[MITgcm-support] restart from pickup files with the change in parameters
Martin Losch
Martin.Losch at awi.de
Mon Jul 13 12:14:41 EDT 2009
Hi Gregory,
I am answering both of your questions in the same email:
> I couldn't find in manuals how to prescribe initial conditions for
> U,V,W or Eta (only found for tracers). Is there a way to do that?
Yes, you need to create ieee-be files with the precision of
readBindaryPrec (default is 32, so real*4) for these files and then
set the parameters uVelInitFile (U), vVelInitFile (V), pSurfInitFile
(Eta) in "data", namelist "PARM05". W will be diagnosed from U and V
(there is no way to prescribe W)
Sometimes it's useful to have a look a model/inc/PARAMS.h and model/
src/ini_parms.F to find out what the parameters are and how the
namelists are defined.
And while I write this, Jody has already given you the same answer. I
need to improve my typing speed (o:
Now for the second question:
On Jul 13, 2009, at 5:19 PM, Georgy Manucharyan wrote:
> Hello,
> Is it allowed to restart calculations from pickup files, but with
> the change in parameters such as forcing fields, diffusion
> coefficients, time step, ..? And is there a way to copy the model
> state and put it as an initial condition for another calculations
> (with different parameters) ?
Yes, that's allowed and regularily done (e.g. for long spinup
integrations with accelerated tracer time steps, or for high
resolution runs to damp out initial shocks with very short time
steps). The only thing to look out for is the correct name of the
pickup files. They have to correspond to the value of nIter0 (which
you specify this directly in data instead of startTime). If you change
parameter you'll encounter shocks at the begining of the integration.
You can use the pickup files and disect them to produce "hot start"
files of your first question. See pickup*.meta to find out what's in
the pickup files and how the fields are ordered. See model/src/write/
read_pickup.F to see how the pickup files are written or read.
Martin
More information about the MITgcm-support
mailing list