Hi all,
I am trying to mosaic same day swaths over a region of interest. My input folders are the daily SEN3 folders. Using the attached XML (which is edited from a file @marpet showed in this thread.
The process work fine, but it appears that for each variable the _FillValue is not applied. I tried a few edits to the xml expression line but that didn’t seem to work.
Thank you.
Jonathan
mosaic_eumetsat_olci.xml (7.4 KB)
Is the no-data (fill-value) missing for all variables or only for some of them?
It could be related to this known issue:
[SNAP-1448] Mosaic does not consider no-data properly in some cases - JIRA
Thank you for the response. I am not sure my issue is entirely the same, or if I explained my issue properly.
In my case, it appears that after mosaicking and plotting TSM_NN, invalid values have a range of values above 250 (with the original _FillValue ==255), and the TSM_NN values are far too high with a range of 126-130, when the “real” range (as seen in the swath data) is 0-10 (see screenshot below).
This is not the issue with the reflectance data, where the mosaic output and the original swath are the same. I am not sure if this is all due to a different FillValue for SPM and reflectance data, or that SPM is stored as log10.
Happy for any insight provided by this community! As a work around in the meantime I made my own mosaicking function using pyresmaple.
Thank you.
At least on part of your problem seems to be the referenced Issue.
I tried to replicate your use case. The results looked good after adding a condition. Using the flag WQSF_lsb.WATER ensures that only valid water pixels are considered.
...
</variable>
</variables>
<conditions>
<condition>
<name>condition_0</name>
<expression>WQSF_lsb.WATER</expression>
<output>true</output>
</condition>
</conditions>
<combine>OR</combine>
...
The other issue (wrong values) occurs when SNAP removes the log10 scaling from the values. In the original NetCDF file the values are log10 scaled, thus the unit lg(re g.m-3). In SNAP the log10 is removed and the unit is only g.m-3.
The devs should have a look at this. @diana_harosa @TomBlock