I agree with John that flag_values should be numeric. That is the intent
expressed in section 3.5. The type specified in appendix A is a mistake
and should be corrected.
Brian
On Thu, Apr 19, 2007 at 11:36:59AM +0100, Stark, John wrote:
> Dear all,
>
> We have a NetCDF file that fails the CF compliance checker on the BADC
> website:
> http://titania.badc.rl.ac.uk/cgi-bin/cf-checker.pl
> The failure concerns the use of 'flag_values' to describe a flag
> variable called 'mask'.
>
> > Checking variable: mask
> > ------------------
> > ERROR: Attribute flag_values of incorrect type
> >
>
> Following this up we ran into a problem. On the list of attributes in
> CF-1.0 appendix A, flag_values is described as a string, but all the
> examples I've seen, including that in CF-1.0, section 3.5, use a list
> (vector) of valid numeric values (presumably matching the variable
> type).
>
> Part of the the ncdump for our file looks like this:
> byte mask(time, lat, lon) ;
> mask:long_name = "sea/land/lake/ice field composite mask" ;
> mask:_FillValue = -128b ;
> mask:flag_values = 1b, 2b, 4b, 8b ;
> mask:flag_meanings = "sea land lake ice" ;
>
> I think flag_values should be numeric rather than a string. The CF-1.0
> standard should be consistent either way...
>
> Thanks,
> John
>
> --
> John Stark SST and Sea Ice Scientist
> Met Office FitzRoy Road Exeter EX1 3PB United Kingdom
> Tel: +44 (0)1392 885059 Fax: +44 (0)1392 885681
> _______________________________________________
> CF-metadata mailing list
> CF-metadata at cgd.ucar.edu
> http://www.cgd.ucar.edu/mailman/listinfo/cf-metadata
Received on Thu Apr 19 2007 - 10:43:34 BST
