Dear Simao Coimbra

See if this video can throw light on your problem: https://www.youtube.com/watch?v=FcETq8OWM0k

Regards

I think Sen2Cor plugin in SNAP will provide an effective tool for your problem.

Dear Simao,

Sentinel level2a products provide BOA reflectance (but multiplied by 10000). Try to divide your values by 10000.

Regards

Deaer Juan Miguel Ramírez Cuesta ,

There are many pixel values larger than 10000. Was that because of the atmoshperical correction error? Could we just regard those pixels as invalid observations?

I downloaded the L2A data from Google Cloud Platform.

https://console.cloud.google.com/storage/browser/gcp-public-data-sentinel-2/L2/tiles?pageState=(%22StorageObjectListTable%22:(%22f%22:%22%255B%255D%22))&prefix=&forceOnObjectsSortingFiltering=false

Thanks

Dear Haijun Li reflectance values might exceed 100% in case of very bright (e.g. specular, dry areas, deep convective clouds, etc.) surfaces.

In order to filter them out you could also use the Scene Classification map embedded in the L2A product to find out which kind of surface is impacted.

Best.

Dear Enrico Giuseppe Cadau ,

I could check out the details in the SCL layer.

However, to simplify the process steps, will it be reasonable to just ignore those pixels values exceeding 100%?

Thanks.

Haijun Li depending on your purposes, yes, it could be a good work-around.