[MITgcm-support] global simulations and resolution

[Martin Losch]

Hi there,

I am running a 4x4 degree global simulation (almost identical to the 
verification experiment global_ocean.90x40x15) in parallel to a 2x2 
degree global simulation (180x80x23). These simulations differ mainly 
in their respective horizontal and vertical resolution. Of course, I 
had (initally) friction parameters adjusted, so that viscAh=5e5 in the 
4x4-run became 5e4 in the 2x2 run.
Both experiments use trenberth winds, ncep heat flux (short wave and 
the rest), SST and SSS restoring to Levitus, GM, KPP, no seaice, no 
Arctic ... All runs are integrated for 3000 years with asynchronous 
time stepping (deltaTtracer=172800 for the 4x4 and 43200 for the 2x2 
runs).
The solutions are broadly similar in terms of circulation (the 4x4 run 
is more sluggish, but I can tune the friction parameters so that they 
give similar ACC transports, for example). What is really different are 
the water masses, and I cannot see a way to make them agree more 
closely:
broadly speaking, in the 2x2-run: when compared to Levitus data, the 
Southern Ocean (south off the ACC) is too cold and too salty. The deep 
ocean (below 3000m) is far too cold. The 4x4-run is too cold and too 
fresh below 3000m, but in the Southern Ocean theta is nearly OK, but 
salinity is too low. (for pictures see 
http://mitgcm.org/~mlosch/run2x2_ts.pdf and 
http://mitgcm.org/~mlosch/run4x4_ts.pdf)

I would like to know, how this difference can be explained, that is, 
how I can tune either model (preferably the 2x2 degree model, because 
is too cold) to be similar to the other. Is there any experience with 
this out there? Is it possible that the vertical resolution in 
combination with surface restoring is responsible for this difference 
(the 4x4-top layer is 50m thick, whereas the  2x2-top layer only 10m.)?
Any suggestion is greatly appreciated.

Martin