Pixel-based calculation of Sentinel-1 products

Microwave Toolbox
#1

Dear all,

I was wondering if it is technically possible to calculate Sentinel-1 products such as Sigma0 in a pixel-based manner, and if yes, if someone has experience doing so. Calculating. I can imaging it’s quite complicated, if not impossible given all the necessary steps to come to end products, but who knows…

#2

Would you pleas to take a look at this tutorial,

#3

Thanks, but I don’t really see how this addresses my question? That workflow still starts from an entire S1 image, although it clips afterwards an AOI. But what would be interested in is to be able to calculate OTF things like sigma0 on a scale as small as one pixel and in a very reasonable amount of time. This is something different than what your tutorial does.

#4

It is not my tutorial it is RUS, Copernicus. But what I get understood from your first post, that you’d need to calculate Sigma0 is that right?

What do you mean exactly of that?

#5

I understand it’ RUS, I’ve been using it :slight_smile:
No, my main concern is not the calculation of Sigma0 itself. My question really is, can we calculate radar products on the fly pixel-based in webservices instead of processing images in SNAP or is this physically impossible (e.g. because we need the entire image in some steps).

#6

You don’t need the entire image, you just need the image metadata and annotation files.
Calibration is actually quite simple. Just find the calibration factor nearest to your pixel in the LUTs annotated in the calibration files, and multiply by your pixel DN^2. As explained here:

Denoising is similar, but has become slightly more complex in the latest versions of the processor (IPF). See
https://sentinel.esa.int/documents/247904/2142675/Thermal-Denoising-of-Products-Generated-by-Sentinel-1-IPF

Accurate geolocation of individual pixels is also possible without having to Terrain Correct the entire image. Only image metadata and DEM needed. Equations are not the simplest and there may be a bit of overhead to geolocate just one pixel on the fly.

#7

Hi css,
thanks for the information, this is what I need, a confirmation that in theory this should be possible. But it might be beyond my (our) capabilities to implement this. We can look for someone who can.

Kristof