NCAR uses a modified version of Paco's file structure for TIGGE output.
All ensemble members found in the file require the same "single" model
initialization time, the same forecast timesteps, and
identical grids types.
A sample file can be found at:
http://dss.ucar.edu/download/tigge/ncep_20100223000000_tp_lev_255_id_528.1.5deg.nc
Here's a snippet from the file header metadata -Doug:
dimensions:
longitude = 240 ;
latitude = 121 ;
string4 = 4 ;
string2 = 2 ;
time = 4 ;
realization = 21 ;
variables:
float longitude(longitude) ;
longitude:data_type = "float" ;
longitude:units = "degrees_east" ;
longitude:axis = "X" ;
longitude:standard_name = "longitude" ;
longitude:valid_min = 0.f ;
longitude:valid_max = 358.5f ;
float latitude(latitude) ;
latitude:data_type = "float" ;
latitude:units = "degrees_north" ;
latitude:axis = "Y" ;
latitude:standard_name = "latitude" ;
latitude:valid_min = -90.f ;
latitude:valid_max = 90.f ;
int time(time) ;
time:data_type = "int" ;
time:units = "hours since 1950-01-01 00:00:00" ;
time:standard_name = "time" ;
time:long_name = "Forecast Valid Time" ;
int realization(realization) ;
realization:data_type = "int" ;
realization:units = "1" ;
realization:standard_name = "realization" ;
realization:long_name = "Number of the simulation in the
ensemble" ;
char forecast_type(realization, string2) ;
forecast_type:data_type = "char" ;
forecast_type:long_name = "Forecast type" ;
char institution(realization, string4) ;
institution:data_type = "char" ;
institution:standard_name = "institution" ;
institution:long_name = "Institution responsible for the forecast
system" ;
char source(realization, string4) ;
source:data_type = "char" ;
source:standard_name = "source" ;
source:long_name = "Method of production of the data" ;
char experiment_id(realization, string4) ;
experiment_id:data_type = "char" ;
experiment_id:standard_name = "experiment_id" ;
experiment_id:long_name = "Experiment identifier" ;
int perturbation_method(realization) ;
perturbation_method:data_type = "int" ;
perturbation_method:long_name = "Perturbation method, GRIB-2 Code
table 4.6" ;
float tp(time, realization, latitude, longitude) ;
tp:data_type = "float" ;
tp:units = "kg m-2 s-1" ;
tp:standard_name = "precipitation_amount" ;
tp:level_name = "Ground or water surface " ;
tp:level_units = "unknown" ;
tp:level_value = 0.f ;
tp:coordinates = "institution forecast_type source experiment_id
perturbation_method" ;
tp:_FillValue = 9.96921e+36f ;
// global attributes:
:Model_Initialization_Date_and_Time_UTC = "2010-2-23 0 UTC" ;
:Conventions = "CF-1.0" ;
:Institution = "NCAR" ;
:Generator = "grib2_to_netcdf" ;
:Created = "Wed Mar 03 04:37:58 2010" ;
:Title = "TIGGE" ;
:References = "For TIGGE, check http://tigge.ucar.edu"; ;
On Mar 16, 2010, at 1:31 PM, V. Balaji wrote:
Jonathan Gregory writes:
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.
http://ensembles.ecmwf.int/thredds/ensembles/stream2/seasonal/atmospheric/catalog.html
has examples of how Paco has it set up.
(esp see the "mother of all aggregations" :-).
It requires the ensemble to share the same space and time
discretization.
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
[email protected]
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
--
V. Balaji Office: +1-609-452-6516
Head, Modeling Systems Group, GFDL Home: +1-212-253-6662
Princeton University Email: [email protected]
_______________________________________________
CF-metadata mailing list
[email protected]
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
_______________________________________________
CF-metadata mailing list
[email protected]
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
