Scene Classification - SEN2COR

Hello guys,
I noticed that Sen2Cor processing compared to level 2A on the copernicus website has different results to “scene classification” produtct. I am using Sen2Cor version 2.8.0 and if I compare the products of both, they are different:



I realized that this happens for multiple images, and not just for a single date. I believe it is a BUG. Any solution?

I think this is the same reason as discussed here: Different colors in L2A images and here SNAP Tutorial
sen2cor outputs can only be compared with the preprocessed L2A products when the the same DEM is used and the ESA-CCI data package is included for sen2cor.

Do you think that different DEMs causes this incompatibility in this product? This worries me, because the classes are very different. I’m not sure which to trust.

I have always used the correction from Sen2Cor v8.0 in the images, but now I suspect which is the most appropriate. I use this classification to filter the clouded pixels of the images.

No, I think the DEM is the minor reason, but rather the CCI Landcover product which is additonally integrated. Again, please read the commend of Jan here: SNAP Tutorial

I wouldn’t trust any of them, to be honest, as they are automatically derived from every image based on robust routines. But they do not replace an adjusted classification. And the cloud masks are rather an indication for image quality, but also not ultimately correct.


I’m trying to download the ESA - CCI from the website ( , but it gives an error, with a message that the website is down. Do you have the executable to pass to me? Thanks a lot!

sorry, I have never used it so far.

Hi @AMuraro, to get a good cloud/shadows mask, you might try MAJA:

1 Like