ECMWF ensemble grib files

Hi. I have ECMWF grib files (grib1) each containing ensemble data (51
members) for a variety of fields at a single forecast time.
The ensemble functionality introduced in v4.3.0 does not work for this
file: NCL sees only 1 value for each field.
 
An abbreviated version of the grib inventory is
> wgrib -v ECMWF_2009021200_0120.grb | grep "GH:500 mb"
1:0:D=2009021200:GH:500 mb:kpds=156,100,500:120hr fcst:ECMWF
stream?:"Height (geopotential) m
2:133930:D=2009021200:GH:500 mb:kpds=156,100,500:120hr fcst:ECMWF
stream?:"Height (geopotential) m
3:267860:D=2009021200:GH:500 mb:kpds=156,100,500:120hr fcst:ECMWF
stream?:"Height (geopotential) m
18:2365650:D=2009021200:GH:500 mb:kpds=156,100,500:120hr fcst:ECMWF
stream?:"Height (geopotential) m
28:3749370:D=2009021200:GH:500 mb:kpds=156,100,500:120hr fcst:ECMWF
stream?:"Height (geopotential) m
29:3883300:D=2009021200:GH:500 mb:kpds=156,100,500:120hr fcst:ECMWF
stream?:"Height (geopotential) m
...
325:45392220:D=2009021200:GH:500 mb:kpds=156,100,500:120hr fcst:ECMWF
stream?:"Height (geopotential) m
334:46642010:D=2009021200:GH:500 mb:kpds=156,100,500:120hr fcst:ECMWF
stream?:"Height (geopotential) m
347:48516360:D=2009021200:GH:500 mb:kpds=156,100,500:120hr fcst:ECMWF
stream?:"Height (geopotential) m
 
And corresponding abbrieviated output from NCL is
ncl 0> file1 = addfile("ECMWF_2009021200_0120.grb","r")
ncl 1> print(file1)
...
   variables:
...
      float GH_GDS4_ISBL ( g4_lat_0, g4_lon_1 )
         center : European Center for Medium-Range Weather
Forecasts - Reading
         long_name : Height
         units : m
         _FillValue : -999
         level_indicator : 100
         gds_grid_type : 4
         parameter_table_version : 128
         parameter_number : 156
         level : 500
         forecast_time : 120
         forecast_time_units : hours
         initial_time : 02/12/2009 (00:00)
...
 
 
I realise that this is likely an issue with the ECMWF file and not with
NCL, but any suggestions/insight would be appreciated.
 
[There is a workaround: use wgrib to split the file into 51 parts - one
for each member - and then read all 51 files into NCL. This works just
fine, but the processing time for all time steps adds up & it's a rather
cumbersome process!]
 
I'm running NCL vs 5.0.1 on a Sun Solaris 5.10
 
Thanks
Isla

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Thu Feb 19 2009 - 07:09:37 MST