Strange Snaphu Unwrapping Result

Hello, I am working on DInSAR application by using Terrasar-X SSC SM data to monitor the deformation.
However, I cannot get the successful unwrapping result after Snaphu-unwrapping. Here is my steps:

  1. Coregistration
  2. Interferogram formation using SRTM 1 sec (Auto Download) to subtract topographic phase
  3. Apply Goldstein filtering
  4. Filter Out Low Coherence Pixels Before Phase Unwrapping

Coherence Map

  1. Snaphu-unwrapping

  2. Unwrapped phase

Sometime the result shows like this pattern
(Phase before (left) and after (right) unwrapping)

and sometime shows like this

the result is so weird. Any suggestion on the workflow and what is going on about the result of shaphu unwrapping. Thank you so much!

I think the unwrapping of traditional DInSAR approaches is not applicable here because you only have thin linear elements with high phase stability which are not really connected (which is an assumption of the unwrapping with snaphu). You could try to mask out all other pixels with low coherence to have the unwrapping only performed on the stable surfaces, this could reduced the error imputed from outside the structure.