Re: [thredds] aggregating GFS data, problem with accumulated and averaged variables

• To: thredds@xxxxxxxxxxxxxxxx
• Subject: Re: [thredds] aggregating GFS data, problem with accumulated and averaged variables
• From: John Caron <caron@xxxxxxxxxxxxxxxx>
• Date: Wed, 05 Jun 2013 08:53:54 -0600

Hi Hien:

can you make the files available do we can reproduce? thanks

john

On 6/5/2013 6:14 AM, Hein Zelle wrote:

Dear Thredds users,

we run thredds version 4.3.16 and use it to deal with GFS forecasts
gracefully from ferret and other software.  We use an ncml file to
catenate all time steps (one grib2 file per step) into a single
dataset.

Most variables work fine, but there is a problem with the averaged and
accumulated variables such as precipitation.  These switch between
3-hourly and 6-hourly accumulations, as can be seen by opening the
individual grib files with thredds.

+03, +09, +15, etc:
Total_precipitation_surface_3_Hour_Accumulation

+06, +12, +18, etc:
  Total_precipitation_surface_6_Hour_Accumulation

When opening the aggregated dataset, I only get the
Total_precipitation_surface_3_Hour_Accumulation variable, however it
only has 1 time step (+03).

It appears that thredds or the underlying netcdf/grib library doesn't
deal with the aggregated and averaged variables properly anymore.
This worked (more or less) with thredds 4.1.4, where the accumulation
interval wasn't appended to the variable name and thredds just
combined them.

Is there anything we can do or try to make this work?  I don't mind
having to re-compute 3 hourly accumulation values by hand, but at the
moment we can't access any data past +03 hours into the forecast.


Thanks for your advice!
Kind regards,
      Hein Zelle


The NCML file looks like this (but much longer until +192):


<netcdf xmlns="http://www.unidata.ucar.edu/namespaces/netcdf/ncml-2.2";>
   <aggregation dimName="time" type="joinExisting">
     <netcdf location="/ncep/gfs/0.5deg/gfs.2013060500/gfs.t00z.pgrb2f00" 
coordValue="0"/>
     <netcdf location="/ncep/gfs/0.5deg/gfs.2013060500/gfs.t00z.pgrb2f03" 
coordValue="3"/>
     <netcdf location="/ncep/gfs/0.5deg/gfs.2013060500/gfs.t00z.pgrb2f06" 
coordValue="6"/>
     <netcdf location="/ncep/gfs/0.5deg/gfs.2013060500/gfs.t00z.pgrb2f09" 
coordValue="9"/>
     <netcdf location="/ncep/gfs/0.5deg/gfs.2013060500/gfs.t00z.pgrb2f12" 
coordValue="12"/>
   </aggregation>
</netcdf>

• References:
  • [thredds] aggregating GFS data, problem with accumulated and averaged variables
    • From: Hein Zelle