Interferogram from Alos palsar 1.1 data


#1

Hello
I am a novice trying ALOS PALSAR interferometry. at present I have two ALOS PALSAR 1.1 imagery downloaded from the Vertex website. I would like to create an interferogram using S1tbx. I tried the orthorectification of the imagery but following the deskewing method as given in the tutorials of ESA I have not been able to proceed with the interferometry process. Please help me!!


#2

I would suggest the following order:

  1. Coregistration
  2. Subset (if necessary)
  3. Deskewing
  4. Interferogram formation
  5. Terrain Correction

Did you receive any error message?


#3

I tried this. First I downloaded two imageries from the vertex website for the years 2007 and 2009, added the VOL-ALPSRP091260440-H1.1A file. For each of the files radiometric calibration was carried out separately. Following this, multilooking and speckle filtering was carried out for each of the images using the HH only. Then I created a co-registered stack for both the two imageries. Till this point there was no error. So I carried on with the ALOS deskewing part. The Deskew was done in 60 seconds! But as soon as I tried the Interferometry procedure Errors began to crop up. The most recent error is that the product must be an SLC product. After this point i have not been able to create any interferogram.


#4

If you are working on InSAR, you shouldn’t calibrate, multi-look or speckle filter your imagery at all. You lose all complex information present in your data.


#5

Okay. So I should directly go for coregistration of the two imageries, and then subset, deskew and go for interferogram creation? But what about the steps of interferogram creation? Any changes to be made for ALOS PalSAR imagery?


#6

yes, try the order I suggested above.
As long as the coregistration was sucessful, the interferogram generation will have no trouble.


#7

Thank you, ABraun for your suggestions. At last I have been able to complete the interferogram! But following this, if I have to come to the PSI/SBAS technique what all steps should I take? My main purpose is to predict the land subsidence rate taking place within the chosen area.
I remain indebted for your sincere help.


#8

Have a look at this overview: About the STaMPS category

Please note that you need at least 20 image pairs for PSI.


#9

Thank you, @ABraun for your response. Since Alos Palsar is available from 2007 till 2011, I might get around 6 or 7 pairs at the most. So if I have to fulfill my purpose what should I do other than the PSI/SBAS technique? Further after the interferometry should I go on with Coherence estimation, Topographic Phase Removal and Goldstein Phase Filter and unwrap the image? If not, how do I generate the displacement from the present pair of Alos PalSAR imagery?

Could you please help me out?


#10

you can simply analyze all image pairs you find and make a time-series out of them as demonstrated here: https://eo-college.org/resources/insar_deformation/
Note that the tutorial refers to an earlier version of SNAP and a couple of steps (unwrapping is easier, also the import of the unwrapped phase) have been improved since then.


#11

@ABraun can I work with ALOS PALSAR in this older version of SNAP? If ALOS images can be read and processed in it, should I follow the steps as mentioned in the tutorials? For example this one: https://earth.esa.int/documents/507513/1077921/2011_NEST_Training_Practicals.pdf/60780248-2ee8-4867-83b9-a66a39bba957

With your help, since I have been able to generate the interferogram, I wanted to perform the unwrapping and export in SNAP itself. But is the current version of SNAP not suitable for unwrapping or for the time series estimation?


#12

I really advise to use the latest version of SNAP. It is the most complete and has the least errors.
Unwrapping is possible with snaphu as described in the tutorials, but neither NEST nor SNAP can do it themselves. So you will need snaphu anyways which means that you can really use the latest version of SNAP.


#13

Thank you, @ABraun. If I have understood correctly, I need to use NEST and not SNAP in case I need a time series analysis. But if I just want to unwrap phase for the current pair of images now after terrain correction, what steps should I follow? Should I follow this sequence in SNAP:
1.Topographic Phase Removal,
2.Goldstein Phase filtering and finally
3.unwrapping in Snaphu?

Or should I start from scratch in NEST and follow the tutorial link which I sent you?


#14

no, you should use SNAP.
The tutorial is just from a time when SNAP was still named NEST. I pointed this out because in the meanwhile some things changed a bit (to the better) and you shouldn’t stick too much to the explanations given in the tutorial.

Regarding your question: Don’t perform terrain correction before you are finished with the interferometry. Make it the very last step.

  1. make an interferogram
  2. remove topographic phase
  3. filter it, if required
  4. unwrap it in snaphu
  5. import the unwrapped phase back into SNAP
  6. phase to displacement
  7. terrain correction

#15

Thank you so much, @ABraun. I am trying to follow your steps. Once I complete it I shall surely inform you. I am not sure how to do a time series analysis in SNAP. So I need a little more help with it. Suppose if I want to extract the displacement from the pixels to develop a graph in excel or otherwise, will it be possible?


#16

yes. Just have a look at the NEST tutorial I posted above. It shows, step by step, what is done and why.

In principle, you make interferograms of all your image pairs: AB, BC, CD, DE, EF, FG…
You unwrap them, convert to displacement, and terrain correct them. Then you can see the displacement for various time increments.


#17

Thank you, @ABraun for the help. I am still processing the imagery. I shall follow your instructions and I hope I can succeed. I am facing some problems with unwrapping the phase as it is taking a long time for SNAPHU to unwrap the phase. I hope that is normal.

Well, if things work out or do not I shall surely inform you.


#18

unwrapping always takes the longest. How many lines does your data have? It is the number after the snaphu command.


#19

As far as I understand 27 lines. Right now it is at the 9th line. :smiley:


#20

that is fine then.