Improved cloud and cloud shadow masks?

[wrignj08]

Hi,

I would love to see improved cloud and cloud shadow masks for the Sentinel-2 collection. I see there have been some requests over the years to add s2cloudless, but may I also suggest OmniCloudMask (OCM). OCM offers cloud (both thick and thin) and cloud shadow identification with SOTA accuracy, high throughput and compatibility with variable processing levels and imagery sources (Sentinel-2, Landsat, PlanetScope, Maxar among others).

I assume part of the reason s2cloudless has yet to be implemented is that the Planetary Computer only maintains an archive of Sentinel-2 L2A data, while most cloud and cloud shadow processes require L1C data. OCM offers a solution to this by being natively compatible with both processing levels.

The integration of OCM would significantly enhance the usability of Sentinel-2 data for researchers and practitioners who need reliable cloud masking.

The OCM paper can be found here, including accuracy comparison to s2cloudless among other approaches.

I believe this addition would be a valuable enhancement to the platform and would appreciate your consideration of this suggestion.

I would be happy to provide technical assistance to any implementation if required.

Thanks,

Nick