Problem in S-1 mosaicing

Dear all:
I have sentinel-1 GRD products .I used the next images:
I made the processing with the next steps:

  1. import the images
  2. apply orbit file
  3. thermal noise removal
  4. calibrate
  5. speckle filter
  6. terrain correction
  7. SAR-Mosaic

I have two questions here:
1.The terrain correction result exists a gap between the two images Before mosaicing in figue 1.
2.After mosaicing, the gap is gone but the values of pixels far from the gap diffe from original values in figure 2.What’s the algorithm about mosaicing ?

Thank you for your help!

1 Like

The filenames reveal that these are consecutive slices of the same S-1 acquisition segment. You should use Slice Assembly to concatenate the products.

In your case the solution as @mengdahl stated is to use the Slice Assembly operator.
However, we are looking into why the terrain correction introduces this gap. It seems to be caused by not having enough pixels at the edge for the resampling and reverting to nearest neighbour.
Generally when mosaicking you don’t have this exact cut between the products. The mosaicking resamples based on the products geocoding. Where there is overlap, a weighted averaging is used to blend the images together.

1 Like

Thank you very much!Now that These are consecution slices of the same S-1 acquisition segment , why does the gap exist?
And I appended the border noise removal between thermal noise removal and calibratoverlap in figure1
if I use Slice Assembly, the Mosaicing wouldn’t be needed anymore.
Thanks a lot!

I truly appreciate your reply!
And another question is that it means Slice Assembly is better than Mosaicing in this case,right?What is the recommended workflow for Slice Assembly in SNAP with Sentinel 1 GRD data?
Thanks a lot !

You could apply slice assembly anywhere you haven’t yet messed with the geometry. So before any multilooking or terrain correction. In you scenario you could put it first or just before terrain correction.

1 Like

Hi, levic
Thank you very much for the help.Not like the gap in figue 1, I also tried to append the border noise removal between thermal noise removal and calibration resulting in overlap in figure3 and some edge dislocation about 2-3pixels exist. We can see some shift distinctly on the road in figure 4 which I zoom in .What am I doing wrong and What 's the function of Border Noise Removal?

figure 3
figure 4
Best regards:slight_smile:

As explained above, if you are mosaicking instead of using Slice Assembly, you are doing it wrong.

What if we want to first preprocess (apply “Terrain-Correction” etc.) by tile then do “SAR-mosaic”? Shouldn’t there some overlap between slices for later mosaic?

Was that mean for the Sentinel-1 data from the same orbit but different slices, we should always do the "Slice-Assembling first?