Rosalyn Hatcher wrote:
>
>>>> My grid mapping is Equidistant Cylindrical. SO in the nc I set as
>>>> GLOBAL ATTRIBUTE
>>>> ----------------------------------
>>>> attname=Conventions
>>>> value=CF-1.4
>>>>
>>>> attname=grid_mapping
>>>> value=Cylindrical
>>>> ------------------------------------
>>>> Some question:
>>>> 1) A collegue of mine told me that for CF 1.4 the grid_mapping is no
>>>> more
>>>> a global attribute but it is a variable attribute named as
>>>> "grid_mapping_name".
>>>>
>>>> BUT I run the CF checker for 1.4 on my file and everything is ok!!!!
>>>
>>> The CF checker is limited in understanding the semantics, it often only
>>> checks syntax.
>
> From my understanding of the Conventions there are 7 standard global
> attributes:
> 'Conventions','title','history','institution','source','reference','comment'
>
>
> However, there is nothing stopping non-standard global attributes being
> defined.
>
> "2.6: This standard describes many attributes (some mandatory, others
> optional), but
> a file may also contain non-standard attributes. Such attributes do not
> represent a
> violation of this standard."
>
> Please correct me if I'm wrong! As stated above there is only so much
> intelligence
> that can added to the checker, often it is checking semantics.
>
> I guess I could get the checker to compare any non-standard global
> attribute with
> the list of defined attributes in Appendix A and if it finds a match
> issue an informational
> statement.
>
> Regards,
> Rosalyn.
Hi Rosalyn:
This is a hard problem, essentially detecting that the user is intending to do something, but incorrectly (this is what i mean by "semantics"). There are 2 approaches that occur to me that the CF checker could do:
1. Look at all the data variables and display which ones have coordinate systems and can be correctly georeferenced. The user sees that their intention is not being fulfilled and digs deeper to find out why. This is the approach that the CDM Validator takes:
http://motherlode.ucar.edu:8080/cdmvalidator/validate.html
2. One notices common mistakes and issues useful warnings, in this case noticing that grid_mapping is a global variable. If you get a lot of the common mistakes covered, the warnings start to answer a significant portion of the easy problems.
If you decide to do more, i would probably recommend doing approach #2, since #1 is already being done. In that case, I would just slowly accumulate better diagnostics as examples come in.
John
Received on Wed Jul 21 2010 - 08:56:59 BST
