\lambda)$ due to absorption (or emission) by the analyte gas with intensity $I_G(\lambda)$. A general OGI equation (Eq. 1) for $\Delta I$ can be derived using a radiative transfer model:

$\Delta I(\lambda) = I_b(\lambda) - I_g(\lambda) = [1 - \tau_g(\lambda)][B(T_b,\lambda) - B(T_g,\lambda)]\quad(1)$

where $\tau_g$ is the transmission coefficients of the analyte gas, and $B(T_b,\lambda)$ and $B(T_g,\lambda)$ are the Planck law functions for blackbody radiation of the background at temperature $T_b$ and the analyte gas at temperature $T_g$, respectively.

We can obtain the transmission coeffecient of the gas plume in relation to its composition, concentration, and geometry using Beer's law (Eq. 2)

$\tau_g(\lambda) = \exp[-\alpha_g(\lambda)\bar{C}_g \ell_g]\quad\quad\quad (2)$

where $\alpha_g$ is the gas's absorption coeffecient, $\bar{C}_g$ is the average concentration of the gas through the line of sight, and $\ell_g$ is the breadth of the gas plume through the line of sight.

Now we can see that the gas plume image obtained by OGI systems requires three things to be detectable:

  1. Absorption (or emission) of infrared radiation in a certain band of wavelengths ($\lambda$);
  2. Plume that is big enough ($\ell_g$) or concentration that is high enough ($\bar{C}_g$) to absorb (or emit) enough infrared radiation to be detected; and
  3. Temperature difference $\Delta T = T_b - T_g$ that is great enough between the background and the gas plume so they absorb (or emit) different amounts of infrared radiation (Eq. 3).

Each of these three factors contribute to the overall contrast observed in the image between areas where there is a gas plume and where there is not. The final point may be mysterious to the reader, but it is due to Kirchhoff's law of thermal radiation. Put simply, if the gas were the same (apparent) temperature as the object behind it (in the line of sight), then it would absorb as much infrared radiation as it was emitting, and would appear invisible to the OGI camera. If the gas plume has a lower temperature ($T_b > T_g$), it will absorb more infrared radiation as it passes through it to the camera relative to the atmosphere around it, thus $\Delta I < 0$ and the gas plume will appear darker. If the gas plume has a higher temperature ($T_b < T_g$), then $\Delta I > 0$ and the gas plume will appear brighter.

For more detail on OGI and its quantitative variant (QOGI), please see this recent dissertation by Michael Nagorski out of Waterloo.

Commercial OGI cameras

Teledyne FLIR LLC, the company you specifically are referring to in your question and who appear to make the industry standard cameras, describe the operation of their OGI and thermographic cameras and sensors in some detail in US Patent No. 9,276,161 B2 and 9,007,687 B2. They have invented arrays of quantum wells of indium antimonide (InSb) that are specially designed to absorb radiation as close to a strong absorption band in the gas plumes that are to be imaged. State-of-the-art camera systems include advanced optics and filters to improve the signal-to-noise ratio, thus increasing the contrast of the acquired image. Directly from US Patent No. 9,276,161 B2:

Generally, the camera system 100 is used for detecting a gaseous compound in the scene 106 when radiation from the scene 106 is received by the lens 104 and passed to the band pass filter 110 to limit the wavelength range of survey scene energy focused onto the FPA 108. In one embodiment, the wavelength range is limited to 10.3 to 10.8 μm. Other wavelength ranges are also contemplated. Each sensor element of the FPA 108 generates an analog photo current value according to a photo current responsivity profile and other factors in response to an irradiance generated by the spectrally filtered scene image formed by the lens 104 at the sensor element active surface. The analog photo current values are read out from each photo sensor element and converted to corresponding digital signal values for rendering a video image frame corresponding to the digital signal values.

In the aforementioned patents they give some details on the ways they have improved upon the principles laid out above.

  • 1
    $\begingroup$ Thanks & Welcome to ChemSE! If it's absorption only, then a methane could would always appear as a simple darkening of the background behind (if thin) or big black cloud if thick. And yet, the images shown in these videos (at these times) look much more complex than that. 1, 2, 3 In the third video for example, the plume from the stack is hotter than the sky behind it, not colder. In this case it seems to be radiating more strongly than the background, not absorbing light from it. $\endgroup$
    – uhoh
    Dec 1, 2022 at 22:05
  • 2
    $\begingroup$ You can just invert the image to make absorption (black) look like emission (white), that is entirely a perceptual difference. Besides, the plume should appear "hotter" than the background because the methane concentration is higher coming out of the engine than it is in the atmosphere. The intensity of the image is due to the concentration of methane which is inversely proportional to the transmission of the infrared light absorbed by methane. $\endgroup$
    – gbacic
    Dec 1, 2022 at 22:56
  • 2
    $\begingroup$ I should also clarify that the image has almost nothing to do with the temperature of the gas being observed. The amount of IR light that is absorbed by the methane is mostly determined by its quantum mechanical properties, and temperature has a negligible effect on these properties. The density of the gas will change as it gets hotter which will change the number of molecules per unit volume, but it won't change how much each molecule themselves absorb. The shape of the plume will also affect the amount of absorption as there is more gas in the middle of a plume than the outside. $\endgroup$
    – gbacic
    Dec 1, 2022 at 23:01
  • 1
    $\begingroup$ Okay I'm familiar with how these cameras work (inside and out) and that the displayed "temperature" is a computed value based on brightness and assumed emissivity and not a real temperature meaasurement. In the third video there is a color scale, we can see that the sky is radiating the least thermal IR and the ground the most thermal IR. The color scale is monotonic, and the intensity recorded from the plume is higher than the background sky. The plume is radiating, not absorbing. Perhaps it is just hot gas not methane. Perhaps this method does not actually specifically flag methane at all! $\endgroup$
    – uhoh
    Dec 1, 2022 at 23:05
  • 1
    $\begingroup$ To avoid confusion, let's not use the word thermal when referring to these images. Thermal IR imaging measures the amount of light that is emitted per unit area due to blackbody radiation. For FLIR imaging we are representing only the amount of light absorbed by a specific wavelength that is characteristic of methane. Thermal IR images would be like a grayscale image while FLIR images would be a colour filter applied to the thermal IR image. In reality, you only collect the light absorbed by methane to increase the signal-to-noise ratio rather than filter all the light afterward. $\endgroup$
    – gbacic
    Dec 1, 2022 at 23:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.