+1，
same here. Yesterday I updated to 7.0.4 and all f a sudden the ESD does not work anymore and gives message as following:
"
java.lang.ArrayIndexOutOfBoundsException: 0
at org.esa.s1tbx.sentinel1.gpf.SpectralDiversityOp.outputESDEstimationToFile(SpectralDiversityOp.java:1328)
at org.esa.s1tbx.sentinel1.gpf.SpectralDiversityOp.estimateAzimuthOffset(SpectralDiversityOp.java:778)
Caused: org.esa.snap.core.gpf.OperatorException: 0"
I don’t know if this is a case by case problem. It seems not be noticed by users?
In my case, 3 PCs, both Windows and Linux have same problem.
not sure about your case.
I tried ESD again and found the error happened when only one burst coregistrated image were processed. It works fine with images more than one burst.
==============edit====================
so now my work around is to include two bursts before ESD stage and subset later.
To make this clear also in here:
ESD should and can only be applied to TOPS products with more than one burst.
Thanks for the information! Did not realize this before your post.
It would be helpful if snap can give some tips in error messages or check burst numbers before ESD process since most of the time we are trying to limit our study area as small as possible.
1 Like
I know about what you said, and I’m sure that I have multi brusts.
In new version with s1tbx ver. 7.0.4, this problem seems be fixed. Thanks for your help.
I have multiple bursts in my image but it still gives the Error: Executor already shutdown. So a lack of multi burst isn’t the only issue. This started since i updated to 7.0.4. I have had to reinstall SNAP to 7.0.0 to use the tool. Very inconvenient
we will look into this.
Hello everyone,
I am also having the same problem. I am using SNAP 7.0.4. I split, applied orbit and then coregistered 54 SLC SAR images and then tried to use the Enhanced Spectral Diversity, but then it gave me this error. Can someone help me find the solution? Is the number of images too much?
Thank you!
please check if the coregistration with BackGeocoding has produced a valid output. Can all bands be correctly displayed?
Hello!
All the intensity bands are able to display (I think correctly also). Also, a correction from the previous message: I am using SNAP 7.0 (not 7.0.4). One notable thing: The master image shows the ocean, but the no slave images show the ocean. I have never seen anything like it before. Can you please help me?
Image of master and slave images are attached as example.
Thank you!
MASTER IMAGE:
SLAVE IMAGE:
what were the steps you applied before BackGeocoding?
Hi!
I downloaded 60 images from scihub website (https://scihub.copernicus.eu/dhus). Then I used the batch-processing tool to split and apply the orbit file to all the images. But SNAP gave error on 6 images, so I deleted them and applied batch-processing on 54 images, which gave me 54 XXXXXX_split_orb files. Then I used the Back-Geocoding to coregister all images, with the first date as the master image. Stack also formed without incident. But after that, ESD gives “Executor Already Shutdown” error, and when I tried to generate interferogram without ESD, it gave me “Java.lang.NullPointerException” error.
Graph used for batch-processing (split and orb) is attached for reference.
Thank you!
are you using images of both Sentinel-1 A and B? Sometimes they are shifted along track, so the burst selected in the graph might lead to different area coverages. Just a thought
Hi!
I am only using Sentinel-1A data from same orbit number. The acquisition is over the same area. I will try to reduce the number of images in a stack and then try to process again.
I have 1 question though: Right now I am using 54 SAR images. I am planning to divide these 54 images into 3 stacks of 18 images each. The master will be same for all 3 stacks. Can you tell me how to combine the 3 stacks so the PSI analysis will yield a velocity map? Is combining even possible?
Thank you!
New update: I created the stack of 18 interferograms. Still the problem is same when I tired ESD. I am planning to re-install the software and then try.
do you want to reduce the number of images per stack because of higher coherence or because of computational reasons?
If the first is the case, I recommend creating three separate stacks with the same reference image from the beginning and then perform the PS analysis for each stack with the same parameters.
Merging them during the PS process would somehow contradict the principle that the initial PS selection is based on the amplitude diversity (which would be different if a larger time-series is used).
I wanted to reduce the stack size for computational reasons. The stack of 54 SAR images yielded the lowest coherence of 0.29, but I was initially planning to keep it as it was. Although the reduced stack (18 images) is not working either, as of now. Once I re-install the software and if the stack works (or doesn’t), I will let you know. I am planning to install SNAP 8.0 now. Hopefully it will work.
Thank you!
the longer the time series, the lower is the coherence of pixels, because it is computed over a large range of images (potentially containing changes). Long periods will give you less PS but the remaining ones have a very stable information and will contain only little noise. So I wouldn’t recommend splitting unless you want to separate the analyzed periods.
But maybe you can create three stacks in SNAP and then apply the StaMPS export into the same target directory in the end.
This seems like a good idea. But let me first re-install the software, because the process has constantly failed since ESD. I cannot go forward from ESD because of the error (Executor Already Shutdown). This error has persisted for both big and small stacks. I will let you know what happens after re-installing.