I am reading Sentinel-3 OLCI Level2 in Python with netCDF4 library
So when i reading geo_coordinates.nc and Oa**_reflectance.nc, in some places of arrays i get repeating coordinates with different reflectance values (sometimes significantly). At the same time SNAP exporting data without duplicates.
I have read the documentation and the physical meaning of keeping duplicates is still not very clear to me.Unfortunately I have not found description about what SNAP doing with duplicated pixels. Calculating mean?
The duplicates in the geo-ccordinates are normal.
For keeping a consistent pixel size across the swath. This is roughly explained here.
In the level 1 data the radiances are duplicated too. There also exist a flag which masks the duplicates. The duplicates get lost in the reflectances due to the further processing and the flag is also not kept in level 2. Only in the geo-coordinates the duplicates are kept.
The duplicates can also be seen in SNAP, if you look into the latitude and longitude bands.
You probably can’t see the duplicates in the exported files because you use the tie-points instead of the coordinates for each pixel.
This is special for the Sentinel-3 data. You can choose between tie-point based GeoCoding and a pixel-based GeoCoding.
For the tie-point you have for each line only have 77 points and in between the values are interpolated. For the pixel-based GeoCoding you have coordinates for each pixel.
In the Options you can switch between both (Tools -> Options -> S3TBX).
I read carefully once again page about spatial resampling (link, you posted)
And OLCI Level0, Level1 ATBD. Unfortunately, these documents do not make it clear enough to me, what to do with situations, where two or three pixels have same coordinates, but visibly different values. Is it normal inaccuracy, given that this is the surface of the sea, not the slope of the mountain?