Re: gribtonc question

• To: "Mathias D. Müller" <Mathias.Mueller@xxxxxxxxx>
• Subject: Re: gribtonc question
• From: Robb Kambic <rkambic@xxxxxxxxxxxxxxxx>
• Date: Mon, 19 May 2003 10:55:15 -0600 (MDT)

On Sat, 17 May 2003, [iso-8859-1] "Mathias D. Müller" wrote:

it works now!!!!!!!!!

Mathias,

Great

i had to compile udunits and netcdf library.

i will have to retune the cdl file or create a tab-file (is there a utility
for the tab file?)

Noop, has to be created by hand. But there is a utility to make the cdl
called gribtocdl.  There's man pages.

data in the netcdf looks good, but some 3d fields are now stored twice in
the same variable by adding a 4 th dimension

e.g. [x,y,z,2] where [x,y,z,0] is the data which is perfect and [x,y,z,1]
is all -9999
anyhow it just wastes a lot of space, but the data is fine.

I think it's the way you created the cdl. gribtonc creates the space
before hand using the supplied cdl. So if you have [x,y,z,2], then
NetCDF will create [x,y,z,0] and [x,y,z,1]. Since [x,y,z,1] is all missing
data values maybe your dimensions need rearranged, ie [x,y,z,1].  Since I
don't know your  data format my suggestion could be wrong. The main point is
the space for the NetCDF file is allocated all at once for each new
unlimited dimension variable (usually the record variable)

the 4 mb grib file is now a 80 mb netcdf, but i can read it much faster
with the idl programming language and it only takes about a second to
create the nc-file. that is awesome.
do you know how i could read the grib file directly with idl as efficient
as your decoder works?

noop, grid is mainly a packing/transportation format.

Robb...

i used to spawn wgrib, dump the grib into a binary file and than read this
binary with idl. but that takes about 2 minutes.
(time is very important, since i do realtime forecasting and have to
process several gigabytes of data every day)



many thanks
mathias

ps: here is a link to the weather forecast results:
http://www.unibas.ch/geo/mcr/3d/meteo/

At 10:50 16.05.2003 -0600, you wrote:
>On Fri, 16 May 2003, Mathias D. Mueller wrote:
>
> > thanks
> >
> > i already had compiled with gcc. i did it by setting the following into the
> > CUSTOMIZE file
> >
> >  CPP_LDM=-I/usr/local/ldm/include
> >  LD_LDM='-L/usr/local/ldm/lib -lldm'
> > #
> > # netCDF header file and library references:
> >  CPP_NETCDF=-I/usr/local/netcdf-3.5.0/include
> >  LD_NETCDF='-L/usr/local/netcdf-3.5.0/lib -lnetcdf'
> > #
> > # udunits header file and library references:
> >  CPP_UDUNITS=-I/usr/local/udunits-1.11.3/include
> >  LD_UDUNITS='-L/usr/local/udunits-1.11.3/lib -ludunits'
> > #
> > # Type of C compiler:
> >  CC='gcc'
> > #
> > # C compiler flags:
> >  CFLAGS=-g
> > #
> > # Installation prefix:
> >  prefix=/usr/local/unidata
> >
> >
> > the UDUNITS and netcdf library i didn't compile but copied a binary
> version from
> > the internet to my usr/local directory, since it seemed to be easier than
> > compiling:-)
>
>Mathias,
>That's probably the problem.
>
> >
> > i probaly have to compile those:
> > which source codes should i use ?
> >
>
>You can get the source from the web page at:
>
>http://my.unidata.ucar.edu/content/software/index.html
>
>
> > are there any special flags needed?
> >
>Noop
>
> > my system: RED HAT 8.0 with
> > Linux ws-eta 2.4.18-19.8.0 #1 Thu Dec 12 05:39:29 EST 2002 i686 i686
> i386 GNU/Linux
>
>I tested it on linux 8.0 last week
>
>Robb...
> >
> >
> > many thanks and best regards
> > mathias
> >
> >
> >
> >
> >
> >
> > Quoting Robb Kambic <rkambic@xxxxxxxxxxxxxxxx>:
> >
> > > On Fri, 16 May 2003, Mathias D. Mueller wrote:
> > >
> > > > hi
> > > > i am a phd student at the university of basel.
> > > > i try to decode grib files from my eta model into netcdf.
> > > >
> > > > i installed the new decoders.
> > > > i created a cdl-file, which seems to work.
> > > > gribdump on the girb file works also and i see all the data in it.
> > > >
> > > > however if i call the
> > > > gribtonc using the cdl-file i get only the message:
> > > >
> > > > May 16 12:40:20 gribtonc[31493]: Starting Up
> > > > segmentation
> >
> >
> >  fault
> > > >
> > > > the nc-file written only probalby only contains the stuff from the
> > > cdl-file
> > > >
> > > >
> > > > do you have any ideas what might be wrong
> > >
> > > Hiya Mathias,
> > >
> > > Another user had a similar problem, it turned out to be the compiler
> > > flags. For linux type systems try:
> > >
> > > % setenv CC gcc
> > >
> > > For Solaris
> > >
> > > % setenv CC cc  or gcc
> > >
> > > If that doesn't solve the problem, send me the machine type and the
> > > version of the OS, ie
> > >
> > > % uname -a
> > >
> > > If that is the problem, NetCDF and udunits need to be recompiled before
> > > recompiling the decoders.
> > >
> > > Robb...
> > >
> > > >
> > > > best regards and many thanks
> > > > mathias
> > > > --------------------------------------------------------
> > > >
> > > > Mathias D. Müller
> > > >
> > > > University of Basel
> > > > Institute of Meteorology, Climatology and Remote Sensing
> > > >
> > > > Klingelbergstr. 27
> > > > 4056 Basel
> > > > Tel: (+41 61) 267 07 56
> > > > my page: http://www.unibas.ch/geo/mcr/3d
> > > >
> > > > -------------------------------------------------
> > > > This mail sent through IMP: http://horde.org/imp/
> > > >
> > > >

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