[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: 20020410: gribtonc CDLs for AVN and could CDL for LAPS work



On Mon, 17 Jun 2002, Tiziana Cherubini wrote:

> 
> On Mon, 17 Jun 2002, Robb Kambic wrote:
> 
> > Tiziana,
> >
> > I made some changes to the gribtonc source, now it appears that the
> > variables are being written.  It kind of hard to tell because some of the
> > variables values are missing and or have very little data.
> 
> Hi Robb,
> 
> thanks very much for the new decoders package. I tried it
> but, I'm sorry to say, that, even if in the log file the
> variable list is now complete, therefore all variables are
> being written, still some fields (i.e. P_lctl or crain and
> many others) are filled with missing values whereas they
> are present in the original grib file.
> 
Tiziana,

As I stated yesterday, many of the initial values are missing. If you look
at the whole set of values returned the data is there.  I checked it
against the data file you suggested the other day.  To look at a
particular variable, use the -v flag on ncdump. ie.

% ncdump -v P_hctl avn.nc | more

> > I did get the above ftp file to use as a test case.
> 
> Have you been able to produce a netcdf file with all the
> fields filled with the proper values?
> 
> There is also something about the size of the file produced
> by the gribtonc that is not really clear to me.
> The original grib files, say:
> 
> -rw-r--r--    1 tiziana  users    22830364 Jun 14 08:58 gblav.t12z.pgrbf00
> -rw-r--r--    1 tiziana  users    25809992 Jun 14 09:00 gblav.t12z.pgrbf06
> 
> where the first one is basicly the analysis file at 12:00 UTC
> and the second one is the t+6h forecast based on the 12:00 UTC
> analysis
> 
> are almost 3M far apart in size. After running the gribtonc
> I obtain:
> 
> -rw-r--r--    1 tiziana  users    116794932 Jun 17 14:05 gblav.t12z.f00.nc
> -rw-r--r--    1 tiziana  users    233561716 Jun 17 13:39 gblav.t12z.f06.nc
> 
> the second one is now almost the double in size than the
> first one. Would you expect it?
> 
That's because the NetCDF file preallocates all the space it needs.  Since
the second file has 2 forecasts, it's twice the size.  Even if there is
only one variable that has 2 records.  This is a con of NeCDF but because
the space is preallocated, it makes for quick access of the data.  There
still seems to be a problem with gribdump though, working on it.

Robb...






> Thanks a lot for your help,
> 
>                               Tiziana.
> 
> > I made new tar.Z file that's located in the ftp dir
> > pub/decoders  it's called  decoders-2.3.9.tar.Z
> >
> > Let me know if this solves your problems.
> >
> > Robb....
> >
> >
> > > although they are listed in the CDL you provided to me
> > > (see mail below). I can easyly convert grib analysis
> > > files but I'm still having problems with the forecast ones.
> > >
> > > Thanks a lot,
> > >
> > >           Tiziana.
> > >
> -------------------------------------
> Tiziana Cherubini
> Department of Meteorology
> University of Hawaii
> 2525 Correa Rd. HIG 367
> Honolulu, Hawaii 96822
> (808) 956-4593, (808) 956-2877 fax
> 
> 

===============================================================================
Robb Kambic                                Unidata Program Center
Software Engineer III                      Univ. Corp for Atmospheric Research
address@hidden             WWW: http://www.unidata.ucar.edu/
===============================================================================