Getting poorly unwrapped, weird interferograms from SNAPHU Unwrapping procedure

[Laerke]

Hello STEP-Forum community!
I’m working on creating displacement maps from the Reykjanes Peninsula, Iceland, using Sentinel-1 SLC data.
I am able to create nice interferograms using the SAR images I have. To unwrap the interferograms, I use Snaphu Export, the Snaphu Unwrapping plugin and Snaphu Import. However, no matter what I do, the resulting “unwrapped interferogram” is very bad. I’ve tried performing my workflow for different timespans, and both 24 days and 12 day intervals. I’ve also tried using different coherence windows when making interferograms, and fourier transforms when doing the goldstein phase filtering. Could there be something wrong with the unwrapping algorithm?

My workflow is as follows:

Because the study area is not covered by one single SAR image, I need to download both a southern and northern SAR image and merge their subswaths/bursts together, which I do with splice assembly seemingly with no problems.

My interferograms usually look like this:

With a coherence like this:

BUT then the unwrapped phase looks nonsensical…

Would anyone know what kind of issue could be causing these unwrapped images to be so weird? I appreciate any help. Should anyone wish to try for themselves, an example of the data I use would be:

S1A_IW_SLC__1SDV_20231216T075839_20231216T075907_051677_063D82_6801

S1A_IW_SLC__1SDV_20231216T075905_20231216T075920_051677_063D82_B70D

S1A_IW_SLC__1SDV_20231228T075838_20231228T075906_051852_064392_01DA

S1A_IW_SLC__1SDV_20231228T075904_20231228T075919_051852_064392_2AF6

[Boorhin]

Maybe not enough overlap? (I am struggling myself and trying to find the best parameters)

[mikhelif]

I have a question about the coherence window size you used. Is it in meters or numbers of pixels ? Because 10-250 pixels is just too large, i usually use 3x10 or 4x12.

Also when comparing your interferogram and coherence map, they doesn’t match. For example you have a smooth interferogram in areas of low coherence.

[Boorhin]

I have been looking at this for a while now and I cannot find the reason why this looks like it.
The gradient is flat after a while, it also doesn’t match the phase observed.

Maybe unrelated but subset doesn’t go through snaphu so cannot really reduce the number of tiles

[Boorhin]

I have less weird if not satisfactory unwrapped result by deactivating the NaN in the sea water while co-registering. I read several times that SNAPHU didn’t like NaN.
Managed to do subset manually (in the GUI) but not in a GPT graph… So it is not that.

[FinPip]

Dear Boorhin,

I am having the same issue. Please look also this new topic.

Snaphu unwrapping not working

Dear all, the SNAPHU plugin does not work properly and gives a wrong unwrapped phase output. I am trying to create a DEM from Sentinel 1 Data, following the ESA tutorial at: https://step.esa.int/docs/tutorials/S1TBX%20DEM%20generation%20with%20Sentinel-1%20IW%20Tutorial.pdf. I am using the same 2 SAR images. I can run the Snaphu Export, Snaphu Unwrapping and Snaphu Import, however the final unwrapped output is absolutely wrong. Here the comparison between the interferogram (which worked well…

With deactivating the NaN you mean to toggle off “Mask areas with no Elevation” during the Backgeocoding?

[FinPip]

Hi Boorhin,

I have actually toggle off “Mask areas with no Elevation” during the Backgeocoding and apparently it worked. I obtained a reasonable (maybe even correct) DEM at the end.
I´ll check if it works also for the generation of LOS displacement maps