ALOS2 processing from ISCE to Mintpy

[Jianming Kuang]

Hi Yunjun,

I pre-processed a stack of ALOS-2 data with ISCE for Mintpy post-processing. However, there is no unwrap file in the output after running the final step run_9_igram. I checked the processing and found that it showed out of memory when running with snaphu in the unwrap step (Please see the error below). But there are still 500GB left in my PC which should be enough.

Running: unwrap

snaphu v1.4.2

Reading wrapped phase from file /home/jerry23/Desktop/ALOS2/Igrams/20160815_20160912/filt_20160815_20160912.int

No weight file specified.  Assuming uniform weights

Reading correlation data from file /home/jerry23/Desktop/ALOS2/Igrams/20160815_20160912/filt_20160815_20160912.cor

Calculating smooth-solution cost parameters

Initializing flows with MCF algorithm

Out of memory

Is it possible that the data size is too big so that snaphu doesn't work since I used small multilook factor (2:1) only? Could you please help me about this issue?

Thanks a lot.

Best regards,

Jerry

[Zhang Yunjun]

Hi Jerry,

It's due to the limited amount of memory size in your computer, as the error message said, not due to the available disk space. Phase unwrapping of big-data is a challenge. The easy solution is either to use a big machine with a lot of memory or to use a smaller data size, by applying more multilooking or cropping into a smaller area, for example. This is really an ISCE processing question, you might want to check the ISCE forum.

Yunjun

[Zhang Yunjun]

Another option is to change the hardwired setup/configuration of SNAPHU in stripmapStack/unwrap.py. Here are some useful discussions from SNAP forum on SNAPHU configuration:

https://forum.step.esa.int/t/snaphu-parameters/8246

https://forum.step.esa.int/t/snaphu-unwrapping/341

Yunjun

[Russell Grew]

At the ISCE level you could consider using ICU in place of Snaphu and then taking advantage of the various phase unwrapping correction techniques from MintPy, if you feel the ICU outcome needs this.

At a glance I cannot see this icu option for ALOS but it is in stackSentinel.py and you may be able to access the routines.

The "memory" or RAM in your PC is probably 8 GB or 16 GB and not related to the 500 GB disk as Yunjun states. You could buy more RAM.

Cheers.

[Jianming Kuang]

Hi Yunjun and Russell,

Thanks for your valuable advice for this issue. I will give it a try for both options: smaller data size and ICU method.

Will inform the result here for further discussion.

Thanks a lot.

Best regards,

Jerry