Re: grib=>nc time axis problem (fwd)

• To: bit-bucket <bit-bucket@xxxxxxxxxxxxxxxx>
• Subject: Re: grib=>nc time axis problem (fwd)
• From: Robb Kambic <rkambic@xxxxxxxxxxxxxxxx>
• Date: Fri, 19 Mar 2004 12:41:58 -0700 (MST)

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

---------- Forwarded message ----------
Date: Fri, 19 Mar 2004 12:51:41 -0500 (EST)
From: Lori Thompson <Lori.Thompson@xxxxxxxx>
To: Robb Kambic <rkambic@xxxxxxxxxxxxxxxx>
Subject: Re: grib=>nc time axis problem

Thanks, Rob. I added the time series to the final concatenated file and
the took care of the additional record frames.

On Fri, 19 Mar 2004, Robb Kambic wrote:

On Thu, 18 Mar 2004, Lori Thompson wrote:

> Hi Robb. I wrote to you a couple weeks ago regarding a conversion from
> grib to nc data. During the conversion using gribtocdl and gribtonc, I
> received an additional record dimension which contained blank data. As a
> solution, after gribtocdl, I edit the cdl file and change valtime_offset
> in the dimensions section equal to 1 rather than 2 and in the data
> section, change valtime_offset from 0, model_time to model_time. This

Lori,

My solution to this is in this msg:

http://www.unidata.ucar.edu/cgi-bin/msgout?/glimpse/decoders/627

> resulted in 1 record frame for the time series. However, the concatenated
> time series file does not contain a time axis. I tried to edit the cdl
> file by replacing record with time, add the time variable with the
> metadata, and add the time under the data section equal to the model time.

I would edit the cdl by setting the dimension "valtime_offset" to the
number of hours in the time series. In the data section set the variable
"valtime_offset" to the actual hours in the time series. ie

valtime_offset = 84, 96, 108, 132, 144, 120, 192, 240 ;

This is how the ldm pqact is set up to decode the model files resulting
with one record per hour. Hopefully this solves your problem.

Robb..

> However, this is producing 2 frames for each file again when it is only
> supposed to be one. I tried to remove the valtime and reftime variables
> and am getting a bus error. There seems to be more time variables with
> this newest version of the decoders and I'm not sure which ones are to be
> retained. Instead of editing the cdl file, can I add a time axis and
> metadata to the singe time series resultant file, since this one produces
> the correct amount of frames?
>
> Thanks in advance for your help.
> Lori
>
> --
> ----------------------------------------------------
> Lori A. Thompson
> Applications Programmer/RS Information Systems, Inc.
> Geophysical Fluid Dynamics Laboratory/NOAA
> Princeton University Forrestal Campus/Route 1
> PO Box 308
> Princeton, NJ 08542
> Phone: 609-452-6563
> Fax:   609-987-5063
> http://www.gfdl.noaa.gov/~lat
>

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

--
----------------------------------------------------
Lori A. Thompson
Applications Programmer/RS Information Systems, Inc.
Geophysical Fluid Dynamics Laboratory/NOAA
Princeton University Forrestal Campus/Route 1
PO Box 308
Princeton, NJ 08542
Phone: 609-452-6563
Fax:   609-987-5063
http://www.gfdl.noaa.gov/~lat