Level-2 format change?

[dem1]

Hi,

It seems that recently there was a change in the format of the OC Level-2 products, for example the /navigation_data/cntl_pt_cols variable and pixel_control_points dimension was removed, do you confirm this (example: AQUA_MODIS.20240708T033001.L2.OC.NRT.nc)?
Are these format change specified somewhere (I did not found anything)? because these changes have impacts on our Copernicus processing chain.
Are these changes applied to all sensors?

Thanks,
Julien

[dem1]

It seems that only MODIS-A L2 ares affected for the moment, both VIIRS did not change.
We use cntl_pt_rows to detect the start of a MODIS bowtie using this equation ((cntl_pt_rows - 1) % 10) == 0
How can we detect bowtie with the new L2 format without cntl_pt_rows?

[dem1]

I also see that the attribute processing_version changed from "R2022.1" to "R2022.0.1" which seems very suspect?

[OB SeaDAS - dshea]

Yes, the L2 file format has changed. Since we have not used sub-sampling of the the lat/lon arrays in over 20 years, we decided to simplify the L2 file. This change also gave other 3rd party programs the ability to geolocate the variables (ie. chlor_a) now that they have the same dimensions as lat/lon. We also added chlor_a:coordinates = "longitude latitude" to help follow the standard and help generic programs with geolocation.

cntl_pt_rows just counted the rows. The best way to know the edges of the bowtie is to use detnum. This variable actually tells you what detector the line was read from. For MODIS 1km bands this has a value of 0 through 9. For VIIRS 0 through 15

Hope that helps.

don

[PODAAC - edward.m.armstrong]

Hi Don,

Hijacking this thread a bit on behalf of the PODAAC. We received no notification of these changes and our software systems broke. Specifically the change from the dimension 'pixel_control_points' to 'pixels_per_line' in the latitude/longitude variables is the issue. Currently there are no Aqua SST L2P (https://doi.org/10.5067/GHMDA-2PJ19) products produced.

What is the schedule for implementing the change in the Terra L2 and VIIRS L2? As of today they both follow the older format.

regards,

Ed Armstrong

[dem1]

I see also a change concerning the unit of the par variable from:
"einstein m^-2 day^-1"
to
"mol photon m^-2 day^-1"
and the new unit is not recognized by the udunits lib and so not compliant with the CF convention (you can check with the udunits2 utility and cut/paste the unit).

Same issue with the ipar variable.

[OB SeaDAS - dshea]

Thanks for pointing the units out. We are deleting the "photons" to make them CF compliant.

don