Sentinel mosaic

Dear all,

I seems there is a difference between Windows and Unix mosaic under Snap: the first one works correctly, the second creates wrong images.
Both system are updated to the latest 6.0.4 version.
Any idea?

Thanks in advance,

can you please demonstrate the difference with screenshots?

are you sure that the values are different? Sometimes, only the colour range is differently adjusted (because of extreme values, outliers, no-data definitions…)

Edit: Is the image below the calculated difference? Hard to understand without context. Maybe a different resampling is applied in Windows/Linux?

The difference is calculated only where are available values for OGVI.
In the southern hemisphere of left image (unix) there are wrong data not available into original S3A L2 OLCI and input files are the same in both results (and also the output resolution, 720*360 pixels).

I am not very familiar with Sentinel-3 products but are there not many quality indicators to mask all kinds of information? Maybe SNAP threats them differently in the corresponding operating systems…

When I created the mosaic I used the same criteria for both results and what is happened in the left image is very strange because it does not depend from quality indicators (also because there aren’t valid data into the compromised southern). It comes only from mosaic procedure.

Using snappy, I obtained the same wrong results with the creation of a resample product.
The problem seems the resample functionality using S3A L2 OLCI LRR (with Full resolution it works without strange areas).

Any idea?

The image above shows the reprojected figure on the left side and the subset in the right side.
The reprojected image contains some strips not contained in the original one.
@marpet: could you please help me?