How can I control quality of each step in SNAP software to decide pass or fail that step?
for example how can I find out ,if “Apply orbit” pass or fail ?
Or “Back Geocoding” works fine ? ( any logfile or method )
If there is any method, how can I interpret it.
TOPS Split: If the output has only VV and a smaller extent, it worked
Apply Orbit File: Check under Metadata > Abstracted Metadata > Orbit State Vectors . The number of entries (orbit_vector1, orbit_vector2) should have increased.
BackGeocoding: Create an RGB image to check if the two images are correctly aligned (example)
Enhanced Spectral Diversity: Open View > Tool Windows > Radar > InSAR Stack Tool and check the RMSE of the coregistration.
Interferogram generation: You can visually check if the inteferogram has good quality (visible fringes, little noise, no trends) and if the coherence is sufficiently high throughout the image. Low coherence indicates phase decorrelation (bad quality). Also select “output elevation” to check the quality of the DEM which is used for topographic phase removal. If it has gaps or errors, the topographic phase will be estimated incorrectly as well.
Unwrapping: Also mostly checked visually - are all areas unwrapped correctly, are there any artefacts (introduced by tiling, check the last page of this tutorial)
Phase to displacement: How does the overall displacement pattern look - does it have overall trends superimposing local displacement. Is the displacement band normalized correctly? See page 25 of this tutorial for an example.
Range Doppler Terrain Correction: Not much to check here - it’s simply a geocoding based on a DEM. Again, does the result look plausibe?
Most of the steps do not result in a right or wrong output or deliver a numeric quality indicator. It is rather the task of the user to question if the processed intermediate products make sense and if they are free of visual errors before proceeding to the next step.
For Back Geocoding , I think it’s very hard to determine co-registration quality all over images with this method.
But I have another question .I need to build a stack of sentinel 1a images for PS method in an urban area ( SNAP+StaMPS). In the Interferogram formation , I have a green patern ( attached image ) at the end of the bursts. To solve this problem, is it necessary to use "spatial subset from view " for each interferogram (all stack interferograms)??
Please let me know if you have another advice.
you can remove it from the product with the spatial subset, yes. But only after debursting and also the extent of all products for the StaMPS export must be exactly the same. That means the stack of intensities must be cropped accordingly.
I have other questions about PSI method (SNAP + StaMPS ) .
1 - for example, my AOI is between IW1 and IW2 ( bursts = 3 4 5 ).
I should process each IW Independently and finally merge then.( 19 images with 1 master)
Now I have some green patterns at the end of the interferograms ( IW1 BURSTS) ( like my last post ).If I use SPATIAL SUBSET to cut green patterns, In my opinion, it will not be possible to merge them( IW1 and IW2 buerst ).Is it correct?
Is there any method to make interferograms without green patterns in the SNAP ? (Nobody has used subset in most of graphs to prepare data for StaMPS in their articles. Does It means that their interferograms are correct or their graphs are general?)
2- I read some articles ( PSI method ) and found that most people use attached graphs to prepare data for StaMPS, If I want to use merge, I should add it in PART B in two points, first between topo phase removal - write ifg and second between TOPSAR deburst - write coreg . Is it right ?
I have never used several swaths in one analysis, so I can’t really recommend the best procedure, sorry.
Have you tried to merge IW1 and IW2 (including the green artefacts) to check how the result looks like?
Subsetting will probably create more problems than it solves. The green artefacts happen to occur in many studies, they are not a result of false processing.
Would it make sense to use ascending instead of descending (or vice versa) so your area is fully contained within one sub-swath?
I use the snap2stamps python package to prepare S1 data for StaMPS processing
It was once announced that future versions will support the merge of different subswaths