[MITgcm-support] using time step less than a second ??
jmc at ocean.mit.edu
Wed Jul 11 21:51:09 EDT 2012
The problem with pkg/cal and non integer time when expressed in second
does not seems very easy (and fast) to fix for me (partly because
some part of pkg/cal are still little bit obscure for me) and
would need to review where are the true limitations.
An alternative would be to allow to use pkg/exf without pkg/cal
(but again, need to evaluate the extent of the required modifications).
But may be, you could try to use pkg/bulk_force (similar bulk formulae
as in exf, coming also from Large & Pond) which is less flexible
than pkg/exf (fewer options) specially regarding forcing fields
time-specifications (e.g., a unique reccord frequency accoss all forcing
files) but does not require pkg/cal.
In the documentation: section 6.4.5 (not really up-to-date)
The only verification experiment using pkg/bulk_force is
with "bulk_force" in global_ocean.cs32x15/code/packages.conf
and "useBulkforce=.TRUE.," in global_ocean.cs32x15/input.thsice/data.pkg
and parameter file "data.blk" in global_ocean.cs32x15/input.thsice
On Wed, Jul 11, 2012 at 09:59:09AM -0500, Abbas Dorostkar wrote:
> Hi Martin,
> I need to run a very high resolution non-hydrostatic simulation with actual wind data. This means, I want to use the bulk formulae of Large & Pond. However, the simulation period is short and covers only 6 days. Probably to resolve this issue, I have to compute wind stresses outside mitgcm ?
> On 2012-07-11, at 8:11 AM, Martin Losch wrote:
> > Abbas, I don't think that this has changed.
> > Do you seriously need the combination of exf and time steps below 1sec? EXF is very convenient for simulations with "realistic" forcing that span months to years to decades. But with a time step < 1s I assume that your simulations are very short in simulation period, so simpler forcing methods (such as the default external_fields_load method) would also be appropriate. Is that the case?
> > Martin
> > On Jul 6, 2012, at 11:55 PM, Abbas Dorostkar wrote:
> >> Hello all,
> >> I remember we could not use time steps lower than 1 sec if we were using EXF package and/or some sort of calendar in simulation. I am wondering if there has been improvement since then?
> >> I was getting this error message back then:
> >> (PID.TID 0000.0001) // execution environment starting up...
> >> (PID.TID 0000.0001)
> >> (PID.TID 0000.0001) // MITgcmUV version: checkpoint61h
> >> (PID.TID 0000.0001) // Build user: hpc1921
> >> (PID.TID 0000.0001) // Build host: sflogin0
> >> (PID.TID 0000.0001) // Build date: Wed Aug 5 11:29:58 EDT 2009
> >> PID.TID 0000.0001) ># *******************
> >> (PID.TID 0000.0001) ># Calendar Parameters
> >> (PID.TID 0000.0001) ># *******************
> >> (PID.TID 0000.0001) > &CAL_NML
> >> (PID.TID 0000.0001) ># TheCalendar='gregorian',
> >> (PID.TID 0000.0001) > TheCalendar='model',
> >> (PID.TID 0000.0001) > startDate_1=20071216,
> >> (PID.TID 0000.0001) > startDate_2=000000,
> >> (PID.TID 0000.0001) > &
> >> (PID.TID 0000.0001)
> >> (PID.TID 0000.0001) CAL_READPARMS: finished reading data.cal
> >> (PID.TID 0000.0001) cal_Set: The time step is less than a second.
> >> Thank you
> >> Abbas
> >> _______________________________________________
> >> MITgcm-support mailing list
> >> MITgcm-support at mitgcm.org
> >> http://mitgcm.org/mailman/listinfo/mitgcm-support
> > _______________________________________________
> > MITgcm-support mailing list
> > MITgcm-support at mitgcm.org
> > http://mitgcm.org/mailman/listinfo/mitgcm-support
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
More information about the MITgcm-support