[Go-essp-tech] CMOR and cell_measures issues

[Karl Taylor]

  Dear CMOR users,

It has been brought to our attention that the CF checker (which is 
different from the CMOR checker) traps an error in CMOR-produced files 
indicating non-compliance with the CF conventions.  The error is that 
the cell_measures attribute is defined, but the area and/or volume data 
is not contained in the file.  [We have included a non-CF attribute 
called "associated_files" that points to the files containing 
area/volume.]  Still, in order to not confuse CF-compliant software we 
plan to immediately release a new version of CMOR in which the 
"cell_measures" attribute is replaced by a non-CF attribute named 
"ext_cell_measures" (indicating that the area/volume data is externally 
available).

Also, some groups have reported that not all their variables are carried 
on the same grid.  (For example, sometimes the u and v fields are 
staggered relative to T.)  In this case it would be necessary to have 
more than one set of areas associated with the model, and this is not 
currently possible since only areacella, areacello, and volcello are 
defined in the CMIP5 request for model output.  We have therefore 
decided to include an option for CMOR users to suppress writing of the 
ext_cell_measures attribute when it would point to areacella or 
areacello incorrectly.  The user would in effect be allowed one set of 
areas to be associated with the atmospheric model and one set of 
areas/volumes to be associated with the ocean models.  The user would 
suppress the ext_cell_measures attribute (and areacella, areacello, 
volcello would not appear in the "associated_files" attribute) for any 
variables where these areas/volumes would be incorrect.

We hope this is acceptable.

Best regards,
Karl and Charles


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/go-essp-tech/attachments/20101019/afbbb58b/attachment.html