Radiometric & Geometric Correction Workflow

Hi everyone,
to find my corner reflectors by S1 SLC, I have tried to do this workflow:

  1. TOPSAR-Split (VV polarisation)
  2. Apply-Orbit-File
  3. Calibration (output beta0 band)
  4. TOPSAR-Deburst
  5. Terrain-Flattening (default parameters)
  6. Speckle-Filter (default parameters)
  7. Terrain Correction (default parameters)
    Then I covert the gamma0 linear to dB.

But I have some questions about this workflow.
I see some papers that use sigma0 band instead gamma0 band to find the corner reflectors, so to do this I need to select as output the sigma0 band in the step 3 and to delete the step 5 (because it converts sigma0 band to dummy); first quesiton: which is the difference between gamma0 and sigma0 and which is better for my goal?

Second question: @ABraun I’ve tried to use SAR simulation terrain correction instead terrain correction, but the graph builder show the following error message: “Source product should include a simulated intensity band.”. Do you know why?

Every suggestion is welcome, thanks in advance.

please have a look here: Difference Sigma, Gamma and Beta nought? - #2 by ABraun

are you sure you used this tool?


Because actually, simulation of an intensity band is part of the operator.

1 Like

Technically speckle-filtering should take place before TF so you should swap 5. and 6.

1 Like

How about the “Border Noise Removal” discussed in the article below?

I wouldn’t consider it mandatory, only if you have this kind of noise and want to get rid of it.

Hello.

I tired doing the SLC steps for generating gamma0 bands for VV and VH. However, the resulting bands are empty. Do you know any further step I need to do for this?

Thanks!

after which step is the product empty?

After range doppler terrain correction
here’s the screenshot:

Are the images OK of after the TF step? Empty images in TC sometimes result from missing DEM issues.

checked the results of TF and it was empty too… I found the problem now, wrong DEM.
Thanks a lot!!!

Hi, thanks for your explanations and links, they really helped me.
Could you please link the source when specifying changes like this:

" NOTE : Since SNAP 8, conversion to Beta0 prior to Terrain Flattening is no longer required, because it has been integrated into the TF operator."

Because I think this has not been implemented yet:


I’m using SNAP 8

Best,
Ognjen

It’s possible that the help menu hasn’t been updated yet(@jun_lu), but the issue is marked as resolved

Yes, but it says “Fix versions: 9.0.0.0”
image
Does it mean it won’t be included unti SNAP 9 or maybe I misunderstood?

According to my understanding, this is already included in the current version. At least I have tested with SNAP 8 and it worked.

For some reason the functionality is not in SNAP 8. We will issue a module update as soon as possible.

You might consider “forcing” beta0 calibration as preparation to terrain flattening, to avoid follow-down errors.

that is how it was solved until SNAP 7, you get an error message that Beta0 is required as an input. The idea for improvement is to check if the data is uncalibrated and include Beta0 computation within the Terrain Flattening step.

It is not fixed in Version 8. For both in graphical model builder and snappy, I get “dummy” after TF, when I input other bands than beta0.

Is this possible now. In the support document it says that the input can be a T3 C3 or C2 matrix but I am still getting the error.

Thanks

Hi,

I applied terrain flattening to the raw asset using the following pipeline xml(1) and then added the processed asset in google earth engine. However I can see the processed asset being shifted a little (2)(as shown below in earth engine code editor output screenshot). I have also attached the xml(3) without terrain flattening which works fine.

  1. RTC and TF xml (Misaligned) - Pastebin
  2. Terrain Flattening - Pastebin

Regards
Manoj