How to prepare Sentinel-1 images stack for PSI/SBAS in SNAP


yes mdelgado.
Gratitude to all those persons on step forum like u who’s comments in time helped me a lot.
I too have completed TOPSAR subsidence detection using DInSAR. Read and watched that tutorial last year.

Now Stamps PSI time series is my aim.
Hopefully will complete it in a week or two.


Hi everybody,

I am facing the same problem - I also get the same warning in StaMPS step 3:
Warning: Not enough random phase pixels to set gamma threshold - using default threshold of 0.3
> In ps_select (line 184)
In stamps (line 356)
resulting in:
PS_SELECT: Initial gamma threshold: 0.300 at D_A=0.07 to 0.300 at D_A=0.41
PS_SELECT: 0 PS selected initially
PS_SELECT: Reestimation gamma threshold: 0.300 at D_A=0.07 to 0.300 at D_A=0.41
PS_SELECT: 0 ps selected after re-estimation of coherence
No PS points left. Updating the stamps log for this*
PS_SELECT: Finished
STAMPS: Finished

I am a bit suspicious due to the output of step 2:
PS_EST_GAMMA_QUICK: gamma_change_change=0.000000

Is there suggested solution to this?

I have already looked into the MAINSAR user group as well, where a similar thread didn’t provide a definite answer.

There isn’t any water in the study area, and lowering the value for filter_grid_size parameterdidn’t result in any change either.

Data preparation for the study area has been done through SNAP and running the modified mt_prep_gamma_snap.

Any assistance would be appreciated.


How long is your time-series and how many images did you use?
How large is your area and what landcover is present here?

I noticed the number of identified PS rapidly going down by extending my temporal window from 1 to two years in a study area with marshlands.


Thanks for the quick response.

  • The time series used covers the whole year 2017 (2017-01-14 to 2017-12-28) and includes 28 images.
  • The area covers about 40 km², is located in the Northern Limestone Alps and is characterized mainly by mixed coniferous forests (50%), grassland and agricultural areas (40%) and some artificial surfaces (settlements and a major highway, 10%).


should at least produce a few PF…
You can try to increase the amplitude dispersion threshold when calling mt_prep_gamma
from 0.4 to 0.42 (maximum according to the manual)


Its gives error if we generate ifg from subset_split_orb_stack_deb
when we generate ifg from split_orb_stack_deb and then subset it, it works fine…


Indeed, that’s what I was expecting, too. The study area does not seem to be extraordinary in any respect.

I have already been using an amplitude dispersion threshold of 0.42, which results in 100k PS candidates.

Purely for testing purposes, I just tried to further release the threshold restraint. Even with absurdly high thresholds (and more than 1 million PS candidates) no PS points remain in step 3, which kinda puzzles me.
In step 2, the result is still PS_EST_GAMMA_QUICK: gamma_change_change=0.000000.

I will have a look at the snap2stamps preprocessing again.


@virk_rana I always used debursting after interferogram generation and would advise to do so.


did you increase the percent_rand variable for a test?
setparm('percent_rand', 60)


should i reexport it after ifg_deb?
if it is going to create problem in future products?


@ABraun yes, but no effect either.


if you change anything in your pre-processing, you need to apply the StaMPS export again, of course.
It takes exactly two inputs:

  1. one stack of coregistered images
  2. one stack with interferograms (including dem and lat/lon bands from topographic phase removal)


Hi ABraun,
Do we need to perform TC before StaMPS exoprt as given in katherines’s doc.


as the lat/lon bands are included since the removal of topographic phase, this is not required any longer.


Hi all

I have a question, somebody know why in the processing after StaMPS, I obtain mean amplitude=0 for several images??? I can recuperate this value or only I have that eliminate this images?
thanks for advance


Dear @gabrielaquintana77,

those values will lead to some issues I think already in step 1 or 2 in StamPS after you have entered Matlab…Whenever I have an image with 0 mean amplitude value, I was not able to process the stack. I think you do not have to delete those images, since it would lead to a significant gap in your case. Instead you have to figure out why these images have 0 mean amplitude, which is nearly impossible. Every time I have had this problem, I looked up the concerning image and it came out that they hold 0 values in each pixel, which makes sense looking at the error, but which should not be the case.

I think the step where these images have been corrupted was the Apply Orbit File step, therefore my advise:

  • If you encounter this error start a new preprocessing
  • before building the single master stack, open all images in SNAP, look each one up, if the screen shows a single color for all values in the image, the image was saved in a wrong way
  • repeat the first steps with those images, until you have all images alright and ready for stacking
  • continue with the preprocessing, the error won’t come up another time.


Thanks @thho for your advice… :grinning:


My advice is in fact aligned with @thho as images with 0 images will dramatically stop your StaMPS procesing in a early stage and not satisfactory way.

For that you should either:

  • remove those files and re-run the mt_prep_snap script
  • try to re-process them to get them in the final analysis, maybe verifying slaves overlap the master splitted area (maybe any SNAP error happened during the processing of those images: i.e. java error, disk space ).

I hope this help


Dear Abraun,
I have exported three stacks of 12,12,08 images of 2018,2017 and 2016 respectively. (Total 32 Images)
I have combines files in rslc,geo,diff0 and dem folders.
Now i Have rscl (containing 64 items), geo(containing 6 items), diff0 (containing 93 items) and dem(containing 2 items).
each geo and dem folders have same items so skipped.
Please elaborate step 9 of Katherine’s work flow.
which results she is saying to “Export result to ENVI or GAMMA format”.


If you included lat/lon and dem bands in the topographic phase removal step, this is no longer necessary. You just need data in the geo and dem folders. But as you describe, these were written correctly.

But I still wonder why you made 3 separate stacks. Do they all refer to the same master image?
StaMPS requires them to have the same number of colums and rows. But the number of files you mention seems fine.