[MITgcm-support] MITgcm-support Digest, Vol 170, Issue 4

Jean-Michel Campin jmc at mit.edu
Sun Aug 6 11:21:36 EDT 2017


Hi Jody and Rahael,  

two remarks:
1) flt_example is probably not the best starting point/template 
  to learn how to use pkg/diagnostics (with or without pkg/mnc):
 It does not use (even compile) pkg diagnostics (one of the reason
 is that pkg/flt has special type of output that does not fit into
 the general case and therefore has no pkg/diagnostics neither pkg/mnc
 interface).
 A better starting point would be a tutorial, e.g.,
   tutorial_advection_in_gyre

2) to Jody:
 Regarding tested forward  experiments (primary and secondary):
  59 experiments are using pkg Diagnostics
  22 experiments are using pkg MNC
 and 19 of theses are using both pkgs
Note that I did not find any experiment that use both packages and 
would not use NetCDF format for pkg/diagnostics output.
In other words, all these 19 experiments use the default, which is to
switch pkg diagnostics output to NetCDF (via MNC) as soon as pkg MNC is used.

Cheers,
Jean-Michel

On Sat, Aug 05, 2017 at 03:29:52PM -0700, Klymak Jody wrote:
> 
> 
> Sent from my iPhone
> 
> > On Aug 5, 2017, at 12:15, rahele shafiee <rahele_shafiee at yahoo.com> wrote:
> > 
> > Hello everyone
> > 
> > How can I change the test case outputs of  ???.data??? to ???.nc???? I'm using test case ???flt_example???, When I enable the package ???mnc???, only outputs such as temperature, salinity, water level, and current in the output file format ???.nc??? . But I want the tracer outputs to be as 
> > 
> 
> Hi Rahael.  
> 
> You want to use pkg/diagnostics.  You can look in the pkg/tracers to see what diagnostics are set up
> 
> It would be nice if more of the examples were setup w netcdf + diagnostics output turned on by default.  My understanding is that diagnostics is the preferred method of output from the gcm but a lot of the examples are setup with the old fashioned output files.  I only personally became aware of all the power available in the diagnostics package
> 
> I'll use this as a bit of an advertisement.  Right now netcdf files are written one per tile using mnc   If you want just one netcdf file for the whole gcm you can try my fork and use nf90io instead of mnc.   https://github.com/jklymak/MITgcm66h. You need a version of netcdf compiled with a parallel hdf5.  It's not well tested at all, so if you have issues contact me directly (through the github page is easiest) rather than bugging the developers.  
> 
> Cheers.  

> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support



More information about the MITgcm-support mailing list