Sentinel-2 processing baseline 4.0

113 views
Skip to first unread message

Paula Lightfoot

unread,
Oct 18, 2021, 12:07:59 PM10/18/21
to RSGISLib Support
Hello, ESA have announced a major upgrade to their Sentinel-2 Level-1C products which is now due on the 15th November: (https://sentinels.copernicus.eu/web/sentinel/-/copernicus-sentinel-2-major-products-upgrade-upcoming).  
They are releasing an update to SNAP software before this transition. I was wondering if you are introducing any changes to ARCSI to deal with these upgrades, or whether existing versions of ARCSI will be able to process them.  
We currently use ARCSI version 3.1.6 to produce analysis-ready S2 data.  ESA shared a sample of the new data which we have tried to process using our version of ARCSI, but the scene provided was 100% sand (which gets predicted as 100% cloud cover) so we are trying to get a more representative scene to test the workflow.  
Thanks, 
Paula

mr.samf...@gmail.com

unread,
Oct 21, 2021, 8:28:17 AM10/21/21
to RSGISLib Support
Hi

I am encountering the same issue as Paula running ARCSI version 3.1.6 on ubuntu 18.04.

The processing command I am using is:

arcsimpi.py
    -s sen2
    --stats
    -f KEA
    --fullimgouts
    -p RAD SHARP SATURATE CLOUDS TOPOSHADOW STDSREF DOSAOTSGL METADATA
    --interpresamp near --interp cubic
    --outwkt <projection-file>
    --projabbv osgb
    -t <temp-dir>
    -o <output-file>
    --dem <dem-file>
    -k clouds.kea meta.json sat.kea toposhad.kea valid.kea stdsref.kea
    -i <header-file>

I've found the following:
Product ID = S2B_MSIL1C_20210517T103619_N0300_R008_T30QVE_20210517T124830 (level 1c data from ESA SciHub using the current Baseline 3.0) generates a cloud cover estimation of <1%, which agrees with the ESA cloud-cover estimate.
Product ID = S2B_MSIL1C_20210517T103619_N7990_R008_T30QVE_20210929T075738 (level 1c data with the new Baseline 4.00 tag that ESA has provided as a one off sample dataset link) generates cloud cover estimation of 100%.

To be fair, I have not looked at the ARCSI source to understand if there is some dependancy file in the ESA SAFE format that the ARCSI cloud estimation algorithm needs that has changed between 3.0 and 4.0 which could be causing this?

Does anyone have any thoughts on what I can try next to investigate?

Thanks in advance,
Sam
Reply all
Reply to author
Forward
0 new messages