Dear Etienne I think the structure you have adopted for the data is fine.
> double pft(pft) ; > pft:long_name = "plant functional type" ; > pft:units = "none" ; > > double npp(time, pft, latitude, longitude) ; > npp:long_name = "npp of carbon for each pft" ; > npp:units = "kg m-2 year-1" ; The specific problem you raise is concerned with the axis attribute. That attribute is really intended for identifying spatiotemporal coordinates; although it may be convenient, it is redundant because they can also be identified in other ways. It has not been extended for non-spatiotemporal axes like pft. In your CDL, the pft axis is identified by its long_name. To make this more reliable, you might want to use a standard_name for this pft coordinate variable. There isn't such a standard_name at present, but area_type is often vegetation type in practice, so you could perhaps use that. We could standardise new area_types by proposals to this email list. Also, there is the new proposal, which I expect will go into the standard_name table, for UN/FAO land cover types, which in many cases are also vegetation types. See http://mailman.cgd.ucar.edu/pipermail/cf-metadata/2012/033507.html The quantities identified by these standard_names are string-valued, whereas you expect a numeric pft. However, a string-valued one could be encoded as a number by using the flag_values and flag_meanings attributes. There is an existing standard name for NPP as well viz net_primary_productivity_of_carbon (kg m-2 s-1) Best wishes Jonathan _______________________________________________ CF-metadata mailing list [email protected] http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
