Error in unwrapped process

[ABraun]

I would skip that part and directly export interfered_stack_ifg_flt_ML for unwrapping

[tangyin]

Thanks,I just do “snaphu export”,I only select"DEFO"and"SMOOTH" ,the unwrapped could progress,once select"TOPO",the unwrpped will failured.
Have you ever encountered the same problem?I don’t know how to deal with it.Pray for help

[ABraun]

I never heard of such a behaviour, sorry.

The only thing I can offer is the standard workflow which should work for most cases for DEM generation:

1. coregistration
2. interferogram formation
3. phase filtering (optional)
4. snaphu export
5. unwrapping in snaphu (TOPO)
6. snaphu import
7. phase to elevation
8. terrain correction

Please try these steps in exactly this order.

[tangyin]

I am very happy you have been helping me to answer my question, thank you.

[bbroschag]

OK,I select TOPO,but unwrapped phase can’t progess,the commend window show the upper question,if i select DEFO ,the unwrapped phase step could progess

What tile cost threshold did you use? You can try with a threshold around 10.000.

[tangyin]

I use 500 tile cost threshold before.I try with a threshold around 10,but it still failured.

[bbroschag]

You should try 10000, sorry for the misunderstanding.

[tangyin]

Oh,I also try 10000,but still can’t progress. This problem shouldn’t related to the software or data.

[ABraun]

if DEFO worked, did you import it in SNAP again and look if it is okay?

[tangyin]

Because TOPO isn’t work,so I export all through the “DEFO” generated DEM.then import it in snap,but the resulting DEM accuracy is not high.

In addition, the accuracy of the generated DEM can be expressed by other means of size, I can only view the two points on the map elevation value.Is it possible to display the maximum and minimum values of the elevation difference between the two images?

[ABraun]

yes in the band maths operator. Right-click on a band, select band math, and subtract the first from the second. This creates a third raster which displays their difference.

[tangyin]

OK,I open the band math,and edit expression,elevation_HH-elevation,Is this right?

[ABraun]

yes. This gives you a raster with positive values where your InSAR DEM is higher and negative values where it is lower than the 30m SRTM.

[tangyin]

Thanks,the new_band_1 is subtract result,but the resulting DEM accuracy is not very good, some point’s elevation difference can reach more than 100 meters. I am going to use MATLAB to make the resulting 3D visualization.

[ABraun]

These errors are not unlikely for repeat pass interferometry.

SRTM was generated with a Space Shuttle and single pass interferometry. The accuray of SRTM is much higher than elevations generated from a single image pair of S1, even if its spatial resolution is higher.

[Tomcater]

Hi ABraun,

Would you like to suggest me, please? I’ve done the next steps:

1. Coregistered the stack of 6 images;
2. Coregistered 2 stacks of 2 dates each (so they use 4 different dates from the big stack);
3. In order to use coregistered images, I changed images in small stacks by coregistered images from the big stack. This makes me facepalming, but I can’t find a better way to prepare different stacks which would be based on coregistered images.
4. Prepared ifgs according to all the steps you had given above.
   The question is. Both of the ifgs show exactly the same numbers for an Elevation_VH band. Would it be right? Would be 2 DEMs be exactly the same? All the rest layers (Coherence, Topo_phase, Phase_ifg) are different, even displacement (Phase-to-Displacement) show different results. I would be better if I could use Phase-to-Height, but it doesn’t work.

Thanks!

[ABraun]

2. Coregistered 2 stacks of 2 dates each (so they use 4 different dates from the big stack);

why do you do this? It would make more sense if you calculated elevation for both dates individually and then average both results.
What is your intention by the way? Elevation or displacement? And what data do you use? I can’t follow…

[Tomcater]

Hi ABraun,
Thank you for an answer. I performed the second step because I must use images which were coregistered once to one master. Because of that I’m taking images from the super-stack and preparing ifg-pairs based on such images. I must prepare all the possible ifg-pairs and I need to find parameters for each pair. Finally, I need to find heights, I would do it as you suggested:

NaN pixels are produced after doing DEM generation

first you retrieve both elevation (from SRTM) and unwrapped phase at random points to calculate a correlation. then you use the correlation as a regression model to estimate the height (y) at every pixel based on the unwrapped phase (x) . [image] Image source That is not the most elegant method but it is a workaround I was successful with. Theoretically, you could also calculate absolute heights from unwrapped phase knowing the ambiguity height, wavelength and incidence angle. [imag…

Actually, that still the same task that we were discussing some time ago.
Averaging? Does it mean Stack averaging?
AFAIK Elevation means that SNAP creates DEM and Height means DSM. I’m looking for DSMs, but my equal elevations in the different pairs confuse me.
All in all, I use 6 images which were went through the next steps:

1. Applying orbit file
2. Split
3. Deburst
   4)Subset
4. Coregister to one master
5. Create stacks based on the images from the previous steps
6. Ifg creation
7. Topo phase removal
8. Goldstein
9. Export (Topo, MST)
10. Import
11. Range-doppler.
    Step 6 is supposed to be a cure for the SNAP’s inability to create ifg with a user-defined master from the stack.
    Thank you!

[ABraun]

I think snap only takes one interferogram in your stack. So you should probably process each image pair seperately and then in the end stack all unwrapped interferograms and average them. But I haven’t tested it myself. I would be interested in your results however.

[Tomcater]

Yes, that’s what I do. After creating all the pairs, I process them separately. Will try averaging after all.