Hi,
I am using MAJA in the frame of the COPERNICUS HR-WSI project. In the frame of this project, we aim to use MAJA 4.8.0 with CAMS at a 120m coarse resolution as our atmospheric correction algorithm. To do so, we need to process the DTMs for all the processed tiles.
The processed L2As are inputs for various processing routines, such as LIS, the snow detection algorithm developed at CESBIO.
When doing round robins on the 32TNS tile, we observed some strange doodle lines on the Fractional Snow Cover product illustrated below.
Detail of an FSC TOC on the 32TNS Sentinel-2 tile computed using LIS 1.11.0 on a L2A raster computed with MAJA 4.8.0 with CAMS at 120m
After further inquiry, we sourced this issue from the slope mask of LIS, the input SWIR from MAJA (and the resulting NDSI).
Eventually, we tracked this issue up to the DTM ASP_R2 mask at least for the slope mask as we know that LIS uses it in its process. We do not have more information on why these same alterations are visible on the SWIR band though. Additionally, this seems to come mainly from the ASP_R2 mask as this artefact is not to be seen elsewhere, and provided that the ASP_R1 mask, though describing the same information at a finer resolution, shows neither negative values - which ASP_R2 does) nor limitroph artefacts.
We also noted that the negative values found in the ASP_R2 mask are coincidental with the limitroph artefacts and the slope mask aberrations.
We wonder where these artefacts are sourced and if they will be taken care of in future releases.
Best regards
Robin BURATTI
