Sentinel-1A GRD product

Thanks Peter. That’s very helpful.

Hello! I’m working with IW GRDH images.

do you guys know if I need to calibrate (in sigma0 values) before or after applying the speckle filter?

I’ve been told that I cannot apply the filter in a image with sigma0 values. Is that true?

Anyway, do I need to calibrate it at all?

And, at last, when I calibrate, the sigma0 band comes in linear scale, right? I wanted them in dB scale, but it is not, since there is no negative value when I choose to see the pixel info.


Calibrate first.

Linear to/from dB conversion can be done afterwards.

Ok, so the right order would be:

Calibrate with Sigma 0 values
Speckle filter
Terrain Correction
Convert from linear to dB.
At the step 3, should I mark the box of apply radiometric normalization? Considering that my data is already in sigma 0…


Oh… and do I need to apply orbit file before calibrating?


when you have calibrated the data and you already have Sigma0 you don’t need to select the “radiometric normalization” in the Terrain Correction module.

Application of the orbit files is not mandatory but often improves the geometric accuracy of your product. It should be done at the very beginning.

  1. Orbit File
  2. Calibrate to Sigma0
  3. Speckle Filter
  4. Terrain Correction
  5. Convert to dB


where will you put thermal noise correction in above sequence?


I am talking about “IW GRD” data.

as early as possible I’d say.



I have to classify different types of crops. Should i do terrain filtering and terrain correction .

For what kind of specific purpose we use these two operation?


Speckle filtering reduces noise-like patterns in your image.
Terrain correction reduces gemetric distortions due to the side-looking geometry of your system.

How would i know about an image whether it has thermal noise or not?


in fact, it always has some. If you don’t de-thermal noise, the image is ok although this step would improve performance in most case

is it necessary to convert sigma0 values to dB ? can i do log transformation on my own e.g in R ?

yes, you can do that later as well. Also it is not mandatory, it just stretches radar backscatter over a more usable range which have nearly a gaussian distribution.


I am getting only positive value of sigma nought (db) in VV intensity (my image is S1A GRD L1 product)

Can you please share a histogram of your data?

Hi Iveci,
Your explaination is very useful.
I have another question for GRD product preprocessing. Since the sigma 0 is in intensity scaling, if Iwant to get the radiometrically calibrated value of amplitude, can I directly calculate the square root the sigma 0 values?

Many thanks for your reply!

Hi, All radiometric measurements from detected SAR imagery (such as Sentinel-1 GRD products) should use intensity values and not amplitude. So radiometrically calibrated value of amplitude is not a physical quantity and it should not be used.

FYI, the only reason why detected products include amplitude values is so that the dynamic range of the processed SAR signal can fit into the 16bits of the product. If intensity values were included in the product, 32 bits would be required per pixel and so the product would be twice the size.


I understand, thank you very much!