Hi Jennifer, in this instance, how do you know its an ensemble
dimension, as opposed to say, a wavelength ? is it
ens:long_name = "ensemble member"
??
On 3/16/2010 12:36 PM, Jennifer Adams wrote:
> In the absence of a standard, I have also made some choices about how
> to handle ensemble metadata with GrADS and the GDS. Below is a
> shortened example of the ncdump output from a GFS ensemble data set
> behind GDS. I put the metadata that GrADS needs in customized
> attributes (e.g. grads_name, grads_dim, etc.) I steered clear of using
> data or coordinate variables of type 'char', because I feel attributes
> are the better place for this kind of metadata, and GrADS and the GDS
> have no interface for handing data variables that are not numeric. I
> have always assumed it would be necessary to add more metadata to make
> this data set CF compliant, but in the meanwhile, the ensemble axis is
> a plain old linear dimension. --Jennifer
> --
> Jennifer M. Adams
> IGES/COLA
> 4041 Powder Mill Road, Suite 302
> Calverton, MD 20705
> jma at cola.iges.org <mailto:jma at cola.iges.org>
>
>
>
> # ncdump -c http://monsoondata.org:9090/dods/gfsens/gfsens.2010031612
> netcdf gfsens {
> dimensions:
> ens = 22 ;
> lat = 181 ;
> lev = 7 ;
> lon = 360 ;
> time = 31 ;
> variables:
> double ens(ens) ;
> ens:grads_name =
> "avg,c00,p01,p02,p03,p04,p05,p06,p07,p08,p09,p10,p11,p12,p13,p14,p15,p16,p17,p18,p19,p20"
> ;
> ens:grads_length =
> "31,31,31,31,31,31,31,31,31,31,31,31,31,31,31,31,31,31,31,31,31,31" ;
> ens:grads_tinit =
> "1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1" ;
> ens:avg = "unweighted mean of all members" ;
> ens:c00 = "unperturbed control forecast" ;
> ens:p01 = "positively perturbed forecast" ;
> ens:p02 = "positively perturbed forecast" ;
> <additional attributes snipped>
> ens:p19= "positively perturbed forecast" ;
> ens:p20 = "positively perturbed forecast" ;
> ens:grads_dim = "e" ;
> ens:grads_size = "22" ;
> ens:grads_mapping = "null" ;
> ens:units = "count" ;
> ens:long_name = "ensemble member" ;
> double time(time) ;
> time:grads_dim = "t" ;
> time:grads_mapping = "linear" ;
> time:grads_size = "31" ;
> time:grads_min = "12z16mar2010" ;
> time:grads_step = "6hr" ;
> time:units = "days since 1-1-1 00:00:0.0" ;
> time:long_name = "time" ;
> time:minimum = "12z16mar2010" ;
> time:maximum = "00z24mar2010" ;
> time:resolution = 0.25f ;
> <other coordinate variables snipped>
> float t(ens, time, lev, lat, lon) ;
> t:_FillValue = -9.99e+33f ;
> t:missing_value = -9.99e+33f ;
> t:long_name = "temperature [k] " ;
> <other variables snipped>
>
> data:
>
> ens = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18,
> 19, 20,
> 21, 22 ;
>
> time = 733848.5, 733848.75, 733849, 733849.25, 733849.5, 733849.75,
> 733850,
> 733850.25, 733850.5, 733850.75, 733851, 733851.25, 733851.5,
> 733851.75,
> 733852, 733852.25, 733852.5, 733852.75, 733853, 733853.25, 733853.5,
> 733853.75, 733854, 733854.25, 733854.5, 733854.75, 733855, 733855.25,
> 733855.5, 733855.75, 733856 ;
>
> }
>
>
>
> On Mar 16, 2010, at 1:22 PM, Jonathan Gregory wrote:
>
>> Dear John
>>
>>> Im not sure if we ever converged on how to write ensemble files.
>> No, we didn't. If I remember correctly, we couldn't agree on the
>> relationship
>> between attributes and coordinates, and that was a sticking-point. I
>> thought
>> that we ought to allow multivalued auxiliary coord variables for ensemble
>> axes, corresponding to attributes that might be used to describe a single
>> model. I would give these standard_names to identify them. Paco
>> started the
>> discussion, and in the end he made his own decisions for what to do
>> for the
>> ENSEMBLES project (because CF had not concluded), so he might comment
>> on it.
>>
>>> Particularly, how does software recognize the ensemble dimension?
>> We didn't agree. But at the end of the discussion, Steve also
>> suggested that
>> it could be identified by an axis attribute, just as you have. I like
>> that
>> idea too. But it could also be given a standard name, like other
>> coordinate
>> variables e.g. ensemble_member_identifier. I suppose the values of this
>> coord var might be either numeric or strings, though it your example
>> it is
>> integer. I haven't looked at the trac ticket to see if we discussed these
>> points before.
>>
>>> Perhaps we should add :axis = "Ensemble" to the model coordinate
>>> variable?
>> I'd suggest lower case, like most CF keywords. The other axis
>> attributes are
>> all single letters (X Y Z T) but I don't think it would be a problem
>> to use
>> a word.
>>
>> Best wishes
>>
>> Jonathan
>> _______________________________________________
>> CF-metadata mailing list
>> CF-metadata at cgd.ucar.edu <mailto:CF-metadata at cgd.ucar.edu>
>> http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
>
>
>
>
> _______________________________________________
> CF-metadata mailing list
> CF-metadata at cgd.ucar.edu
> http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <
http://mailman.cgd.ucar.edu/pipermail/cf-metadata/attachments/20100316/4f9d8979/attachment-0002.html>
Received on Tue Mar 16 2010 - 15:30:26 GMT
