Subswath mismatch after merging S1 mosaics

Dear SNAP community,

I am producing coherence images by processing several raw S1 SLC tiles (mosaics) using the TOPSAR Merge processor. However, I have noticed a consistent mismatch between IW1 and IW2 across my data of several of these images. Some exhibit strong mismatches (B), while others show just a small black hairline (A). The error also occurs with other swaths and only between IW1 and IW2, while IW2 and IW3 are consistently good merged.


Does anyone know how the error occurs and how it can be resolved?

I do the processing in two steps, as I had weird errors in the graphbuilder when I tried to squeeze everything into one single graph. SNAP was fine with the two-step-method, which is why I applied it. All coherence images use data with 12 day baseline. Amount of raw SLC scenes is the same for both dates of each coherence pair. Find the graphbuilds below.

Thanks for any help and ideas!