Radiometric Calibration - what can I use to implement the formula?

s1tbx
#1

Hi there,

I’m a Masters Student a bit new to this, and I’m using the S1tbx to calibrate an image I have. It is a GRD IW Sentinel-1A product. The purpose is to prove that by doing this manually I can get the same result as the inbuilt automatic calibration tool as part of the project. I know that in ENVI the Band Maths option can be used to calibrate an image using the formula provided in documentation. I was wondering whether this was also an option on SNAP, or if not what to use?

I have used the formula in https://sentinel.esa.int/documents/247904/685163/S1-Radiometric-Calibration-V1.0.pdf for a distributed target and obtained a sigma0 value of around 571. When trying to use this figure in the Band Maths window expression box I have been using expressions like “Amplitude_VV*571.4289793” and “Amplitude_VV^2/(6.6117035e02)^2”. I am not sure of the correct usage of the number now that I have worked it out, and none of the outputted bands look any different to the original. When I look at the pixel info for corresponding pixels across the images, on the outputted bands it lists under the “Bands” tab the corresponding values between images, and although the values displayed are orders of magnitude different, they don’t physically look different on the image produced. I was wondering why this might be? They are also a magnitude of 10 out, compared with the automated calibration image.

Firstly, am I going about this the right way using Band Maths? Secondly, is the use of the sigma0 number and the band name in either of the expressions “Amplitude_VV*571.4289793” and “Amplitude_VV^2/(6.6117035e02)^2” correct? Lastly, is the option of “Units” important in the Band Maths window, as I don’t know what to put?

Many thanks for any help,
R

#2

Hi,

Sigma0, in dB, has a range from approximately +1dB to -25dB (or 1.26 to 0.003 in linear units). So I’m not sure how you obtain a values of 571. Perhaps you could explain how you obtain this value. Section 4.1 of the document in your link gives an example of how to calculate sigma0.

Peter

#3

Hi,

I obtained that value by using the formula in Chapter 4, I inputted an averaged value for the amplitude and the same process for the LUT value, this gave 571. I’m not working with data in dB units for this, though I know I can convert to dB if I needed to.

However, when looking again this morning, I think this is essentially pointless to work out, as the formula that gives the best results places the Amplitude_VV band as the averaged amplitude (), so the latter of the two variations I asked about, “Amplitude_VV^2/(6.6117035e02)^2”. Therefore I’m not necessarily sure the 571 or whatever I worked out is any use at all.

Thanks for your help.

#4

Hi,

You will see for the equations in Chapter 4 use the average intensity and not the average amplitude. For SAR data the square of the average amplitude is not equal to the average intensity since the distribution of a uniform region is a gamma distribution. So for your distributed target (e.g. a field, forest etc), you need to square the amplitude value of all the pixels in the target and then take the average. If you do not follow this step, then the resultant sigma0 will be wrong. If you use the average intensity, you should obtain in more realistic value for sigma0.

Peter

#5

like peter said, the problem is you use amplitude rather than intensity