Hi Jonathan, Brian:
A time-dependent lat,lon coordinate is a good idea, though probably a lot of clients wont understand it (my wont right at the moment). I'll think about how to represent "time varying grid_mapping information". I suppose if we allowed a WKT representation of the projection, we could have that be time dependent, eg char projectionWKT(runtime, strlen).
Thanks for the advice
Brian Eaton wrote:
> Hi John,
>
> I agree with Jonathan, and would like to stress that a conforming CF file
> must have the 2D auxiliary lat/lon coordinates; the grid_mapping
> information is optional. Jonathan's suggestion of adding a time coordinate
> to the lat/lon coordinates is the natural way to concatenate these time
> slices.
>
> There is no provision in CF-1.0 for time varying grid_mapping information.
> That's a good use case to keep in mind moving to CF-2. Thanks for bringing
> it up.
>
> Brian
>
>
> On Mon, Oct 09, 2006 at 07:28:22AM +0100, Jonathan Gregory wrote:
>
>>Dear John
>>
>>
>>>The projection is changing its parameters
>>>from one run to the next
>>
>>The data variables should have 2D auxiliary lat and lon coordinates, pointed to
>>by the coordinates attribute. Can you use these for plotting, slicing, etc.? If
>>you want to put the data into one variable with a time dimension, for example,
>>you could give the auxiliary lat and lon coordinates that dimension as well
>>e.g. lon(time,y,x).
>>
>>However if you want to do this and still retain the grid_mapping information,
>>we would need a way to represent a variable grid-mapping, by e.g. adding a time
>>dimension to its attributes, or making them point to 1D variables.
>>
>>Cheers
>>
>>Jonathan
>>_______________________________________________
>>CF-metadata mailing list
>>CF-metadata at cgd.ucar.edu
>>http://www.cgd.ucar.edu/mailman/listinfo/cf-metadata
>
> _______________________________________________
> CF-metadata mailing list
> CF-metadata at cgd.ucar.edu
> http://www.cgd.ucar.edu/mailman/listinfo/cf-metadata
Received on Mon Oct 09 2006 - 18:19:36 BST
