L3A Artefact

I would like to know why there is some artefacts on the L3A composite on 20 m bands ?
I used 15 images along a year and some of them (3 or 4 images) have a completely clear sky without any clouds…
for info i put the value 25 for the parameter synthalf, don’t know if it have an impact, i don’t really understand this parameter.

Many thanks

Dear Anouk,
Did you check the L2A products used to generate this composite ? Can you see the same artefacts in one of the L2A ? I suspect it might be linked to the previous GIPP parameters (that we updated in the v1.8.2).
Please let me know,

This parameter is used to define the temporal window of the composite. The symetrical temporal window is defined with 25 days beofre and 25 days after the central date of the composite.


Thanks for your response,
I don’t see anything on L2A quick looks…
I am using the 1.8.2 version.
When I launched the L4A process, i observed the same shape artefact on the PCA tiff image… And when I make the crop mask with only cloud free images (5 images instead of 15), I don’t observe this error.

thanks for the clarification.

Hi, I don’t manage to resolve the problem, i tried to make a composite on an other tile close to this one and I also observe artefacts as in the previous capture.

Dear Anouk,
When you installed the v 1.8.2, could you confirm me that you did the full installation (including the GIPP files)?

Yes I downloaded all data correctly. For the moment, I uninstalled all the system and I’m waiting for the new version. I will re-install the new version properly and re-process my data next week, I will keep you informed.
I published another post related to the questions I asked yesterday during the webinar, L3A Composite multi tile
(posted from another account).
I thinks the two problems are linked.

Hi Anouk,

I would like to try and recreate those artefacts using the same inputs. Therefore, could you please tell us the tile ID as well as all of the dates of the products you used to in order to create the composite?

Kind regards,

images used are on the tile 31UFR (Belgium).
You will find here the list of inputs:

  • S2A_MSIL1C_20170126T105321_N0204_R051_T31UFR_20170126T105612.SAFE
  • S2A_MSIL1C_20170215T105121_N0204_R051_T31UFR_20170215T105607.SAFE
  • S2A_MSIL1C_20170314T104011_N0204_R008_T31UFR_20170314T104411.SAFE
  • S2A_MSIL1C_20170327T105021_N0204_R051_T31UFR_20170327T105021.SAFE
  • S2A_MSIL1C_20170516T105031_N0205_R051_T31UFR_20170516T105322.SAFE
  • S2A_MSIL1C_20170526T105031_N0205_R051_T31UFR_20170526T105518.SAFE
  • S2A_MSIL1C_20170705T105031_N0205_R051_T31UFR_20170705T105605.SAFE
  • S2A_MSIL1C_20170923T105021_N0205_R051_T31UFR_20170923T105022.SAFE
  • S2B_MSIL1C_20170707T104019_N0205_R008_T31UFR_20170707T104022.SAFE
  • S2B_MSIL1C_20170829T105019_N0205_R051_T31UFR_20170829T105633.SAFE
  • S2B_MSIL1C_20170925T104009_N0205_R008_T31UFR_20170925T104008.SAFE
  • S2B_MSIL1C_20171015T104009_N0205_R008_T31UFR_20171015T104525.SAFE
  • S2B_MSIL1C_20171018T105029_N0205_R051_T31UFR_20171018T105026.SAFE
  • S2B_MSIL1C_20171107T105229_N0206_R051_T31UFR_20171107T124713.SAFE

Thank you for the quick answer, I will try and reproduce it.
From what I can see already though is that you’re using L1C products from a bigger period than the synthalf of 25 (days) you stated earlier. Your first product is from January '17 and the last one from November '17, resulting in a synthalf of approx. 5months or 150 days.