[MITgcm-support] "gluemnc grid.t*.nc" omits many variables

[Ryan Abernathey]

I would say that you definitely identified the problem--old version of  
nco.

I think that the ncecat -u option is needed for glueing grid files  
because they don't have a common record dimension by default. That's  
probably why the script worked for the state files but not the grid  
files. I think this was one of the hacks I had to do to get the grid  
files to glue.

You could either update your nco or try to rewrite the script to avoid  
using that -u flag.

Good luck,
Ryan

p.s. cc'd to the support list for posterity.



On Sep 27, 2010, at 11:20 AM, Maarten Buijsman wrote:

> Hi Ryan
>
> At GFDL we use
> (ncecat -r) "NCO netCDF Operators version "3.1.4" last modified  
> 2006/06/09 built Dec 11 2006 on hnv by hnv"
> which is old and the cause of my trouble.
> I am currently checking if they want/can update it..... (cannot mess  
> with it myself)
>
> -u does:
>
> "A new record dimension is the glue which binds the input file data  
> together. The new
> record dimension name is, by default, “record”. Its name can be  
> specified with the ‘-u
> ulm_nm’ short option (or the ‘--ulm_nm’ or ‘rcd_nm’ long options)."
>
> This functionality is not present in the olde ncecat....:
> ncecat [-4] [-A] [-C] [-c] [-D dbg_lvl] [-d ...] [-F] [-H] [-h] [-l  
> path] [-n ...] [-O] [-o out.nc] [-p path] [-R] [-r] [-t thr_nbr] [- 
> v ...] [-x] in.nc [...] [out.nc]
>
> -4, --4, --netcdf4 Output file in netCDF4 storage format (HDF5)
> -A, --apn, --append Append to existing output file, if any
> -c, --crd, --coords Coordinate variables will all be processed
> -C, --nocoords Associated coordinate variables should not be processed
> -D, --dbg_lvl, --debug-level dbg_lvl Debugging level
> -d, --dmn, --dimension dim,[min][,[max]] Dimension's limits in  
> hyperslab
> -F, --ftn, --fortran Fortran indexing conventions (1-based) for I/O
> -H, --fl_lst_in, --file_list Do not create "input_file_list" global  
> attribute
> -h, --hst, --history Do not append to "history" global attribute
> -l, --lcl, --local path Local storage path for remotely-retrieved  
> files
> -n, --nintap nbr_files,[nbr_numeric_chars[,increment]] NINTAP-style  
> abbreviation of file list
> -o, --output, --fl_out fl_out Output file name (or use last  
> positional argument)
> -O, --ovr, --overwrite Overwrite existing output file, if any
> -p, --pth, --path path Path prefix for all input filenames
> -R, --rtn, --retain Retain remotely-retrieved files after use
> -r, --revision, --vrs, --version Program version and copyright notice
> -t, --thr_nbr, --threads, --omp_num_threads thr_nbr Thread number  
> for OpenMP
> -v, --variable var1[,var2[...]] Variable(s) to process (regular  
> expressions supported)
> -x, --xcl, --exclude Extract all variables EXCEPT those specified  
> with -v
> in.nc Input file name(s)
> [out.nc] Output file name (or use -o switch)
>
> so I guess updating is the best ......
> although there may be a workaround?
>
> thanks
> Maarten
>
>
> Ryan Abernathey wrote:
>> Hi Maarten,
>>
>> On Sep 27, 2010, at 9:40 AM, Maarten Buijsman wrote:
>>
>>> ncecat: invalid option -- u
>>
>>
>> This error is telling you that the script is attempting to use an  
>> invalid option with the ncecat command. That is probably because  
>> you are using a different version of nco from the one I used when I  
>> modified the script. Just because it works with the state files  
>> doesn't rule out this possibility. (The grid files are actually  
>> more complicated because the variables have different sizes, rather  
>> than the state files, where the variables are all the same size.)
>>
>> I will log on to my server this afternoon and tell you which  
>> version of nco I used. In the meantime, check out the manual for  
>> ncecat and try to figure out how it works.
>>
>> Cheers,
>> Ryan
>>
>>
>
> -- 
> ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~
> Maarten Buisman (maarten.buijsman at noaa.gov)
> Geophysical Fluid Dynamics Laboratory
> 201 Forrestal Road
> Princeton, NJ 08540-6649
> Phone: (609) 452-5305
> Fax: (609) 987-5063
>