As the title said, I have 24 S1 images to process PS Insar. My steps are split --> apply orbit–> S1 back geocoding–> deburst–> subset–>interferogram–> topo phase removal–>export to StaMPS. However, at the deburst step, my PC have spend so many time(7 days passed and still not finish). I have opened up the task manager, and checked the usages of CPU, Memory, and Hard disk. I found that CPU was nearly idle, about 8GB memory was used by SNAP, and Hard disk was reading frequently without any writing. Is it normal? How could I improve the performance?
The specifications of my PC are CPU: Intel Xeon E5-2609v3 6 cores 1.9Ghz, Memory: 16GB, Hard Disk: RIAD5
Thanks for your attention
I think the problem is that you process to many images for once, in my case, I have 12G RAM, but I only process 5 images(the minimal number for StaMPS export) once a time, the more images you process, the slower the process speed, and if you process to many images, just like your case, you will run out of memory and like you said, hard disk was reading frequently, but no writings, the reason why you look only 8G memory was used is because all the rest of memory has become stand by memory, which you can understand it as kind of memory use.
And maybe you should do interferogram before deburst.
Hope this helps
Yes indeed, I also noticed that I have processed too many images. But as your case, you mean that I should separate my 24 images to several parts? And the master image should be included in each part?
The answer to your questions are Yes and Yes
Maybe this can help you:
Thanks for your reply, I will try soon!
I think in your processing chain, the interferogram should be right after backgeocoding.
Did you overcome with this issue? I have same problem and we need to use more than 20 SLC for PSI that was explained in the Andy’s PSI instruction. So that, I used 32 SLC slave and 1 master and stack proccess was finished succesfully. On the other side, I could not achieved debust process, I wrote 64 GB data but process was not progressing. So I am trying interferogram formatin first. Did you use any other method?
I am afraid I didn’t overcome this issue. I still strictly followed the traditional instructions to make sure everything would be right.
However, I found that, in some step(e.g. deburst), SNAP would take up a lot of memory, and it won’t be released by itself. If I closed the SNAP completely after each step end, the resources would be fully released, the processing would speed up significantly. In the second time, I spent about 2 days to finish the processes in SNAP.
Hope this helps
@YuL this problem was already reported on the forum by some other users. It is already being investigated and hopefully will be addressed in the next releases.