OLCI GeoTIFF Export bad georeferencing

When exporting Sentinel3 OLCI --> RGB view to Full Scene GeoTIFF, the georeferencing seems to be far off.
I compared with the Footprint given in the xfdumanifest.xml. I uploaded Screenshots from QGIS including the Footprints.

EFR is somehow in the right place but rotated (maybe that is a projection issue), but ERR is stretched from -5 to 202°.

$ gdalinfo RGB.tif [...] Corner Coordinates: Upper Left ( -5.4245711, 86.6952913) ( 5d25'28.46"W, 86d41'43.05"N) Lower Left ( -5.4245711, -65.4526406) ( 5d25'28.46"W, 65d27' 9.51"S) Upper Right ( 202.427, 86.695) (202d25'37.35"E, 86d41'43.05"N) Lower Right ( 202.427, -65.453) (202d25'37.35"E, 65d27' 9.51"S) Center ( 98.5012353, 10.6213254) ( 98d30' 4.45"E, 10d37'16.77"N) [...]

Using pconvert from CLI gives the same output.
pconvert -f tif -p tristimulus.rgb -o $OUTPUT_DIR xfdumanifest.xml

Am I missing something or what am I doing wrong?
Thanks a lot in advance.

1 Like

Similar problem for me, this is what I obtain exporting as GeoTiff in SNAP.
Advices welcome.

I just tried it and I have experienced no problem.
This is how the images look like after the export.

And I have a good geo-location when importing these files either in QGIS or in Google Earth.
For the RR:


For the FR:

I’ve used the latest updates SNAP 5.0.
In both cases I first reprojected to WGS_84, then created RGB and exported it using the Export View function. The products I’ve used are from the SciHub

  • S3A_OL_1_ERR____20170328T091755_20170328T100158_20170329T161045_2643_016_036______LN1_O_NT_002.SEN3
  • S3A_OL_1_EFR____20170328T092800_20170328T093100_20170329T161304_0179_016_036_2159_LN1_O_NT_002.SEN3

Which products have you used?

Hi marpet,

thanks for your Reply. I think I missed the “reproject to WGS84” part. I just created the RGB and exported it.
How can I do that from the CLI using pconvert and/or gpt?

Regards,
Rouven

You have to use gpt.

Call

gpt Reproject -h

for specific help on the reprojection operation and call

gpt -h

for general help on gpt.
In a separate step you can then call pconvert on the reprojected products.