Interferogram averaging for DEM generation

Well, I don’t know if it’s the right place but in general, It would be a good SNAP QoL improvement to have a button “copy to clipboard” for any displayed data/tables


That’s an excellent idea!

1 Like

I realize I’m jumping in on a thread that might be stale, but what @ABraun describes regarding pairing slaves to form interferograms is exactly what I need at the moment. Is there a timeline for such a capability?

Thanks for any advice, in general, I’m grateful for what SNAP provides.

Hi ABraun,
I’m trying to average interferograms, creating according to “Create a DEM using Sentinel-1 Data” reciepe. So, as far as I understand, you’ve stacked unwrapped interferograms after snaphu import. Which tool have you used? Radar - Coregistration - Stack Tools - Create Stack doesn’t work because of “Phase band should not be selected for co-registration” error.

I’m not sure if averaging them before unwrappng might make more sense because the fringes are still present and could be pronounced by the averaging. But I don’t know if snaphu takes anything besides the exported interferogram.

I’m following these steps for a series of S1 SLC IW data (M-S1, M-S2 and so on)

And then I want to do:

Is it a right way to create interferograms stack or may be I missing something?

give it a try!

That’s what my question about. I can’t create stack from imported unwrapped interferograms. I use “Create stack” from Radar - Coregistration - Stack Tools, but receive a mesage “Phase band should not be selected for co-registration” error. What tool did you use for interferograms stacking?

just look above :slight_smile: Interferogram averaging for DEM generation

coregister all images to one master and then stack them with coregistration > create stack.

Ok, I’ve tested this approach, but have been dissapointed by results. So, I’ve come to conclusion, that averaging unwraped interferograms has as little sense, as averaging resulting DEM.

I’d like to ask you again after long time, did you get any result in this topic?

No. I found that areas with suitable coherence area already fine with one pair while areas of low coherence can only hardly be improved by including multiple interferograms. But I am still interested in this point. Any suggestions for a suitable study area? Should contain various surface types and be small enough to be included within one subswath for a first try.

I’m now trying out to implement an average of multi-interferograms, so called stacking , to get one result including the less atmospheric effects, in addition to the long - term ground deformation average. Do you think there is still no chance to do so?

Averaging the interferograms should be no problem in SNAP but I am not sure if the snaphu export module takes this average as an input for the subsequent unwrapping.

How to implement it?

I guess it will be one averaged infgrm, so what is the matter in this case?

Also I’d like to draw your attention to our colleague answer, might be you have comment to get me well cover this topic.

Calculate interferograms for images pairs of different dates (all at 6 day intervals at best) and stack/coregister them. You can then calculate the phase average in the band maths. However, I am not sure how the snaphu export will recognize this band maths result as the correct image to export and process within snaphu, because usually the input for the snaphu export has i, q, intensity, phase (and sometimes coherence). So we would need to find a work around to replace it by the averaged interferogram. One solution could be to perform the snaphu export on a “normal” interferogram and then substitute the exported img by the average raster. But the coordinates have to be the very same.

I have seen the answer of johngan, he basically suggests the same. But his post doesn’t include the unwrapping step (as far as I understood it).

It is possible to average the unwrapped phases, of No.s infgrm?

Could be an option as well, you are right.
According to my understanding, the unwrapping suffers from bad interferogram quality and so, unwrapping errors add up. So I would see that as an alternate option, but not the best choice. Averaging interferograms before the unwrapping should reduce errors way better.

Yes, you are totally right, but we have two issues now,

First one is, how it is possible to create stacking of intfgms to reduce the errors? Using SNAP
Second is , how it is possible to to get the average of deformation for NO.s infgms? Using SNAP

Could implement both in one, or separately. Now the issue in between the hands @marpet or @lveci or any one else of our colleagues could help in this issue.

As the interferogram product contains an intensity band, coregistration should work.

Deformation can be a problem, yes, but if the investigated period is short (3 months) it is probably negligible.