Re: Descriptor Files

[Jaison Kurian <jaison@xxxxxxxxxxxxxxxxxx>]

Hi Kasper,
             As Ansley pointed out "load" command will make the 
time-independent variable (from the first descriptor file) available
even after opening a new descriptor file (this is the easiest method). 
Few more work arounds...

  1. Use the time-independent variable from the first descriptor file
           in some "action commands" (like list/plot/fill) before opening
           the second descriptor file. In ferret..........

      set data first.des
           
      let t_indp_vars = TSEQUENCE({"lon","lat"})  ! time independent vars
      let t_indp      = `t_indp_vars,return=lend` ! num of time indep. vars
      
      REPEAT/l=1:`t_indp` ( let var = `t_indp_vars` ;\
          list/nohead var[i=@sum,j=@sum,k=@sum] )        

      set data second.des

    If "t_indp_vars" is supplied with all time-independent variables then
       above method will take care of all of them.

   2. Time-independent variables will be "same" for all files in a 
         particular experiment (say for 5km run). Then you can save all
         these time-independent variables from a *.des file to a new netcdf 
         file in Ferret using SAVE command  : 
         
         set data exp_5km.des ! or any individual file in the series
         SAVE/file=time_indep_var_5km.nc/append lon 
         SAVE/file=time_indep_var_5km.nc/append lat 
         cancel data/all
         set data exp_15km.des
         SAVE/file=time_indep_var_15km.nc/append lon
         SAVE/file=time_indep_var_15km.nc/append lat

         ..now you can open these new files for the time-independent 
         variables of the corresponding *.des file.


 With Regards 

 Jaison 



On Wed, 15 Sep 2004, Ansley Manke wrote:

> Hi Kasper,
> This seems to be a bug in how Ferret handles time-independent variables in
> descriptor files.  We'll look into fixing it, but in the meantime I 
> think I've found
> a way to work around it.  Load the non-time-dependent data from the first
> set before the "set data" command for the second set.
> 
>    set data  model_1.des
>    load mask[d=1]
> 
>    set data model_2.des
>    list mask[d=2]
>    list mask[d=1]
> 
> In the example I cooked up, the two mask variables are correct when loaded
> in this way.
> 
> Ansley Manke
> 
> Gian-Kasper Plattner wrote:
> 
> > Hi all,
> >
> > I have problems using descriptor files in ferret. I would like to 
> > compare model output from two different versions of our ocean model, 
> > which are run on different grids (e.g. 15km and 5km horizontal 
> > resolution), in order to analyse the resolution-dependence of the 
> > results. The netcdf files for the two model versions have the exactly 
> > same output variables (names etc), some of which are time-independent 
> > (masks, lon, lat etc.) and most are time-dependent. To calculate 
> > climatological averages over several years, I have created separate 
> > descriptor files for both model versions. That works fine and I can 
> > load the files with "set data *.des" and use them in ferret as dataset 
> > 1 and 2. However, it seems as if all the variables that are 
> > time-independent are overwritten each time I read a new descriptor 
> > file into ferret. Therefore when I read the 15km file first [d=1] and 
> > then the 5km file [d=2], the values of all the time-independent 
> > variables in dataset 1are set to the values from dataset 2!
> >
> > Did anyone encounter similar problems while using descriptor files? Do 
> > I have to adapt my descriptor files?
> >
> > Thanks for your help,
> >
> > Best regards, Kasper
> >
> >------------------------------------------------------------------------
> >
> >
> >&FORMAT_RECORD
> >D_TYPE        = '  MC',
> >D_FORMAT      = '  1A',
> >D_SOURCE_CLASS       = 'MODEL OUTPUT',
> >/
> >&BACKGROUND_RECORD
> >D_EXPNUM      = '0001',
> >D_MODNUM      = '  AA',
> >D_TITLE       = '2 year simulation with 15 km grid',
> >D_T0TIME      = '01-JAN-2000:12:00:00',
> >D_TIME_UNIT   = 1.0,
> >D_TIME_MODULO = .FALSE.,
> >D_CALTYPE     = '360_DAY',
> >/
> >&MESSAGE_RECORD
> >D_MESSAGE     = ' ',
> >D_ALERT_ON_OPEN      = F,
> >D_ALERT_ON_OUTPUT    = F,
> >/
> >&EXTRA_RECORD
> >/
> >&STEPFILE_RECORD
> >s_filename    = 'usw15_5_flux_euph_avg_Y1.nc.AVG_EUPH_LAYER',
> >S_AUX_SET_NUM = 0,
> >S_START       = 3.117324e+08,
> >S_END  =        3.415404e+08,
> >S_DELTA       = 1296000.0,
> >S_REGVARFLAG  = ' ',
> >/
> >&STEPFILE_RECORD
> >S_FILENAME    = 'usw15_5_flux_euph_avg_Y2.nc.AVG_EUPH_LAYER',
> >S_AUX_SET_NUM = 0,
> >S_START       = 3.428364e+08,
> >S_END         = 3.726444e+08,
> >S_DELTA       = 1296000.0,
> >S_REGVARFLAG  = ' ',
> >/
> >&STEPFILE_RECORD
> >      s_filename    = '**END OF STEPFILES**'
> >/
> >  
> >
> 

-- 
___________________________________________________

    Jaison Kurian                           
    Centre for Atmospheric and Oceanic Sciences
    Indian Institute of Science
    B A N G A L O R E   560 012
    Ph: +91-80-3942505
        +91-80-3600450
    Fax:+91-80-3600865
___________________________________________________