I have been trying to create a DSM of a coast and some islands, and have run into some issues regarding how the data looks in the unwrapped interferogram (which affects my height data). I have been using the following method to process my SAR data using esa's SNAP software:

1) Co-register 2 images
2) Interferogram formation 
3) Deburst + Merge 
4) Goldstein phase filtering 
5) Phase unwrapping 
    -SNAPHU export,
    -run SNAPHU on ubuntu,
    -SNAPHU import

The final steps after successfully importing from SNAPHU are:

6) Phase to Height
7) Range Doppler Terrain Correction

After step 5, when I observe the unwrapped interferogram (of an island and coast), bands appear (along the sea) that seem to skew the data and I don't know why (please see image below: coherence image on the left , phase image on bottom right and resulting unwrapped interferogram on the upper right).


The bands follow a gradient from left to right and can be observed after phase to height and terrain correction. The gradient of heights of course is not representative of the area at all.

Does anyone have an idea why this is happening?

  • Could be a result of the filtering. Coherence is low over the ocean and the filter may have interpolated or filtered more heavily. What does the interferogram look like pre-filter? – Benjamin Aug 9 '16 at 13:18

The coherence is likely low over open water, so the filter interpolated or filtered heavily based on available data at the top and bottom of your interferogram. The best solution would be to create a mask based on coherence, and apply the mask to the unwrapped interferogram. It's possible that the unwrapper can also take a mask to prevent errors from this region from being propagated.

| improve this answer | |
  • The coherence is likely low over open water. That is an understatement :D. I ran some difference calculations based on two different LIDAR sources recently, in combo with some edge detectors, and the areas around water had some very amusing artefacts. I hate to thing what the intererogram would look like. – John Powell May 25 '17 at 6:59

I have very little experience in using SAR data for DEM extraction, but from looking to the resulting unwrapped interferogram (upper right), the pixelated area seems to be the no data black area of the coherence image (water) where the software interpolated the elevation values between land areas. That can be normal due to the no-data nature of water areas.

As for the anomalous gradient, it could be related to atmospheric effects or something gone wrong in the orbit file and parameters.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.