top of page

Why is a VNIR camera useful on a TIR mission?

Nov 3
2 min read

Having a VNIR camera on the same satellite can help with Thermal Infrared (TIR) camera calibration challenges.


Geometric improvements

If you have both VNIR and TIR cameras on the same satellite, the VNIR imagery can be geolocated against a VNIR reference. FarEarth can then correlate TIR imagery with your geolocated VNIR imagery. This way, TIR no longer needs its own dedicated reference. This can help solve the reference data gap for your TIR cameras (see previous blog)

Example of good correlation between VNIR (left) and TIR (right) with tiepoints.
Example of good correlation between VNIR (left) and TIR (right) with tiepoints.

Radiometric improvements

Most small satellites only have one TIR band. In this case, you can't separate temperature from surface emissivity using multi-band analysis because it needs at least two thermal bands. Without this, emissivity is usually assumed to be a single fixed value across the whole scene, while in reality it varies from one surface to the next.


An alternative is to use the near-infrared band on the VNIR sensor to calculate an NDVI-based emissive estimation. Our example below shows the estimated emissivity based on NDVI. In this scene, the value over water, for example, is 0.95, but below 0.9 for desert areas.

Landsat-9 RGB image (left) and emissivity estimate based on NDVI (right).
Landsat-9 RGB image (left) and emissivity estimate based on NDVI (right).

AOT estimation

Historical atmospheric data or weather forecasts are often used to estimate aerosol optical thickness (AOT) during processing. Historic values, however, are not an accurate reflection of the conditions at the time of the acquisition.


VNIR data can also be used to estimate AOT at a point in time without the need for a separate reference dataset.


Having VNIR onboard can therefore allow for more reliable bottom-of-atmosphere (BOA) processing of TIR imagery.

The example below illustrates the effect of aerosols on an RGB image. TIR images are affected similarly.

Example of a top-of-atmosphere (TOA) image (right) compared to a bottom-of-atmosphere (BOA) product. Notice how the atmosphere makes the image look bluer due to blue light scattering.
Example of a top-of-atmosphere (TOA) image (right) compared to a bottom-of-atmosphere (BOA) product. Notice how the atmosphere makes the image look bluer due to blue light scattering.

An experiment by FarEarth

We processed Landsat-9 thermal images in two different ways – with and without the help of VNIR. Look at the real-world impact:

  1. The image on the left uses both emissivity and AOT estimated from VNIR data.

  2. The image on the right uses fixed default values for both: 0.95 emissivity and an AOT of 0.125.

The difference between the two results is 3°C, enough to be the difference between a healthy crop and one in need of attention. 

Landsat Band 10 TIR image processed using AOT and emissivity retrieved from reflective data (left) and processed using a fixed AOT and emissivity (right).
Landsat Band 10 TIR image processed using AOT and emissivity retrieved from reflective data (left) and processed using a fixed AOT and emissivity (right).

To recap

Having a VNIR camera onboard a TIR mission can be very useful for calibration. VNIR data can be used to improve the geolocation accuracy of thermal imagery. It can also be used to estimate emissivity and aerosol optical thickness to improve radiometric accuracy.

 
 
bottom of page