I stumbled upon an issue with the OLCI data. It contains duplicated pixels and duplicated geolocations too. This leads to the case that geolocations are ambiguous.
When copying pins to an exact subset (without cropping) the pins are shifted by one or two pixels, depending on how many adjacent pixels are marked as duplicated. But the pins should stay at the same pixel location.
This is difficult to fix for SNAP, anyway I wanted to mention this. Maybe by ignoring the duplicated pixels in the geolocation bands and interpolating between the valid locations when reading the data.
I never understood why the duplication is applied to the geolocation in the OLCI products. Does not really make sense to me.
Hi Marco,
I see the issue and understand the consequences - technically.
The pin has a location in lon/lat which is present several times in the geolocation data. The measurements at each of these locations are duplications of a single geophysical measurement record - that is my understanding of the format specification.
As consequence, scientifically it is not important, which x/y location is chosen, as long as it is the exactly same measurement record. For serious processing, you would exclude the duplicated anyway.
Though, from an engineering point of view, it would be nice if the pin also is at a constant x/y location - the one on which it has been created.
But we won’t fix, it’s not a bug, instead it is a consequence of the redundant data storage.
Tom
1 Like
I agree that is not really a bug and actually an issue of the data.
That’s why I said:
It can be very confusing for users who don’t have the deep insights. And I also posted this to make the ESA aware of this problem.
However, thanks for your reply.