Multitemporal mosaic of S3A_OL_2_LFR with per-pixel geocoding turned on

Hi,
I am trying to create a 300 m multitemporal S3A_OL_2_LFR composite using the Binning operator with MIN_MAX aggregator. For the composite to look fine, the OLCI per-pixel geocoding option has to be turned on. However, with this option turned on the super_sampling parameter of the Binning operator has no effect when the Binning operator is connected directly to Read operators, i.e. the Moire effect is visible and exactly the same regardless of which number super_sampling is set to. See the first screenshot below.

A workaround which I found is to first use Reproject operator on all the input images before using then in Binning. However, Reproject produces artifacts at the north and south edges of PDUs and those edges are than visible in the output composite. See the second screenshot below.

Is there anything else which I could try?

Thanks!
Rado


Moire effect


Reproject artifacts

1 Like

Yes, unfortunately the super-sampling is not effective. That’s a known issue (SNAP-1267)

But the artifacts you have at the edges should not be present with the latest SNAP 7 updates installed.

Thanks for the clarification @marpet! However, it appears the the reproject artifacts are still present in latest SNAP (7.0.4). With the attached graph I get the following output:

reproject.xml (2.0 KB)

Oh, I thought this was fixed. Then probably only in the SNAP8 release. Maybe I mixed something up.