Correct SAR calibration steps

Hello forum

I found that the tool SAR processing > Geometry > Terrain correction > range doppler terrain correction, is able to carry out the geocoding and the calibration of a SAR scene in just one step (or am I wrong?).

A very quick question: when is it better to apply first the SAR processing > radiometric > calibrate and then range doppler terrain correction?

Wich are the differences between the SAR processing > radiometric > calibrate and the range doppler terrain correction calibration procedures?

Thank you very much

Hello Giacomo,

The SAR processing -> radiometric -> calibrate process converts the pixel data to actual backscattering values. It should be performed first in whatever processing chain you do. If you need to do polarimetric work after with your data, you need to check the box marked «Save in complex» when calibrating. After this step, your image has backscattering intensity pixels and is in ground range or slant range geometry.

The SAR processing -> Geometry -> Terrain correction -> Range Doppler Terrain Correction process corrects the geometric distorsions in the image caused by the SAR system (slant view, etc) and yields a geocoded image. After this step, your image has backscattering intensity pixels and is in a map coordinate system. This step is usually done at the end of the processing, depending on your application.

Any multilooking or speckle filtering is usually done between these two steps.

All this is taken from the SAR basics tutorial .

I hope this helps you.


Could you please explain why the map projection set on WGS84 or UTM Zone 12 /WGS84 for the terrain correction would result in the NAD 27 when the terrain correction is done, for Rdarsat 2!!! I have tried it on several radarsat 2, still I get NAD 27!


I think the question of Giacomo is still open: Is there a difference in the outcomes regarding calibration results between

  • the radiometric calibration module
  • the radiomertic calibration within the Range-Doppler TC module

I wondered about this as well and think it has not been clearly answered.

1 Like

It depends on the mission. For ENVISAT and ERS the difference is that the local incidence angle is used instead of the incidence angle from the ellipsoid. For Sentinel-1 I since the calibration comes from a look up table, I believe there is still an open question on applying the local incidence angle.


First I have to apologize for the terrific delay of my reply. Sorry and thanks everybody on this forum who paid attention on my topic. I think that Louis answer could be applied to all the SAR sensors, I mean: if you want to calibrate your image using the ellipsoid as ground reference you have to apply Radiometric > Calibrate, instead if you want to use the “real” ground reference (accounting topography) you must use Geometry > range Doppler terrain correction.

My opinion is that YOU DON’T HAVE TO APPLY BOTH (or am’I wrong?), I mean: since Calibration and Range Doppler terrain correction are both able to carry out the calibration step (both procedures produce backscatter), you don’t have to this twice (don’t Calibrate and then Range doppler terrain correction).

Thank you

1 Like

yes, I think we all agree that double calibration would lead to nonsense results (maybe even leads to an error before executing).

Calibration and terrain correction are separate steps that should both be performed (calibrate first).

so what is the point of the calibration checkbox in the terrain correction module at all?

Range Doppler Terrain Correction does not have a checkbox for “calibration”, but there is a selection for “radiometric normalization”, which is not the same thing calibration. Calibration is a purely radiometric operation, while terrain correction is a geometric one that affects radiometry via geometric effects.

1 Like

you’re right, that seems totally legit.
What confuses me is that checking the normalization in the RD Terrain Correction module still produces Sigma0 from SLC intensities.

Ok, we know that in the correct ingegneristic language calibration doesn’t mean terrain correction that doesn’t mean normalization, but we are all S1TB users and it seems that there is a semantic issue (very important IMHO) in our software to be solved. I use to work with SAR data on totally flat area. Backscatter data produced with “calibration” is (almost) the same that the one produced with “terrain correction”. I think that terrain correction carry out also a proper inner calibratrion procedure.

1 Like

A question about radiometric calibration for Sentinel-1A data with VV and VH polarization using SNAP, the radiometric calibration of VH polarization is right, but the VV polarization is incorrect with the outputs of zeros.