How is L2A data generated for AWS?


I downloaded S2B_MSIL2A_20201115T185649_N0214_R113_T11TLM_20201115T211121 from s3://sentinel-s2-l2a/ using the “sentinelhub” Python library. I wanted to compare the B04 image against the L2A B04 image I have created locally using Sen2Cor 2.5 (with default configuration). The results seem to be different and the locally generated image seems to be a better image visually that that retrieved from AWS.

While I double check that I haven’t done something stupid my end It would be useful know what process is used to generate the L2A data in AWS.

There is some discussion here ( about using the MTD file to work out which version of Sen2Cor was used but I don’t see an MTD file under the product or tile key in AWS. I do see one for the locally generated data.

The original announce ( talks about using Sen2Cor in some areas but also talks about using “a run-time optimized statistical atmospheric correction”.

How do I tell what has been done to produce each L2A product?

Thanks in advance

All L2A data on AWS after December 2018 are processed by ESA and replicated to AWS. ESA is using Sen2Cor 2.8. I suggest you contact them for the details.

Ok, thanks for the response

I have a problem to run Sen2Cor on Sentinel-2 L1C images from AWS-sentinelhub.
There is no MTD_MSIL1C.xml file!! That is the data I have.

Not sure if you are aware, but there is full archive of Sentinel-2 L2A data available on AWS, globally since 1.1.2017.

For the mentioned file, see this thread: