Sentinel-1A GRD product

Hi everybody,

I have downloaded one Sentinel-1A Level 1 GRD product in zip format.

  1. What’s the unit of digital number (DN) value of an image
  2. What is the equations/formulas to convert that DN value into signma naught and amplitude?

I hope someone can answer it.

  1. Since the DN is roughly just the square root of the backscatter coefficient (aka normalized radar cross section or sigma nought), which is a dimensionless quantity, the DN itself is dimensionless as well.
  2. There calibration equations for Sentinel-1 are described in detail in this document and in the SNAP help as well.

There is, of course, no need to do the calibration yourself. You can use the Calibration operator in SNAP (Radar→Radiometric→Calibrate) to convert DN to sigma/beta/gamma nought. You might want to add the Radiometric Terrain Flattening operator in conjunction to it, too, to further improve the radiometric calibration by removing some topographic effects that are not compensated for in the Calibration step.


Thanks for your explaination. But little confused; what the following math gives? backscattering in db?

By the way, it is necessary to do noise filtering as well?

Yes that returns the dB.

Noise removal should already be applied to GRDs from the IPF. If you are creating a detected product from SLC then you may want to apply the noise removal.



Since toolbox’s help manual is not clear about how it performs terrain flattening. Can I know formula for terrain flattening using external DEM data?

1 Like

The radiometric terrain flattening procedure is described in detail in the article cited at the end of its help page:

David Small, “Flattening Gamma: Radiometric Terrain Correction for SAR imagery”, IEEE Transaction on Geoscience and Remote Sensing, Vol. 48, No. 8, August 2011


@valgur So, radiometric terrain flattening is also called slope correction, ya?

I don’t know what exact procedure you have in mind, but I guess you could call it that. You could also use that term for calculating sigma and gamma nought in the general sense as well, I think.

1 Like

@valgur @lveci So, can I do radiometric calibration and topographic correction after clipping the whole GRD image into ROI (polygon)?

1 Like

Yes, definitely. you should be able to use any SNAP operator after cropping
to a subset, I think.

1 Like

A subset GRD product should be no different in terms of what you can do with the product. The only place to be careful is with TOPSAR SLC products which should be handled by burst and swaths. The TOPSAR Split operator should be used for this.

1 Like

Why I can’t calibrate Level-1 GRD prouct?It appears Java heap space.But I think my computer have enough space.

That could be related to the amount of RAM in your system. 16GB is the recommended amount and the tool generally works well with 8GB. Less memory than that can lead to problems.

1 Like

After calibrating GRD-product to sigma-nought and calculating statistics i hadn’t found any negative values. Is it possible ? If [ valueDB = 10*log10(valueDN)] true, we can probably get negative value. How it could be explained ? Thanks.

The expected range for the radar cross-section (Sigma0) of distributed targets in Sentinel1 imagery is from about -25dB to about +1 dB. The lower value is for noise (usually over oceans under low wind conditions with HV or VH polarisation) while the upper value is typical for land ice regions such as edges of the Greenland ice sheet.

What types of sigma0 values are you obtaining and what type of surface are these for? Can you give an example of the DN and the corresponding sigma0 values? A numerical example of how the calculate sigma0 can be found in Section 4.1 of the ESA document “Radiometric Calibration of S-1 Level-1 Products Generated by the S-1 IPF” at

sigma nought is a relative measure of intensity (or power), so can never be negative. The use of log scaling (dB) is because intensity can vary several orders of magnitude, i.e. from 0.001 (-30 dB) to 10 (+10 dB).

1 Like

Thanks. Does calibration by the SNAP convert DN into sigma nought in dB or result of this process is dimensionless ?

Hello everybody,
I have a question about the time of acquisition of images from Sentinel 1 A.
The Sensing Date I see in the Scientific Data Hub (for example: Mission: Sentinel-1; Instrument: SAR-C; Sensing Date: 2016-05-05T05:11:28.052Z) is the local time of that area?



The acquisition time indicated in S1 data is UTC time. In your example 2016-05-05T05:11:28.052Z, the time is 05:11:28.052 where Z refers to Zulu time. A search in Google gives the following definition for Z:

Zulu (short for “Zulu time”) is used in the military and in navigation generally as a term for Universal Coordinated Time (UCT), sometimes called Universal Time Coordinated ( UTC ) or Coordinated Universal Time (but abbreviated UTC), and formerly called Greenwich Mean Time.



I also have a question to S1 GRD.

In the documentation of Level 1 data it’s written, that the ground range detected is projected onto the ellipsoid ( likely WGS84). So is there an ellipsoid correction from the S1TBX still needed? I work over ocean so the terrain correction is not important for me.
If the terrain/ellipsoid correction is not needed, how can I reproject the image. If I use the “reproject” function of the Toolbox, it says, that I have to do the terrain or ellipsoid correction. So what to do?

Many thanks!