MAJA crashes with "no miniLUT has been generated for this angle zone" #14
Comments
|
Sorry to reply so late, your message arrived during my holiday period and I did not notice it. However, there seems to be something weird with the L1C data. The "angle zones" refer to the detectors (a line of S2 detectors has 12 linear detectors) which have participated to the observation of the tile. It is strange to see that an image was observed with detectors [0,3,4,5], as the numbers should be contiguous , probably [2,3,4,5]. I'll chek that. Did you process with start_maja ? Could you please provide the command line ? |
|
|
Hello Dion, Sorry you have to wait so long. Best regards, |
|
Sure. Which files exactly do you need from the SAFE? |
|
Hi, again quite a delay, but this time we've got some news: Can you please show us the contents MASKS folder of the Level-2 product just before the date where it failed? Sorry again for the delay! Kind regards, |
|
Sorry, I'm not quite following - where should I find the MASKS folder? In the MAJA If you mean the original SAFE, that one does contain the DETFOO files. |
|
To clarify, the MAJA processing chain rolls on happily if I remove this one product, so I doubt it has much to do with the products before or after. |
|
Yes Dion, we would need the DETFOO from the original SAFE. Thanks, |
|
Sure. Here's the whole |
|
Hi, is there any news on that issue ? We still have this problem. |
|
Dear @olivierhagolle , we got the same error, i.e.,
from a MAJA run. What we did find out is that there were two L1C scenes provided by ESA for this date (T34SEJ: 20190703), containing different parts of the same scene:
Hope this may give you an insight for these cases. Christina |
|
Hi Christina, |
|
Interesting. This might actually be why we were getting this issue in the first place - the split products might have issues with the metadata for the view angles etc. Those angles are given in the metadata XML files on a 5 km grid with slight variations for each sensor and maybe the production sometimes does not get them right on these products. From my experience, the same tile published in two parts is actually not a rare glitch, but a pretty frequent pattern. I can see you also discussed this elsewhere a while back @olivierhagolle: https://forum.step.esa.int/t/sentinel-2-weeding-out-bad-products/15041 So, does MAJA stitch the two together or only process one of them? I guess it all depends on the input script that does the bookkeeping and compiling of the lists of input files? And then you need to do the spatial merging afterwards... I think when I discovered this issue for the first time, I also tried to monkey-patch the products - merging the JPEG2000 files and trying to trick downstream software in believing these were the original products. But I was unable to get the file format and headers right with GDAL - another really nasty implementation detail in the Sentinel-2 product format... |
I run MAJA on some S2 scenes, and all work well except this one (
S2A_MSIL1C_20180724T103021_N0206_R108_T32UNF_20180724T142006). Do you have any idea what could be causing this error?Full output:
The text was updated successfully, but these errors were encountered: