[MITgcm-support] newbie question about mnc output

Patrick Heimbach heimbach at MIT.EDU
Thu Mar 1 12:39:46 EST 2007


Hi Shafer,


On Mar 1, 2007, at 12:03 PM, Shafer Smith wrote:

> Hi,
>
> Great model, super well documented for the most part,

Where did you find that manual?
;o)

> but perhaps for
> ultra-newbies, a few things are hard to figure out....
>
> I have compiled and run exp0 on an ibm  js20 cluster,  single  
> processor,
> and am trying to understand how to read the output.
> Though I believe I have read the manual, I cannot seem to figure  
> out the
> following:
>

I assume you are referring to the experiment "tutorial_barotropic_gyre/"
which is what exp0 evolved into some 10 months ago.

> 1.  Why are there four of every file?  e.g.
>
> state.0000000000.t001.nc
> state.0000000000.t002.nc
> state.0000000000.t003.nc
> state.0000000000.t004.nc

If you look in SIZE.h, you'll see that the
Nx*Ny=60*60 domain is actually split up into nSx*nSy=2*2=4 virtual  
tiles,
i.e. you have
Nx*Ny = sNx*nSx * sNy*nSy = 2*30 * 2*30
The mnc output is currently always a per-tile (or per-processor) output
which is why you're getting 4 files (even in single-processor mode
if you're using several virtual tiles per proc).

> 2.  How do I make the model save basic snapshots for analysis at some
> interval?
>
> pChkptFreq (permanent checkpoint - i can make multiples by setting  
> this
> less than run time)
> chkptFreq (rolling checkpoints -- overwriting same files??)
> dumpFreq ... I don't understand how to use this parameter.  I also
> found an old post about this, but it seemed to advanced for what I'm
> asking....

dumpFreq should be very basic.
It yields model state snapshots every "dumFreq" seconds.
Since you're using mnc, that output is in files state.*
(try ncdump state.0000000000.t001.nc to check what's in there).

Other alternative is to use the diagnostics package
which is very versatile in usage.
It's used in some verifications, e.g. ideal_2D_oce,  
tutorial_global_oce_biogeo
and documented in chapter 7.1
http://mitgcm.org/r2_web_testing/latest/online_documents/node251.html


> 3.  mnc_assembly.m  seems to expect a series of files differing by
> increasing integer values in its file name.  If I set pChekptFreq to
> save multiple checkpoint files (which I expect is not what I want to
> do), they differ by some larger factor:
>
> pickup.0000000720.t001.nc
> pickup.0000001440.t001.nc
> pickup.0000002160.t001.nc
> pickup.0000002880.t001.nc
> pickup.0000003600.t001.nc
> pickup.0000004320.t001.nc
> pickup.0000005040.t001.nc
> pickup.0000005760.t001.nc
> pickup.0000006480.t001.nc
> pickup.0000007200.t001.nc
> pickup.0000007920.t001.nc

I haven't used it so far,
someone else needs to fill in.

Cheers
-Patrick

---
Dr Patrick Heimbach | heimbach at mit.edu | http://www.mit.edu/~heimbach
MIT | EAPS, 54-1518 | 77 Massachusetts Ave | Cambridge, MA 02139, USA
FON: +1-617-253-5259 | FAX: +1-617-253-4464 | SKYPE: patrick.heimbach





More information about the MITgcm-support mailing list