RE: [ferret_users] saving from a descriptor file

[Jaison Kurian <jaison@xxxxxxxxxxxxxxxxxx>]

Hi Jaclyn,
           In the repeat command, you have to use "grave accents" like 
/L=`L` (not single quotes "'" but "`"). Then the repeat stuff will 
work without any error.

   Please note that the regridding to a new time axis does not mean 
that you got the data correctly. If there is an error from in the 
descriptor file time axis, you will carry over that error in all the
write-outs from Ferret. So check whether the descriptor file time axis
is correct. You can do this by making a plot or listing the time axis
values as

    set data mydata.des
    list/l=1:122 t[gt=myvariable]

  Without knowing the exact error messages that you received, it is
difficult to deduce what is happening with your data. If it is the
memory problem, you can increase the memory within Ferret session as
 
    set memory/size=500mb

 or starting ferret with more memory cache as

    [user@machine]$ ferret -memsize 500mb

Hope this helps,

Jaison 


On Mon, 6 Nov 2006, Jaclyn Brown wrote:

> Thanks Jaison,
> I have to save the data one time step at a time because it is so large. So
> doing save/file=u_1949.nc/append uvel won't work.
> Also the L='L' returned an error.
> 
> I tried defining the time axis again (define axis/t=..), and assigning it to
> the variable (u=uvel[gt=taxis@asn]) even though this seems to be redundant.
> Nevertheless it did the trick. No idea why. I think there must be some
> problem with my descriptor file that I can't see. 
> 
> I made the descriptor with make_des.pl one_file.nc>collated_files.des and
> then added the extra files by typing in extra stepfile_records. I did this
> because the descriptor file wouldn't join the files together automatically.
> I guess this is where some error has been introduced.
> 
> Cheers,
> Jaci 
> 
> -----Original Message-----
> From: Jaison Kurian [mailto:jaison@caos.iisc.ernet.in] 
> Sent: Monday, November 06, 2006 1:22 PM
> To: Jaclyn Brown
> Cc: ferret_users@noaa.gov
> Subject: Re: [ferret_users] saving from a descriptor file
> 
> Hi Jaclyn,
>            If you want to write the data for l=1:122, there is
> no need to open the output file with /LLIMITS qualifier. Just
> go like this :
> 
>   use "/kuroshio/jb842/POP/ful.....
>   set reg/l=1:122
>   save/fie=u_1949.nc uvel
> 
> That is all ! 
> 
> In your method, The following line is fine
> 
>    save/file=u_1949.nc/clobber/llimits=1:122/l=1 uvel
> 
> I think the next line should be
> 
>    repeat/l=2:122 (save/file=u_1949.nc/L=`L`/append uvel)
> 
> Because, each time you have to specify the L index or indices
> that you are appending to the file opened with /LLIMITS. This
> method is used only if there is a basic difference between the 
> time axis of source data and that of the final output one or 
> say you want to do some processing only for a give period. In
> normal case "set reg" or specifying region as 
> "save/L=1:122/file=somefile.nc/append uvel" will work just fine.
> 
> Hope this helps,
> 
> Jaison 
> 
> On Mon, 6 Nov 2006, Jaclyn Brown wrote:
> 
> > 
> > Hi,
> > I am trying to save a subset of a descriptor file I created. When I open
> the
> > descriptor file in ferret the time axis is as it should be. When I save
> some
> > of it (see below) the new file does not have the correct time axis. Every
> > point has the same time 2-Jan-1949, which is the first time step. 
> > Any ideas where I am going wrong?
> > Cheers,
> > Jaci
> > 
> > 
> > use "/kuroshio/jb842/POP/full_output_July2006/Pacific1949.des" !reads in
> > datafiles
> > 
> > save/file=u_1949.nc/clobber/llimits=1:122/l=1 uvel
> > repeat/l=2:122 (save/file=u_1949.nc/append uvel)
> > 
> > 
> > 
> 
>