Increasing value of displacement with increase height

[jossisko]

hi
Please bear with me, as I am a new user with SAR data, and in some my post the answers haven’t been really clear to me.
I doing a project determing vertical displacements using 2 S1A images using instruction “Sentinel-1 TOPS Interferometry”-Michael Foumelis.

I found a problem with values of displacement because when I export result in .kml file and open it in Google Earth application I see that values of displacement have correlation with values of terrain height. I don’t understand how it can be correlated because in processing I do “Topographic fase removal” step which should remove topographic influence on values of displacement.
Can anybody please help me to solve this problem? Thanks

[ABraun]

I see that values of displacement have correlation with values of terrain height.

can you describe this a bit more? Maybe a screenshot would help.

[jossisko]

thi is screenshot from Google Earth application.
I’m doing processing for area of Zagreb city (Croatia). On the image the orange and yellow part represents centre of city and it’ neighborhood and for that area values of displacement are good. Green and blue part of picture represents area of mountain Medvednica (max height 1033 m) and you can se that from going to north values of displacement get bigger and bigger??

[ABraun]

this kind of ‘ramping’ is often observed in Sentinel-1 repeat pass interferometry. It is probably caused by insufficient topographic phase removal but also from atmospheric effects.

Was your coherence good in your image? You could mask out areas with coherence lower than 0.3 for example. The resulting pixels could contain higher information content.

How large is the temporal gap between both images?

[jossisko]

Temporal gap between images is 16 months (First image:20151022, Second image 20170213).
Can you please describe me how can I mas out areas with coherence lower than 0.3?

this is statistic from coherence

[ABraun]

How to mask out pixels is described here:

Subsidence map in 3d view

I explained it in the post above. Just entering the coherence is not a valid logical expression. In this field (valid pixel expression) you define at which locations your result is usable. As coherence is a measure for quality of the interferometric product you only want those pixels at a location where the coherence is larger than a threshold, let's say 0.4. Thus, just putting the coherence there is pointless because the expression is lacking a condition. Enter this instead: coh_23Nov2012_08F…

Of course you need the long temporal baseline in order to observe changes. But the coherence won’t be high over the full image.
Regarding the atmospheric content: Maybe you have the chance to investigate another image pair within a similar period. If it contains the same ramping as your example above, it is probably caused by the topographic phase removal (did you use the 1Arc Sec SRTM instead of the 3Sec?). If it is different atmospheric effects dominate in the distortion of your displacement values and you could have more success with another image pair.

[jossisko]

huge thanks for fast reply!!!

I’ll try with other pair of images

I don’t have option to use 1Arc Sec SRTM Here is all what I can use

[ABraun]

I just re-arranged the term

SRTM 1Sec HGT (Auto Download)

From your screenshot I saw that you performed Split and Orbit File as separate steps. There is also a module which coregisters your S1 scenes at once.

Radar > Coregistration > S1 TOPS Coregistration > S1 TOPS Coregistration

[jossisko]

I know for that module but I thought it is the same result doing it together or as separated steps, am I wrong?

[ABraun]

yes, it’s the same. Just a suggstion.

So did the masking work out?

[jossisko]

I’m stil trying.
If I understood, I need to mask out unwrapped phase which I get after interferogram formation?
can you please suggest me what value of coherence I should put in coh_VV_22Oct2015_13Feb2017 > ??

[ABraun]

yes, after unwrapping it would be best. I’d say 0.3 or 0.4.
You can also make those areas zero

Coherence Mask/Threshold for Unwrapping

I would use the 'valid pixel expression' definiton in the band properties. For example: coh_HH_11Feb2008_04Mar2008 > 0.4 If you want to calculate a new raster you can also do it in the band maths: if coh_HH_11Feb2008_04Mar2008 < 0.4 then 0 else Phase_xx

To include the coherence layer in the final product you need to select it in the Range Doppler Terrain Correction module.

[jossisko]

Sorry but I don’t understand
I get unwrapped phase after using SNAPHU module in virtual box what is about on the half way of processing, and Interferogram formation is third step in processing immediately after Radar > Coregistration > S1 TOPS Coregistration > S1 TOPS Coregistration. So do I need mask out pixels after on the begining after I ger Interferogram formation, or I need to mask out pixels after unwrapped phase??

[ABraun]

after unwrapping or at the very end. It is just to exclude the pixels in the final product with low coherence. So you could also do it after terrain correction.

[jossisko]

Ok, thank you for explanations
I will post results as soon as I get them.

[jossisko]

I couldn’t get results on this way. After unwrapping to the end of proces I don’t have band coh so I can’t use coh_VV_22Oct2015_13Feb2017 >0.3

[ABraun]

actually, coherence is exported by SNAPHU export and imported with the unwrapped product again.

Folder:

after import in SNAP:

[jossisko]

Oh I see. Thank you

But how you put all these files i,q,Intensity, Phase,Coh and Unw in one? I can only import one by one and than they are separated

[ABraun]

you need to import it via Radar > Interferometric > Unwrapping > SNAPHU import

Please see the manual at step 10, page 34
http://sentinel1.s3.amazonaws.com/docs/S1TBX%20Stripmap%20Interferometry%20with%20Sentinel-1%20Tutorial.pdf

[jossisko]

I already have this instructions and I did import by these steps, but my problem is that I cant’t see on page 36 what is the name of 1-Read-phase folder???