[CF-metadata] Getting back to ensembles

From: John Caron <caron>
Date: Mon, 20 Nov 2006 12:49:56 -0700

Hi Brian:

Brian Eaton wrote:
> Hi John,
>
> On Mon, Nov 20, 2006 at 11:12:00AM -0700, John Caron wrote:
>
>>let me munge your examples slightly:
>>
>>1)
>>float temperature(rdim,time,lat,lon):
>> temperature:coordinates = 'realization time lat lon' ;
>>
>>char realization (rdim, strlen):
>>
>>assuming that time,lat,lon are the usual CF conventions, then the new thing
>>is that I have a string valued coordinate, but i dont know what it is
>>intended to mean. But if I unambiguously indicate it as:
>
>
> The "new" thing isn't new at all. This is a label coordinate as described
> in CF-1.0 section 6.

yes, not new to CF, but new to the CDM model of Grids.

>
>
>>char realization (rdim, strlen):
>> realization :standard_name="realization"; // or whatever we decide
>>
>>Now I know its an ensemble dimension, and can present it to the user for
>>slicing and dicing, in a way that the user knows for sure what the
>>dimension is used for. I would require that the strings are unique, so
>>there is a 1-1 mapping. What the string values are doesnt matter to me.
>
>
> CF places no restriction on the uniqueness of labels in a label
> coordinate. Typically one would "slice" a label coordinate by asking for
> the subset of elements whose labels matched the string "xxx". It's not
> clear to me why you want to require a 1-1 mapping (I'm assuming the mapping
> you refer to is between the labels and the array indices).

I think there's 2 points i'm making:

1. I need to understand any "extra" dimensions that are not part of the space-time coordinate system. In this case, understanding that the dimension represents an ensemble of model runs.

2. I would prefer to have a 1-1 coordinate variable for that dimension, so I can let the user slice it by name in addition to index value. In this case the "label coordinate" is acting like a coordinate variable.



>
> Brian
Received on Mon Nov 20 2006 - 12:49:56 GMT