Avoiding unwrapping over areas of low coherence

Hi all,

I have been successfully producing Sentinel interferograms using the toolbox but am looking for some advice on unwrapping. There have been some similar, but inconclusive, threads on this topic.

I want to avoid unwrapping over areas of low coherence as this is giving me lots of unwrapping errors. I know that SNAPHU uses the correlation during unwrapping, but this isn’t doing a lot to stop it from unwrapping more or less everywhere. Is there anything that can be implemented prior to exporting to SNAPHU to remove these low coherence areas?

During interferogram formation I estimate coherence using window size [40,11] and use the Goldstein filtering step twice. When analysing the interferograms in Matlab I have then used the coherence to mask out some of problematic areas. However this approach isn’t very robust and I still have lots of unwrapping errors. I have tried playing around with bandmath to set pixels where the coherence is < 0.2 to NaN, but haven’t got this to work.

Are others at this stage, and if so do you have any tips?