[MITgcm-support] Restart a simulation using finer spatial simulation (interpolation)

[Jan Klaus Rieck]

Hi Juncheng Zhang,

You don't need to worry about the grid-related files (XC, YG, etc.).
When you set up the 1km-resolution configuration the model will
automatically generate these files when you run the simulation. 

In principle, you just need to interpolate the pickup-files and specify
to use them in 'data' of the 1km-simulation. In practice (at least in
my experience) this is very often unstable and the model will crash.
But I guess that depends on the complexity of your model. If it
crashes, you could start the 1km-simulation with a very small time step
or higher viscosity etc. and adjust later on, sometimes that works.

Another way to do it would be to start the 1km-run from scratch and use
the interpolated output temperature (salinity if you have) of the 5km-
run as initial conditions ('hydrogThetaFile' etc.). This is often more
stable if you use output averaged over some period, probably because
the fields are smoother and the model can "create" its own matching
velocities etc. instead of being forced to use the interpolated
snapshots from the pickup files. But you need to spin-up again before
your 1km-run is in quasi-equilibrium (but this spin-up will be a lot
shorter compared to starting the 1km-run from rest).

Regarding the interpolation it works better to use "nearest neighbour"
methods instead of linear interpolation in my experience. I assume that
is because the linear interpolation can result in unstable
stratification in some cases because it has no information about the
equation of state while you will have only water columns that existed
stably in the 5km-run when you use "nearest neighbour". Again, this
depends on the nature and complexity of your configuration.

Cheers,
Jan


On Wed, 2024-11-13 at 00:51 +0000, Juncheng Zhang wrote:
> Hi all, 
>  
> I am trying to restart a simulation (5km resolution) with 1 km
> resolution, and I understand I need to do the interpolation,
> But I am wondering what are the files that I need to change, I saw
> the output directory has files such as XC, YC, DXG,DXF, DYC,DYF, T,
> Eta, Layer_HS, U, V, W …
> Should I change all the files in .meta? and also, I find the pickup
> files also has dimension in coarse resolution, should I do
> interpolation to the pickup file as well?
>  
> And any suggestion for interpolation would also help, currently what
> I did is read the ‘mds’ data and do interpolation and rewrite it to
> an mds meta/data file set.
>  
> Thank you, I appreciate any help from you
>  
> Kind Regards
> Juncheng Zhang
> PhD Student, 
> Earth, Atmospheric, and Planetary Physics
> 
> Department of Physics
> University of Toronto
> McLennan Physical Laboratories
> 60 St Gorge St
> Toronto, Ontario, M5S 1A7
> 
> juncheng.zhang at mail.utoronto.ca
> 
> Pursue STEM
>  
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support