S1 TOPSAR Interferometry on S1TBX - Further improving interferogram coherence

[smithalas]

Hi all. I have processed an image pair using the S1TBX TOPSAR Interferometry tutorial and I am quite happy with the results (there seems to be a clear signal) however I was wondering if there are any additional steps not listed in the tutorial that could improve the interferogram further? For instance, would any of the following steps improve coherence? :

-Remove antenna pattern
-Thermal noise removal
-Speckle filtering
-Spectral (range and azimuth) filtering

Thanks in advance!

For some context here is the phase image I obtained after completing all steps listed in the TOPSAR interferometry tutorial:

[ABraun]

however I was wondering if there are any additional steps not listed in the tutorial that could improve the interferogram further?

you could, theoretically, use more image pairs of the same region, process like above and then create an average interferogram out of all. This clearly reduces the influence of atmosphere on interferometric phase.
This however requires that your surface didn’t change much between the image pairs’ acquisitions.

[smithalas]

Thank you for your reply! This will definitely be something to keep in mind.

[smithalas]

Sorry I don’t think I fully understand this. What would be the utility of the average interferogram? Would the interferometric image pairs used in the averaging have to cover more or less the same period?

[ABraun]

I asked myself the same question some days ago - but unfortunately didn’t come to a solution so far:

Interferogram averaging for DEM generation

I wondered if it's possible (and reasonable) to calculate multiple interferograms and average them for later phase to elevation calculation. I have Sentinel-1 A/B interferograms (6 days temporal baseline each) in an arid region with high coherence. I would like to exclude atmospheric effect by a combination of interferograms. But I struggle to do it in SNAP. When would be the best step to do the averaging? Directly on the filtered interferograms (I doubt that this makes sense) After unwrap…

[mauleon.fb]

Hi. Need some help. I think something is not right with my InSAR processing. I followed the steps in S! TOPSAR Interferometry tutorial as follows:

1. RAW Data downloaded from Sentinel
   

2. Coregister the two dataset using S1 TOPS Coregistration with TOPSAR Split at both IW2 and Backgeocoding using SRTM3Sec:
   

   
   

3. Interferogram formation:
   

   

My interferogram does not seem to show visible “fringes” so even though i proceed with topo phase removal and goldstein filtering it seems something is not right from the interferogram formation

Can someone tell me what is wrong? is it with the parameters i used? i just use the same in the tutorial

@ABraun

[mauleon.fb]

@dsmilo @johngan

Thanks appreciate all the help for a beginner like me in InSAR processing

[mauleon.fb]

I still proceeded with the topographic phase removal ->goldstein filtering -> multilooking -> phase unwrapping using SNAPHU then Terrain Correction and the result was:

Someone else did an InSAR process on the same data and he was able to produce this interferogram (upper right part of the previous image)

So I’m wondering what was wrong on my process.

@ABraun @dsmilo @johngan

[ABraun]

I wonder why you performed topographic phase removal. What is your overall aim?

You can increase the window size in the interferogram step to get better fringes.

How does your coherence look? No coherence, no fringes

[johngan]

HI,

As ABraun said mentioned, if your coherence is low, no fringes are formed, hence, you have just random phase. By looking at your SAR image, it looks like the coherence is low apart from a few areas in your image. The quality of the interferogram is heavily depends on the coherence.

I guess that what you are trying to do is to estimate movement/subsidence on the ground ? if yes, then areas with random phase (low coherence areas) cannot be used for estimating movement or subsidence. You may need to look at your coherence map.

In terms of the second image, it looks like the coherence is high, hence, the nice fringes.

[ABraun]

I think coregistration also plays a major role in the resulting quality of the interferogram. Besides that, Multi-looking, filtering or the quality of the DEM could have differed in the cited study which all increase or decrease quality of your fringes.

[dsmilo]

@mauleon.fb I can’t imagine the coherence being that bad with just 6 days between passes. Did you use the same subswaths in each scene? Do they sit on top of each other well, or is one subswath different from the other? Did you use S1 TOPS Coregistration? Did you check the modeled coherence after coregistration? (Radar > Interferometric > InSAR Stack Ov erview) Add your files and run the overview. The modeled coherence should be as close to 1 (or 100%) as possible.

Are you sure that the interferogram you are trying to replicate was processed using the same exact datasets? Was ESA SNAP used for this?

[mauleon.fb]

@ABraun I performed the topographic phase removal yes because i needed to create a subsidence map after an earthequake which is on July 6.

As for the coherence, this is how it shows on SNAP

@dsmilo I use the same subswath (which is IW2) and use S1 TOPS Coregistration and the modeled coherence showed as:

I did use the exact dataset mentioned in the paper and he SNAP was mentioned as his tools. Screenshot from his paper as:

That’s why I don’t know why he was able to create a very good interferogram and mine shows grainy results. I can’t figure out what i missed doing

[mauleon.fb]

If some can try here is the link of the downloaded Sentinel files.

https://drive.google.com/drive/folders/0B3if9bjQvNR8WThmNmFGQjdtY0E?usp=sharing

Hoping someone can show what process i miss or did wrong to come up with a good interferogram and subsidence map

Thank you all

@ABraun @dsmilo @johngan @smithalas @sharifulgeo

[ABraun]

@mfoumelis is one of the authors of this document and also member of this forum. Maybe he can explain how these nice fringes were achieved.

[mauleon.fb]

Yes i have see that @mfoumelis is one of the authors maybe you can help explain the process made that would really be a big help…

Thank you!

[mengdahl]

That scene has a lot of water in it, perhaps the ESD-correction did not work properly?

[ABraun]

this is what I achieved with this data and the given processing steps:

[dsmilo]

Yes, highly recommend reaching out to the authors directly to see where you are falling short. I am eager to see what was done to accomplish the results shown in the paper.

[dsmilo]

It does look like a subset was applied by looking at the report result boundaries, however.