Hi,
I am currently in the middle of running an ocean colour training course, and we noticed that SNAP does not retain geolocation of EUMETSAT level-2 OLCI WFR products after subsetting. The latitude and longitude of the extracted subset do not match the expected coordinates from the original product at all. When activated, the “reticle”, that is supposed to collocate across scenes is clearly drifting in the wrong place. The reported pixel lcoations for the same features in each image are very different. Correct geolocation is only retained if the top left corner of the image is included in the subset; implying that any extracted box starts at i-0,j=0 for the lat lon grids and not as the i/j index of the actual extraction point.
Worryingly, we were also able to recreate this in SNAP 12. The scene we are comparing is:
S3A_OL_2_WFR____20241024T154049_20241024T154349_20241025T220902_0179_118_225_2520_MAR_O_NT_003.SEN3
The level-1 product used to generate the above level-2 product subsets without an issue.
Oddly, pins will transport between the two images at the correct point, which seems to indicate that the geolocation of the actual product is ok, but the lat/lon used to report the location of the image in SNAP is the problem.
Cheers,
Ben