Earth observation guideKinds of data

Thermal imaging

A thermal sensor measures the heat the ground gives off, so it works by night as well as by day.

The basics

A thermal sensor ignores reflected sunlight and measures the heat the ground gives off, so it works at night as well as by day. It is the same idea as a thermal camera that shows a warm hand against a cold wall. From orbit it shows how hot a roof, a field, a street or a chimney is, and from that it can reveal crops short of water, hot spots in a city, active fires and volcanic activity. Its pictures are coarser than optical ones, because heat radiation carries little energy and a sensor has to collect over a wider area to see it.

Why it matters

Heat tells you things colour cannot. It shows crops short of water, the parts of a city that stay hot, active fires and volcanic activity. Because the sensor needs no sunlight, it can image at night as well as by day.

Warm houseCool pondTreeThermal sensor, at night
A thermal sensor sees warmth, not colour. At night a house still glows warm and a pond reads cool.

In depth

Where the infrared bands sit

Infrared is a family of bands with different physics. The near infrared (0.7 to 1.4 micrometres) and the shortwave infrared (1.4 to 3 micrometres) carry reflected sunlight, and they are treated like the visible bands. The mid-wave infrared (3 to 5 micrometres) and the thermal infrared (8 to 14 micrometres) carry radiation that the surface emits itself. The two thermal ranges are atmospheric windows. The gap between them, from 5 to 8 micrometres, is largely closed by water vapour absorption [1, 2].

Emission and temperature

Every body above absolute zero emits radiation. A perfect emitter, a blackbody at temperature TT, emits spectral radiance given by Planck's law [3], with the Boltzmann constant kk.

Bλ(T)=2hc2λ5 1exp⁡ ⁣(hcλkT)−1B_\lambda(T) = \frac{2 h c^{2}}{\lambda^{5}}\,\frac{1}{\exp\!\left(\dfrac{h c}{\lambda k T}\right) - 1}
(1)

where Bλ(T)B_\lambda(T) is the spectral radiance of a blackbody, in W m−2 sr−1 m−1\mathrm{W\,m^{-2}\,sr^{-1}\,m^{-1}} (divide by 10610^{6} for the value per micrometre); λ\lambda is the wavelength, in m\mathrm{m}; TT is the absolute temperature, in K\mathrm{K}; hh is the Planck constant, 6.626×10−34 J s6.626\times 10^{-34}\ \mathrm{J\,s}; cc is the speed of light, 2.998×108 m s−12.998\times 10^{8}\ \mathrm{m\,s^{-1}}; and kk is the Boltzmann constant, 1.381×10−23 J K−11.381\times 10^{-23}\ \mathrm{J\,K^{-1}}.

Figure 1 plots equation 1 for three temperatures. Each curve is divided by its own peak so that only position and shape are compared, since the Sun's total emission is vastly greater than the Earth's. The peaks move to shorter wavelengths as the temperature rises, following equation 3 on the light and the spectrum page. A fire near 1000 K peaks near 3 micrometres, at the short end of the mid-wave band, which is why that band is used to find fires [2], and the Earth peaks within the 8 to 14 micrometre window. Integrated over all wavelengths, the emitted power per unit area rises with the fourth power of temperature, with the Stefan-Boltzmann constant σ=5.670×10−8 W m−2 K−4\sigma = 5.670\times 10^{-8}\ \mathrm{W\,m^{-2}\,K^{-4}}.

M=ε σ T4M = \varepsilon\,\sigma\,T^{4}
(2)

where MM is the radiant exitance, the power emitted per unit area of surface over all wavelengths and directions, in W m−2\mathrm{W\,m^{-2}}; ε\varepsilon is the emissivity of the surface, between 0 and 1 without unit and equal to 1 for a blackbody; σ\sigma is the Stefan-Boltzmann constant, 5.670×10−8 W m−2 K−45.670\times 10^{-8}\ \mathrm{W\,m^{-2}\,K^{-4}}; and TT is the surface temperature, in K\mathrm{K}.

3 to 58 to 140.3131030Wavelength (µm)Sun, 5772 K1000 KEarth, 288 K
  • Vertical axis: emission, each curve scaled to its own peak
  • Shaded: the two atmospheric windows, in µm, where thermal sensors look
Figure 1. Planck curves for three blackbody temperatures, each scaled to its own peak. The shaded bands are the 3 to 5 and 8 to 14 micrometre atmospheric windows in which thermal sensors operate.

Real surfaces are not blackbodies. Their emissivity ε\varepsilon, between 0 and 1, depends on material, moisture and roughness, and it multiplies the blackbody value. A sensor measures radiance LL, and the simplest product inverts Planck's law for the temperature a blackbody would need to produce it, the brightness temperature TbT_b.

Tb=hcλkln⁡ ⁣(1+2hc2λ5L)T_b = \frac{\dfrac{h c}{\lambda k}}{\ln\!\left(1 + \dfrac{2 h c^{2}}{\lambda^{5} L}\right)}
(3)

where TbT_b is the brightness temperature, in K\mathrm{K}; LL is the spectral radiance measured in the band, in W m−2 sr−1 m−1\mathrm{W\,m^{-2}\,sr^{-1}\,m^{-1}} like BλB_\lambda; λ\lambda is the central wavelength of the band, in m\mathrm{m}; and hh, cc and kk are the constants of equation 1.

The brightness temperature is lower than the true surface temperature when ε<1\varepsilon < 1, and the atmosphere adds its own emission and absorption. Retrieving the land surface temperature therefore needs an estimate of emissivity and a correction for the atmosphere, done either with two neighbouring thermal bands (the split-window method) or with one band and atmospheric data. The methods and their accuracy are reviewed in the literature [4, 5, 6].

References

  1. Rees, W. G. (2012) Physical Principles of Remote Sensing. Cambridge University Press. https://doi.org/10.1017/CBO9781139017411
  2. Kuenzer, C. and Dech, S. (2013) Thermal Infrared Remote Sensing. Remote Sensing and Digital Image Processing. https://doi.org/10.1007/978-94-007-6639-6
  3. Planck, M. (1901) Ueber das Gesetz der Energieverteilung im Normalspectrum. Annalen der Physik, 309(3), 553-563. https://doi.org/10.1002/andp.19013090310
  4. Dash, P., Göttsche, F. M., Olesen, F. S. and Fischer, H. (2002) Land surface temperature and emissivity estimation from passive sensor data: Theory and practice-current trends. International Journal of Remote Sensing, 23(13), 2563-2594. https://doi.org/10.1080/01431160110115041
  5. Jiménez‐Muñoz, J. C. and Sobrino, J. A. (2003) A generalized single‐channel method for retrieving land surface temperature from remote sensing data. Journal of Geophysical Research: Atmospheres, 108(D22). https://doi.org/10.1029/2003JD003480
  6. Li, Z. L., Tang, B. H., Wu, H. et al. (2013) Satellite-derived land surface temperature: Current status and perspectives. Remote Sensing of Environment, 131, 14-37. https://doi.org/10.1016/j.rse.2012.12.008