Loading differences between collections

When loading data in the VITO backend using the same bbox:
{'west': 590647.7761173819, 'east': 595388.4372280851, 'south': 5490355.029930396, 'north': 5494155.316432625, 'crs': 'EPSG:32633'}

I get two different areas, where one of the two is downloaded using the same crs ("SENTINEL2_L1C_SENTINELHUB"). This netcdf uses the correct bounding box.

The other ("GLOBAL_SURFACE_WATER") is using EPSG 4326, and does not adhere to the bounding box (the bounding box of this downloaded netcdf is much larger). Any ideas how / why?

The current behaviour on vito backend is that downloading a single netcdf results in the full chunk that was processed rather than cropping it to the bounding box.
This is something I might be changing, as users are always confused by it.
Geotiff does crop to the bounding box.

Thanks for the explanation. Could I have found it anywhere in the docs?
Good to know!

Not in the docs, but perhaps on this forum.
Basically, when implementing like that, I reasoned that people wouldn’t mind to get a bit of extra data, but somehow it does seem to be a problem.
Google Earth Engine for instance also takes some liberty in interpreting what it will return.

Yes I know. Same with images and the clip functionality there. Also gives some confusion. In this case it was just some confusion for me, I do not mind it.