search

NOAA-GFDL / MOM6-examples

Example configurations for MOM6 and SIS2
Other
86 stars 145 forks source link

Running MOM6 in a coarser resolution #252

Open whuiskamp opened 5 years ago

whuiskamp commented 5 years ago

Hi everyone, I was wondering if anyone has experience running MOM6/ a C-grid model at resolutions coarser than 1-degree. My group is looking to use the wetting and drying scheme in MOM6 for deglacial simulations, but these necessitate a coarse resolution (we're currently using MOM5 with a ~3 degree resolution) to maintain speed. Does anyone have experience here or could advise if this is feasible? Thanks. Willem

ashao commented 4 years ago

Also, with regard to the speedup part, I wonder if using wide halos in the barotropic solve might buy you a little gain? It's a tradeoff between message passing and doing more calculations, so it's definitely not a guaranteed to be an improvement.

! === module MOM_barotropic ===
BT_USE_WIDE_HALOS = True        !   [Boolean] default = True
                                ! If true, use wide halos and march in during the barotropic time stepping for
                                ! efficiency.
BTHALO = 0                      ! default = 0
                                ! The minimum halo size for the barotropic solver.

(Note: this doesn't show up in MOM_parameter_doc.all, but only in MOM_parameter_doc.layout but you should be able to add it to MOM_input or MOM_override as per usual. We should probably change that at some point).

whuiskamp commented 3 years ago

Hi everyone, I have created an updated grid (now extends to 90S, with the intention of one day performing Antarctic melt experiments) and run the model to equilibrium over 1400 years. Attached are some figures of global overturning and mixed layer depth from the 'best' state I've been able to achieve. The AMOC is weaker than I would like/expect, while in the Southern Ocean we see strong Weddell Sea convection, but an absence of any in the Ross Sea. I'll add that I've only been able to achieve this by disabling THICKNESSDIFFUSE and using very low background diffusivities, surprising given the resolution (which makes me a little nervous). If anyone has suggestions on how I might improve this, I'd be grateful - my tests have so far been unsuccessful (re-enabling thicknessdiffuse yields an AMOC of ~10Sv). mld gmoc

MOM_parameter_doc_all.txt