Hi all,
I am working with S3 OL_1_EFR, OL_2_WFR and SL_2_WST products in snap v7. All overlapping the same area.
I would like to remove cloudy pixels from my analysis in SL_2_WST_products.
As the quality_level_cloud flag does not seem reliable over my area of interest (there is a clear cloud over-flagging near the coast when comparing to clouds in OL_1_EFR or OL_2_WFR products acquired at the same time), I wanted to compute a cloud_mask from OL_1_EFR products with the Idepix OLCI processor. Then my idea was to export the output mask values to the corresponding (same ground track) SL_2_WST products. This used to work in a previous SNAP version. Now, the idepix cloud mask gets transfered to the WST product but I am then unable to open any band from the WST product!
Before exporting the mask values, I have subsetted both OL_1_EFR and SL_2_WST products to the same area of interest and reprojected them with the same crs.
I conducted another test by trying to export some mask values from a OL_2_WFR product to a OL_1_EFR product (same ground resolution, same sensing time) and even this fails. I get the following error (which is the same message I get with Idepix, only the symbol name changes):
Failed to open image view
java.lang.IllegalArgumentException: expression: undefined symbol ‘WQSF_lsb.CLOUD’
This transfer mask technique to compute a more suitable mask for WST products used to work really well.
Do you know what I could be possibily doing wrong here? If this functionnality is broken in Snap v7, has it been fixed in v8?
If none of the above works, would you have a suggestion to compute a more reliable (and less conservative) cloud mask for SL_2_WST products? I have tried using probability_cloud_single_in==1 but this is still too conservative. I have then tried using probability_cloud_single_in==1 and sea_surface_temperature<=empirical_threshold but this is not ideal either.
I would be very grateful for your help. I would be glad provide with more inputs/details if needed!