I wanted to download single bands from the collection Sentinel2_L2A, so I used as output format the netCDF one. But I was quite surprised of the result :
(red,green,blue,nir and then RGB)
I tried the GTIFF format but get the sames results…
job id (netCDF) : vito-j-fbaab6ecf7304bf2be91ee7a59b57cdd
job id (GTIFF) : vito-j-03a9ccfe967041448aed7cf09a45db53
I tried the same process graph with a synchronous download but get the same results too.
I also tried another temporal extent but I still get some of these white pixels / artefacts. (see below)
The thing that I don t understand is that in the openEO preview tool, everything looks fine, so something happens during the download I guess…
OpenEO preview of the job id (GTIFF) : vito-j-03a9ccfe967041448aed7cf09a45db53:
Hi Selim,
thanks, it seems that the data was fetched from sentinelhub, and we used data type uint16.
The derived from links indicate that Sentinel-2 L2A with processing baseline 4.0 was used:
S2A_MSIL2A_20221130T103411_N0400_R108_T31TFJ_20221130T163008
So my guess is that these white nodata pixels are actually negative values. The preview probably just has different visualization.
I logged an issue, please let us know if you have more info, or if we should prioritize a fix. (It seems a relatively low impact issue, assuming SENTINEL2_L2A is mostly suitable for land applications.)
As Jeroen said, on newer Sentinel2 some pixels can become negative.
When requesting those with SENTINEL2_L2A_SENTINELHUB, they get discarded as noData (probably showing as the bright background of your tiff viewer)
You can also do the same request with with SENTINEL2_L2A, negative pixels will be visible, and only pixels that have exactly reflectance ‘0’ will become noData.
The difference is visible in the following picture:
