StaMPS - SNAP - DInSAR - Questions

Hello everyone, My name is Daniel, this is my first time in this forum, I’m studying Remote Sensing Master in Universidad Católica de Manizales (Colombia).

I want to make a project about PSI and SBAS techniques for DInSAR analysis in Sentinel-1 SAR images and I have read that I need a tool named StaMPS to generate those techniques in SNAP software, for that reason I have these questions:

1- Is StaMPS tool only used on Linux Ubuntu?

2- Do I need matlab to use StaMPS?

3- If I need matlab, what is the version that I have to use?

4- Does StaMPS apply for PSI technique only or can I use it for SBAS technique too?

5- Do you have a guide that I can read to learn how to use StaMPS?

6- What is the difference between StaMPS and viStaMPS, or are the same tool?

7- Does StaMPS apply only for Sentinel-1 SAR images?

8- What other tools do you recommend me to apply PSI and SBAS techniques?

I would like that you can helpme, thank you.

1- Is StaMPS tool only used on Linux Ubuntu?
there are cases where it worked on Windows (here), but Linux is recommended

2- Do I need matlab to use StaMPS?
yes, and also triangle and snaphu (links are given in the in the stamps manual)

3- If I need matlab, what is the version that I have to use?
Doesn’t really matter, but the signal processing toolbox must be included in the installation.

4- Does StaMPS apply for PSI technique only or can I use it for SBAS technique too?
StaMPS allows SBAS as well, but currently you can only prepare single-master approaches with SNAP.

5- Do you have a guide that I can read to learn how to use StaMPS?
Please check here: StaMPS - Detailled instructions

6- What is the difference between StaMPS and viStaMPS, or are the same tool?
viStaMPS is something entirely different and was developed for the visualisation of results generated with StaMPS. But StaMPS also allows visualization, so you don’t ultimately need it.

7- Does StaMPS apply only for Sentinel-1 SAR images?
No, any prepared SAR time-series can be used.

8- What other tools do you recommend me to apply PSI and SBAS techniques?
Please check the list of resources above. Once you have Matlab and StaMPS installed correctly, you are ready to go.

1 Like

Thank you for your advice mr.

In your experience,
what is your recommendation, work only with Linux Ubuntu or with Windows Subsystem for Linux (WSL) to apply StaMPS?

If you have access to Linux and know the basic things (command line, installing packages, compiling make files), I highly recommend Linux.

Have a look at these wolderful guidelines - the best around in my opinion: SNAP - StaMPS Workflow Documentation

Thank you for the information.

Mr ABraun.

Do SNAP program and StaMPS work for L-Bands sensors?

yes, it should work with all wavelengths, as long as you have enough images and all products are from the same track.

Thank you Mr ABraun

Mr Abraun, good day.

I have 2 questions about PSI.

When I use PSI technique, can I use it with just 4 SAR images, or do I need more to have succesfull results?

If I have an ascending and a descending image orbit, Can I use both to PSI analysis, or is neccesary that both images have the same pass or orbit?

unfortunately, both is noth possible. PSI relies on the identification of scatterers with stable information. This cannot be guaranteed with only 4 images, you need around 20 images to get reliable and noise-free time-series. Same applies for ascending and descending. The scattering mechanism is not the same if a signal is retrieved from different look directions so this is neither comparable nor technically feasible (coregistration).

Thank you Mr. Good day.

Mr Abraun.

You said that I need at least 20 images to have good results with PSI, but my doubt is that if those 20 images need to be used in one set, or can I use more sets to process them?
For example, I have 20 images, but can I use them in 5 sets of 4 images, that could be possible? or is necessary that all 20 images must be in the same images-set to be processed?

Thank you.

The all have to be in the same stack, otherwise the selection of scatterers will not work

Thank you Mr.

Good morning Mr Abraun, how are you?

I have few questions about DInSAR:
I am processing 23 SAR images in SNAP, first I did the split to de image and then I applied the orbit file, I know that the next step is the corregistration but I want to know something:

1- Can I cut out the images in the interest zone before I apply the corregistration (Back Geocoding)? I asked because for me that would speed up the work.

2- After of corregistration do I have to apply TOPS-Deburst? The thing is that I don’t understand what deburst is.

3- What is the perpendicular baseline? Is the distance between radar satelites for the same area but in different times? I am confuse with that.

Thank you Mr.

  1. if you are using Sentinel-1 data, you can only reduce the size by TOPS Split before coregistration
  2. deburst is required to remove the stripes between the different bursts, it merges them seamlessly (a more technical explanation is given in the SNAP help file)
  3. the perpendicular baseline is the distance between the position of the satellite at time a and time b. As the image are taken from the same track, this difference is mostly between 15 and 150 meters. For DInSAR, sall perpendicular baselines are preferable, but PS approaches are robust to variations.

Thank you Mr.

Mr Abraun, good day.
I have a question, PSI technique could be used in mountainous areas and rural areas?

Mr ABraun, another question.

I know that I need at least 20 SAR images to apply PSI, but how many interferograms do I need to use PSI?
As well for SBAS, how many do I need? In my case for my work I have 23 SAR images.

The situation is that I have a big confussion, some colleagues have told me that for a time-series of 23 SAR images I have to create many and many interferograms but I thought that I just needed to create one interferogram using the 23 images. I would like that you can help me with that confussion.

I want to share you an image of what I am trying to show you, I have an stack of 23 SAR images as you can see.

When I apply back geocoding I am going to make a corregistration, I have chosen a master scene, that is ok; then I will apply Enhanced Spectral Diversity, and finally the generation of the interferogram, the question is , Will I get one interferogram or 22 interferograms?

Or do I have to generate the interferograms image per image, image 1 with 2, 1 with 3, 1 with 4 …2 with 3, 2 with 4…separately?

As you can see my studio zone has mountains and is rural, so I will repeat this question, Can I use PSI and SBAS here?

Sorry to make many questions but I really need to understand this, thank you for your patience.

I’ll try to clarify some points.

The applicability of PS or DInSAR in general is mostly limited by vegetation. If there are not enough surfaces which are characterized by coherent scattering (bare soil, urban areas, low vegetation) results will probably not be reliable.
I cannot telll for your study area, but it might be worth a test.

PSI and SBAS are based on the same technique, but different principles.
PSI uses one reference image (mostly in the middle of the time series) and computes one interferogram to each of the other images (figure on the left), you get n-1 interferograms. In your case 23 images will result in 22 interferograms. Each of them calculates the phase difference and these are then corrected for systematic errors (topographic, orbital and atmospherical) and summarized to an average displacement over the entire period and often also a time series of displacement for specific locations. You can do this with StaMPS, for example. The interferograms would be AB, AC, AD, AE,…

SBAS allows to calculate interferograms between all image pairs, based on thresholds of perpendicular and temporal baseline (e.g. separated by maximum 48 days and 160 meters). This is displayed in the image in the middle. This allows you to analyse more pairs, so 23 images can easily lead to 50 to 80 interferograms. This allows the retrieval of more stable information on deformation with less noise, but of course requires more processing time. The interferograms could be AB, BC, AC, BD, CE…
The third example is based on even more image links (explanation in source). Currently, their preparation cannot be done with SNAP.


image source: https://www.mdpi.com/2072-4292/12/9/1364/htm

If you want to understand the principles and the technical processing, please go throught he list provided in this topic: StaMPS - Detailled instructions
especially the first two points (by @Gijs and @m477h145) give excellent instructions on how to pre-process and analyze the data based on PSI.