[kianicka]

Dear SARPROZ Team,
I am just starting to learn SARPROZ software together with InSAR methods, particularly PS InSAR processing with Sentinel-1 sensors.
I have already quite successfully processed stack of Sentinel-1 SLCs from Copernicus program, more than 100 images mostly following instructions in provided tutorial from 2018, from the SARPROZ Web page: Tutorial2018_QYTutorial4.pdf
There was quite a lot stable scatterers, good temporal coherence and GCP stability indicated by the histograms, that the peak is at zero elevation.
However, the areas of interest were a bit poorly covered by amplitude stable points (I have used the thresholds from the tutorial Amp.Stab.Index 1 – sigma/mu – 0.85 for APS, and 0.65 for sparse points processing).
Second, we have in the area already some corner reflectors mounted. None of them was recognized by the algorithm.
I have played around a little by lowering, the threshold, turning off the mask in the preliminary analyses step.
But they were still not identified, and lowering of thresholds start to spoil inputs by noise and start to lower overall temporal coherence.
I have found that there is possibility to generate a custom mask to e.g. process corner reflectors since 2022:
https://www.sarproz.com/groups/sarproz-group/forum/topic/sarproz-release-2022/

Question 1: However, I do not understand how to incorporate the artificial scatterers into selection of points for APS and Multi-Image Sparse Point processing. Even when I turned off mask generation in “Mask for sparse point selection”, there were still only about 10 000 points selected in the area 20x20km – and our Corner Reflectors were not in there.
I would need to merge about 20 corner refletors points with points selected by amplitude stability threshold and process them in APS estimation and then in Multi-Image Sparse Point processing modules.
Is this somehow possible in SARPROZ?
(I know that using Matlab one could manipulate in/out mat file – however, I do not have Matlab, only compiled version of SARPROZ)

Question 2: When conducting my experiments I compare results with European Ground Motion Service:
https://egms.land.copernicus.eu/
Very interestingly – I was using only descending orbits acquisitions and there, when displaying just descending orbits, 2 of 4 corner reflectors in vicinity of Ruska Nova Ves were identified. And yet, they were build by spring 2020, thus only part of computational time period was available.
We can see how the amplitude has stabilized. See attachment:
PSInSAR_EuropeanGroundMotionService_descending_RuskaNovaVes_CornerReflector.jpg
And in EGMS they were using exactly the same data as I was using in SARPROZ. How it was possible that they detected such point? We also see that there are several points corresponding to the corner reflector. And we also see that the corner reflector is stable without significant movement.
Contrary, from SARPROZ I have no stable points corresponding to the corner reflectors – seems like there is missing amplitude stability, see attached:
PSInSAR_processing_descending_RuskaNovaVes_NoCornerReflector_timeSeries_SARPROZ.jpg

Thank you very much for any kind of help.
Kind regards
Jan Kianicka
SGUDS

Attachments:

You must be logged in to view attached files.

[periz]

hi

* if you want to cherry pick only a few pixels of interest, you can use this module: https://sarproz.com/manual/kml2mask.html to generate your own mask of pixels, eg by importing a simple list of geographic coordinates

* if you want display data (like amplitude or coherence) of cherry picked pixels, you can use this module (visualization tools): https://sarproz.com/manual/class_sito.html

* if you want to process and analyze amplitude time series data (eg to detect temporary scatterers) you can use this module: https://sarproz.com/manual/amp_series_proc.html

all these topics are discussed in the Sarproz training course
best

[Author]

Posts