RCM terrain correction issues

[Deaner]

Hello,

I wanted to revisit an issue that myself and it seems several other people in the forum have had over the years.

There appears to be an problem when terrain correcting various types of RCM imagery. In my case, the images are ScanSAR 50 or 100 GRD images. Here is a list of various scenes that fail for me:
RCM1_OK2034581_PK2275823_1_SC100MA_20220928_081406_HH_HV_GRD
RCM1_OK2135956_PK2324192_1_SC100MA_20221103_081208_HH_HV_GRD
RCM1_OK2135956_PK2324192_2_SC100MA_20221103_081208_HH_HV_GRD
RCM1_OK2258374_PK2412317_1_SC100MA_20230111_074938_HH_HV_GRD
RCM1_OK2144244_PK2400813_1_SC100MA_20221230_074939_HH_HV_GRD

I have subsetted and calibrated previous to the terrain correction and am fairly certain it is unrelated to anything in these steps, as there are other ScanSAR scenes over my study region that successfully terrain correct. During terrain correction it seems to get hung up with CPU usage becoming very high, but progress seemingly halting. Out of 100s of scenes I have downloaded and am pre processing, the issue appears to happen approximately once for every ~5 to 10 images.

Links to unresolved threads of others who potentially have had the same issue:

Terrain correction failing sometimes for RCM radar imagery

Hello, I’m wondering if anyone has insight about why terrain correction is failing for some RCM images (working for most images, but failing occasionally - always on the images I need!) I’ve tried processing problematic images on multiple computers, and through the gui and command line interfaces. The process stalls at ~10% complete, and uses most of the CPU and memory on the computer. This occurs even when I subset the image to a small area, and no matter what DEM I select. Note that I can lo…

RCM Low Noise 100m Terrain Correction Issues

Hi! I’m trying to perform Terrain Correction on RCM (RADARSAT Constellation Mission) Low Noise 100m GRD images and processing freezes or throws an error. My pipeline is as follows: Calibration with only one polarization Amplitude and Intensity bands selected. Range-Doppler TC. As my images are in the polar regions, I’ve tested all the DEM available (e.g. all except SRTM). Other functions, requiring DEM: Radiometric Terrain flattering, SAR Simulation, SAR Simulation TC stack as well. I’m run…

RCM processing crashes when projecting (ellipsoid correction)

I am trying to process some RADARSAT Constellation Mission (RCM) SAR images. I am able to successfully calibrate and speckle filter them, but when I project them (Average height range doppler Ellipsoid Correction) SNAP just gets stuck. I have tried with several images and the processing bar gets stuck at around 20% (I left one running for about 8 hours and it made no progress). This happens both using the graph builder/batch processing and doing the task individually on the main GUI. Any ideas…

Any thoughts or help appreciated. I’ve reached out to the Canadian space agency and they think it may be an issue on the software side, though I’m not convinced.

[diana_harosa]

@djagula can you have a look at this? Thanks

[jun_lu]

A JIRA ticket ([SNAP-3724] - JIRA) has been created to track the issue. We will look into it. Thank you.

[Deaner]

Thanks for taking a look.

I see that there has been a fix committed. Is there a rough time frame of when it may get released in an update?

[jun_lu]

The issue has been fixed and the fix should be in the next release. I am not sure about the time of the release. But it should be very soon. Thank you

[Deaner]

Hi, just wanted to check in if there is any update on when the next release may be?

Thanks.

[jun_lu]

I think the fix is already in SNAP 10 release

[Deaner]

It seems I’m still encountering the issue. The progress bar looks like it might be going further, but for a scene I just tried to terrain correct again it failed to complete after several hours (usually takes less than 10 minutes). This time around CPU didn’t get to 90% plus, which it had in the past so maybe something has been changed?

Would the fix be part of the Microwave toolbox plugin update? My application is up to date.

[jun_lu]

The fix is actually in SNAP 11 release which should be soon. Sorry about my previous response above.

[Dawson]

Looking forward to the SNAP 11 release! I too am encountering this bug with certain RCM SAR scenes, and I really appreciate how quickly the bug was tracked down and fixed.

As a temporary workaround I’ve found that I can configure the snap.parallelism to some arbitrarily large value (say 1200). The processing still hangs indefinitely; however, I can see that eventually the output Geotiff stops growing in size and is valid. I then exit snap (to kill the processors stuck in the infinite loop). Opening the output Geotiff file, I can see that most of the swath product is processed–with some tiles missing. Usually, 80% of the scene data is available.

[Deaner]

No problem. Just happy to hear there is a fix in the pipeline. Do you know if it is likely to be weeks or months?