Problem in Snaphu Import section

Gopika 2015-10-13 09:49:01 UTC #21

Hey,
I have the same snaphu import export problems as described in this section and would like to share my list of problems in hopes to find a solution for them. I used the SNAP toolbox to process (coreg_ifg_deb_dinsar_flt) two images of the sept 2015 Chile earthquake (20150731 and 20150917, IW3, Rel orbit 156).
- I usually run the SNAPHU export and other SNAP processes on the SNAP 2.0 beta-07 toolbox (installed on a 16 GB RAM (Intel(R) Xeon(R) CPU E5-2620 v3 @ 2.40 GHz, 2 processor) computer with 64-bit windows 7 system.
- The snaphu unwrapping on a VM on this computer never worked. Out of memory and my computer froze each time.
- I am now trying the snaphu unwrapping on another 32 GB RAM, 8 processor system (Intel(R) Xeon(R) CPU E5-1620 0 @ 3.60 GHz) with a 64 bit Windows 7 operating system where I have a UBUNTU Virtual machine configured to use 30 GB RAM and 4 CPUs processors.
- Snaphu unwrapping using MST goes on for days and days without stopping, and the snaphu unwrapping using MCF stops after 2-3 seconds with a 'OUT OF MEMORY' error.

Is increasing the RAM or dividing the subswath image the only solution? And if yes, how much should the RAM be increased to? For me, even 32 GB does not seem to be enough.

Having said that, I must also point out that I once cancelled the snaphu unwrapping (MST) after 2 days and tried importing the hdr product into the SNAP toolbox. This did not work, and after reading this section, I tried the snaphu import of the same product in the S1-toolbox. This worked, but of course, the snaphu unwrapping was extremely coarse but atleast now I know that the unwrapping with MST ran for a short while before encountering a memory problem and freezing.

I would appreciate it if the other people who had similar problems could share their solutions, whether it is increase in RAM, processor, VM configuration etc.

Thanks for taking time out to read this!

Fran 2015-10-13 10:07:25 UTC #22

Hey Gopika, I had similar problems until I created an even smaller subset of the area, ran snaphu export and executed the unwrapping further on Cygwin and there seems to be progress, it is much faster now. I am working off of a 16GB RAM but increased the size of my Virtual Memory using steps outlined here: http://windows.microsoft.com/en-za/windows/change-virtual-memory-size#1TC=windows-7 which solved my "out of memory" problem.

Maybe you should try sub-setting even further too.

Gopika 2015-10-13 11:15:35 UTC #23

Hey Fran,
Thanks for the brilliant tip. How much did you increase your virtual memory to?

Fran 2015-10-13 11:26:42 UTC #24

I increased it by 120 GB, 60 GB from the C drive on my computer and another 60GB from the D drive. It has not given me the "out of memory" error since the increment.

Increasing the Virtual Memory makes the computer run slower than normal but helps with memory shortages.

Gopika 2015-10-13 13:36:06 UTC #25

Thank you Fran!! I increased my Virtual machine's memory to 169 GB (49 MB recommended) and ran the same snaphu unwrap routine (MST) but no joy! MCF still stops immediately with an out of memory error. I guess I will now look into splitting the images further. Thanks for the help though!

lveci 2015-10-14 08:05:59 UTC #26

Some of the memory hogging comes from the GeoTIFF reader for the Sentintel-1 reader. I've added an option to use a file cache to conserve memory. It'll will be available in teh next release.

Fran 2015-10-15 06:35:58 UTC #27

Hey guys,

Has any of you managed to create a DEM from the unwrapped interferogram phases? How does it look? My DEM does not look good at all, I am a bit confused, I was expecting something a bit more than this. See the attached before and after geocoding. I understand that the amount of coherence plays a role in how my output DEM would look but the elevation values in the color scheme for the DEM do not make sense either, they don't correspond to values on google earth. I am assuming that I made a mistake somewhere with my processing.

mengdahl 2015-10-15 08:10:51 UTC #28

For DEM-generation high coherence is needed, say 0.75 and above. Even with perfect coherence the DEM will still suffer from distortion caused by atmospheric artefacts. These artefacts can be mitigated by averaging many generated DEMs.

Fran 2015-10-15 08:30:12 UTC #29

Alright, Thanks Mengdahl.

Fran 2015-10-20 14:20:28 UTC #30

Hello,

I created a much better DEM by selecting a much flat lying area with higher coherence. I want to know the expected resolution of the output DEM and the accuracy of the DEM generation tool in SNAP or does it vary with every output? If it varies with every output, how to I estimate the accuracy of my results?

Thanks.

djl 2015-11-15 13:34:42 UTC #31

hi Fran
thanks for contributing! I am used snaphu to unwrapping.and i have seen your reply. but i still can‘t finish my work! Do you have a better way to unwrapping than before! Could you give me some advice？
thanks a lot！

Fran 2015-11-20 08:03:49 UTC #32

Hello djl,

On free softwares i have only used the snaphu unwrapping method with cygwin command line which was successful, you can try using MATLAB but I don't fully comprehend the procedure. I have also unwrapped using the new ENVI SARscape 5.3 and Imagine SAR Interferometry toolboxes but these toolboxes need to be purchased, try to see if you can get a software trial for either of them by maybe contacting esri and hexagon.

Or maybe explain where you had difficulties with the snaphu through cygwin unwrapping and I will try to help you with it.

djl 2015-11-20 10:20:24 UTC #33

thank you Fran！！how much time do you cost on unwrapping?

Fran 2015-11-24 09:42:21 UTC #34

Hello,

Depending on the scene subset, it usually takes a couple of hours.

djl 2015-11-24 13:36:41 UTC #35

thank you very much! Fran!
On the beginning, it takes me one day to unwrapping, and end in failier!
However, when i take a step of "Multilooking" , the result is satificatory, it only takes about one to tow hour!(RAM:16G, Mulitilooking:32:8, data: Single polarization)!

Fran 2015-11-24 13:41:26 UTC #36

Alright djl,
Glad to hear that its fine now.

gabrielaquintana77 2016-02-04 20:39:36 UTC #37

Hi all

I have the same problem that Medhi: I did the export step, but I do not know which file must be inserted in first tab of "Snaphu Import" (1-Read phase). When I insert the Phase file created in the export step, I receive following message:
"3-SnaphuImport: SnaphuImportOP requires at least one product with InSAR metadata"
When I insert the Phase in dim format, I have the following message: "Operator 'SnaphuImportOP': source product not found".

I have problem with adding metadata of Phase file to UnwPhase file.

Now i am using NEST 5.1 with the tutorial Subsidence monitoring (NEST), i hope can solved!!!

Best regards

Gabriela

mengdahl 2016-02-05 09:39:31 UTC #38

NEST is an old unsupported toolbox - the current ESA toolbox for interferometry is SNAP 2.0. It is highly recommended that you switch to the new toolbox.

gabrielaquintana77 2016-02-05 20:10:59 UTC #39

Hi all

I can not add the metadata of wrapping phase to the unwrapping phase!!! please help me with this!!!

I am using SNAP 2.0

gyLona 2016-02-08 10:07:38 UTC #40

I had the same problem. Upon one above mentioned idea (Oswaldoguax) I used Sentinel 1 toolbox (1.1.1) to the snaphu import step. For me it was working. You can do the whole process again, or maybe try to add only the unwrapped phase that you have already created.