Analysis of Image Pairs

Hi gentlemen, I am confused about analysis of Image pairs for time series displacement monitoring. I want to show how subsidence has propagated through the study period. I feel I am missing out something on my understanding. Would be thankful if someone could give me a right direction on my problem. I have 7 Alos-1 images of different times of years (2007 to 2011) randomly. The total pair combinations came out to be 21 with older image as Master. I have done the LOS processing for all the 21 combinations following the tutorials. I am getting results in some pairs and not in others. Now my problem is with the analysis.
Does it make sense in doing the study with 2007 image as Master and seeing its combination with other years. For e.g
12 January 2007-27 February 2007 ( this doesn’t show anything in interferogram)
12 January 2007-01 March 2008
12 January 2007-16 April 2008
12 January 2007-17 January 2009
12 January 2007-04 March 2009
12 January 2007-10 March 2011

OR like I got some results from these pairs.
12 January 2007-16 April 2008
27 February 2007-17 January 2009
01 March 2008-10 March 2011
16 April 2008-10 March 2011
04 March 2009-10 March 2011

Since I do not have the consecutive pairs like for e.g Jan 2007 to Jan 2008, Jan 2008 to Jan 2009, Jan 2009 to Jan 2010, Jan 2010 to Jan 2011. I fail to understand how I can justify the subsidence in my area.

  1. How can I justify the subsidence between Different months of different years. Should it be always consecutive to do time series analysis of subsidence.
  2. I did Terrain correction after unwrapping which shows LOS. Is it possible I can extract the LOS values from the image to a table so I can put in my report.
  3. How can I calculate vertical displacement from LOS. Where do I get the local incidence angle? When we do Terrain correction there was an option to select local incidence angle which got added in my displacement bands. Should I use this band for calculation of vertical displacement in Band-Maths.
  4. What kind of analysis can I do from LOS or vertical displacements to show time series results.

I hope I could explain my problem. I apologize for such a big post, but I need to derive some conclusion from the results.

Look forward to your advice soon.
Thanks.

the big problem with pairwise DInSAR is decorrelation. It all depends on wether you have clear fringes in your interferogram or not. If the interferogram produces random patterns, it does not make a difference if your use consecutive images or pairs with one fixed master. If the interferogram is empty, the coregistration failed, so you can try to repeat this with higher number of GCPs.

Can you maybe show an example of how the interferograms look like?

About the analysis: Please have a look at these resources (available after login): https://eo-college.org/resource/insar_deformation/

Although the slides are quite old and refer to earlier versions of SNAP/NEST where many things were more complicated, it gives an outlook at the end on how to combine the displacements of many images for a time-series analysis.

About the conversion of LOS to vertical displacement: There are many discussions on this in here, you will find them with the search function.

Thank you for your response Mr. ABraun. Highly appreciated. So that means if I show the results from march 2008-Mar 2011 or Jan 2009 to Feb 2011 it should make sense. Not necessary to be consecutive.
Below are the screenshots of what I got till now.

  1. Phase from Filtered Subset (Mar2008-Mar 2011)
    image

  2. Unwrapped Phase (Mar2008-Mar 2011)
    image
    (does this phase unwrap make any sense; Min=0)

  3. LOS on Google Earth (Mar 2008-Mar2011)
    image

  4. Phase from Filtered Subset (Mar 2009-Mar2011)
    image

  5. Unwrapped Phase (Mar 2009-Mar2011)
    image

I’m afraid these interferograms are not usable. There is high coherence over the city, but the noise outside introduces weird patterns which are not related to subsidence.

Besides, the pattern of the fringes indicate ionospheric effects superimposing the actual displacment (Interferogram from Alos palsar 1.1 data)

Did you perform topographic phase removal?

Yes I did topographic phase removal.If this is noise is there other way to get the interferograms. I dont see any prominent fringes in my images as I see in the research papers.

can you try one with the shortest temporal baseline possible?

For around one month baseline i get the following
image

And this is the coherence that i get
image

this doesn’t look so good…

Either the coregistration was not successful or there are ionospheric effects. These kinds of linear fringes are not caused by deformation and it cannot be extracted from theim either.

Did you apply ALOS Deskewing before coregistration?

No I did not do alos deskewing. I will try to see how to do it and then I will get back to you. thanks

Thanks Mr. Abraun for the advice.I did Alos deskewing before coregistering. But the interferogram still is the same.

image

Is there any way to remove ionospheric defects in Snap.

i subsetted my image to just focus on the city. I did as suggested by mdelgado in the post you referred to change the polynomial degree and gcp to higher side, I get the following interferogram now. I dont see the line patterns now. But still no visible fringes.

image

and the coherence

image

this looks better, although we cannot say yet if these patterns are caused by artefacts or real displacement.
Can you please open the InSAR stack tool to view the perpendicular baseline and the altitude ambiguity?

image

Is there something I can do about this. I am desperate to get some interferograms in my images.

you have no impact on the satellites’ positions, but you can test different image pairs.
Unfortunately, ALOS has no baseline information on available products (like the baseline tool of Sentinel-1, for example).

Thank you very much ABraun. I will try more pairs. I have already checked with 12-15 pairs. I dont see any fringes in any of those except the ones posted earlier which you said looked like ionospheric errors. What might be the problem. And there is defintely subsidence reported in my area.

the largest problem I see is temporal decorrelations. There might be fringes over the urban areas but total phase decorrelaton over its surroundings caused by vegetation.

This might be interesting to you: Echoes in Space - InSAR error sources