Sentinel 1 Sub-swath differences

My dear SNAP users,
I have the following situation: I am aiming at thresholding water bodies from a Sentinel 1B image cross-polarized (VH), but I am observing some “artificial” difference in signal along a line within the Sentinel 1B GRDH tile, which can seriously bias the thresholding approach.
I decided to break the tile up in sub-swaths and check the problem. As a result, I used the S1B Single Look Complex IW1, IW2 and IW3 and applied the following workflow:
1)I Apply Orbit File
2)S1 TOPS Deburst (merge bursts and sub-swaths)
3) Multilooking 8 by 2
4) Speckle Filter Lee 3 x 3

My observation is that this low signal stripe occurs exactly at the sub-swath stitching point (1st sub-swath on the right). Could you suggest a method to tackle this issue? I am very grateful for any hint.
Useful info which might be relevant to the scattering in the image: location is Namibia, vegetation is predominantly savanna (grasslands).
This is SLC:

This is GRD: