Dear Ben, if you tried also to decrease the product coherence tresholds (which is different from the coherence itself), I think it's because 8 images are not enough to get valuable results. If you encountered an error, you would have a warning panel. Anyway, I haven't used yet SARscape on S1A imagery, but i don't think this is the probelm.

Best regards,

Pierluigi

Dear Pierlugi

Thank you for the reply. Do I set the product coherence thresholds at the graph connections step?

Just wondering, do you have any idea if PSinSAR takes longer to process than SBAS for the same number of SLC images ?

Regards

Ben

Hello Ben,

usually you cannot perform PS analysis with Sarscape with less that 20 images, while for SBAS the number is at least 8, if I remember correctly. Are you sure you are doing everything right? About the processing time of Sentinel data, processing the full frame can take an huge amount of time. I suggest you to first make a small crop of the images and check if the processing is working. At this point you can process the full frame.

Hope it helps,

Simone

Dear Simone

Thank you for the response. I will attempt your suggestion and post the results if I can get it.

On another note, for DINSAR and PSINSAR analyses, it will be recommended to use passes with the same viewing geometry (e.g. ascending/descending) right? What happens if I use 1 ascending and 1 descending pass? I have attempted this, and the results gave me smaller displacements. Why does this happen?

Dear Ben, you cannot combine for any DInSAR processing ascending and descending images. In that case, you're results are completely wrong. I suggest you to give a look at the several tutorials available on the SARMAP website before starting new processing. http://sarmap.ch/tutorials/PS_Tutorial_V_0_9.pdf

Best regards,

Pierluigi

Dear Pierluigi

I see. Therefore, would it be correct to say, for interferometric processing in SARSCAPE, SLC images have to be in the same view geometry (i.e. ascending or descending) as well as along the same relative orbit (or path)?

Hello,

to be correctly processed, the images must be in the same acquisition geometry (ASC or DESC) and of the same track. You can also process images of the same track but with different frames if you crop them in the overlapping area.

Simone,

I see, that clears things up.  'Track' means 'path' (i.e. Sentinel-1) right? That also means that 'frames' are really not critical I presume?

Hi Ben,

I'm encountering the same problem as you do/did. I'm processing 26 interferogramms of S1A and it gets stuck at 386/546 during the first inversion step of PS processing.

I'm curious to know if you resolved your problem by now?

It seems to me that the SubAreas are not created, and I'm wondering if it could be due to a DEM problem. Indeed, I performed the same processing before, using GEO-GLOBAL coordinates, and it did work (processing time ~ 1,5day). However, I had some issue with the created SubArea grid and couln't display it above my results. So I changed GEO-GLOBAL to UTM-GLOBAl coordinates for the DEM Import, and now it is stuck.

I would be grateful for any answer that could bring me further!

Regards,

Clémence

Clemence

Which DEM are you currently using? SRTMv4? Did you check the CPU usages? Mine was hovering about 14% (32gb RAM CPU). However, I ran out of patience after processing 21 scenes (no crop) after 3 days.

Are you using mixed frames? Subareas might be a problem. I intend to run a stack of same orbits and frames, while increasing the coherence thresholds to limit the candidates.

It is however strange to me that SBAS wasn't a problem, but PSI is.

Regards

Ben

Hi,

yes, I'm using SRTMv4. My CPU utilization is also of about 13-14% - same RAM- which seems to be way too low. For my previous processing -which worked-, the CPU utilisation was of about 96%!!

I'm using only ASC data, same track, and croped them so that I just have my region of interest. With SubAreas, I  mean the ones that are created on my cropped image during the first inversion step. As I understand, for you, there should have been in the first case 567 SubAreas created, but your processing got stuck by searching for the best PS in your 245th SubArea. Maybe it is indeed a threshold issue, but I would think that SARscape would produce an error message if it was so...

Let me know if your new process works!

Regards,

Clémence

Hi Clemence

I have tried to increase the coherence in hopes of reducing the number of preliminary candidates, but unfortunately the processing is still stuck. I have also perform the sampling using the general tool under the sarscape module. I will try another DEM. Any luck on your side?

Regards

Ben

Hello everyone,

I'm working on 23 slc scenes sentinel1-a with SARscape 5.2 software for Envi 5.3 in the process of P.S. Interferometric stacking.

I'm working on one Burst enclosing my area of interest.

As regards the software processing times, I am the following (the process is still in progress):

Connetcion GRAPH: about 30 minutes;

Interferometric PROCESS: about 3 days;

INVERSION FIRST STEP: 7 days and is still being processed.

I'm using a machine with 32 GB of RAM.

The process goes on smoothly although it is asking the infinity of time.

A results will update you about it.

Greetings,

N.A.F.

i have managed to run the processing of 21 scenes of S1A, took about 5 days. Huge chunk of data space being used up!

Hello, I've read you've been using Sarscape 5.2 . Is there a way to get this version of the software? (I need 5.2, not a  lower one, since from version 5.2 Sentinel-1 import module is correctly implemented).

Thanks

Hi Sebastian,

Please check if your institution has it. Else, you can use open source sar processors.

Folks, some processes in InSAR stacking use the GPU not CPU. Another issue is constant input/ouput (read/write to temp files). If you are finding processing is taking a very long time for some steps you can try to use an SSD for the temp file disk and improve your GPU speed (by a better graphics card). Not everything is determined by CPU speed and not all processes are optimised for multi-core multi-threading.

Hi Ben L Clemence Dubois @Sebastian Balbarni

I am also facing the same problem of PS Analysis: First Inversion "preliminary candidates searching". Its stuck at 808/1131. I took 30 S1A images cropped properly to required area. It took 3 days for Interferometric Process. My CPU usage is only 6.3% for main_persistent_scatterers.exe.

With RAM of 128 GB and Nvidia Quadro M2000 and 2 - Intel (R) Xeon (R) CPU E5-2630 v4 processors. Any of you who succesfully did PS First inversion, please suggest any solution or just i should wait.. Regards

Hello Ben L ... I too faced the same problem, what I tried to change was "SubArea for Single Reference Point" and it worked...However if we double the area i.e. from 25 sq. km. to 50, Reference points are reduced to almost half. (in my case Reference points were reduced from 1131 to 567). Now I am going through Second Inversion. Changing SubArea to 16/25/36/49/64 may work, give it a try. All the best.

HI does anyone has access to sarmap dataset given with PS_tutorial.

I really need to see their output. My output doesnot give good results.