PAR in VIIRS R2018.0 reprocessing

[gnwiii]

I'm comparing primary production calculations using monthly level-3 binned data between VIIRS R2014.0.2 and R2018.0.  In the new processing there are fewer bins with PAR values.   Looking at a level-2 file (V2017068164200_L2_SNPP_OC.nc) the  ice mask is blocky in the new processing and smooth in the image processed with version 7.4.

[OB.DAACx - SeanBailey]

It's masked for seaice...perhaps erroneously, but unless I'm missing something, the two images look the same to me - both have blocky blotches.

Sean

[gnwiii]

Oops -- they look the same because they are the same - somehow I managed to save the same image under two names.  Fixed now, but I realized the R2014 file was extracted and cuts off at about the ___location of the missing data blotch.  The real difference seems to be the ice mask resolution.   The attached image was generated using
the OCSSW 7.4 software.  The "ice edge" in the R2018.0 image is blocky, but smooth using the 7.4 software.

[OB.DAACx - SeanBailey]

The reason the data show a blocky patch is indeed the seaice masking.  We're using the ice field in the OISST ancillary file for flagging seaice.
Previously we used an ancillary file we generated from NSIDC inputs.  They two are not the same, although use the same SSMI source. 
In the image below, the right side is the current processing.  The left is the same scene processed with the "older" ancillary file.  Both mask
what could be considered valid data, but the current masking is a little more conservative (and blocky).  It is what it is :wink:

Sean

[gnwiii]

Thanks.  Knowing were to look, it is easy see the change in the L2 NetCDF metadata.   For a global level-3 data set, given that you are already using OISST, it makes economic sense to use the OISST ice field as well (one less file to download and store).