[MITgcm-support] Seeking Assistance with Numerical Instability in MITgcm Simulation of a Plateau Lake

[Jody Klymak]

Hi Hong Tang

I agree with Martin. I deal with similar model scales all the time, and your time step will need to be order 1-s, certainly not 100-s.

Cheers,   Jody

On Aug 1, 2024, at 7:59 AM, Martin Losch <Martin.Losch at awi.de> wrote:

Dear Hong Tang,

since it is a lake I assume that the salinity is zero and you don’t have to worry about salinity advection (so you could set saltStepping = .FALSE. to save some time), but your temperature advection will be done by the default central differences (tempAdvScheme = 2), which requires stabilisation by diffusion, which is unset and will default to zero.

So either you change your advection scheme to one with numerical diffusion to stabiliize it, e.g.
 tempAdvScheme = 3, 30, 33 or 7 (have a look ad GAD.h for all options)

or (and?) you need to set diffKhT to some value > 0

and your diffKrT (vertical diffusion) should also be non-zero. With 1m vertical resolution, the vertical cfl will probably be the limiting factor for the timestep (monitor adv_cfl_wvel, etc.)

Martin

On 1. Aug 2024, at 16:48, 唐鸿 <tanghong22 at mails.ucas.ac.cn> wrote:


Dear MITgcm Support Team,


I hope this message finds you well.


I am currently working on a simulation project using the MITgcm to study the dynamic processes of a lake situated on the Tibetan Plateau, which experiences freezing during winter. The lake, located at an altitude of 4200m with a depth of 28m and a maximum width of 38km, presents unique challenges due to the strong radiation on the plateau that leads to noticeable warming of the water beneath the ice.


I am writing to seek your expert advice as I have encountered significant numerical instability in my simulations despite my efforts to adjust model parameters. My current setup is that the horizontal grid size is 500m and the vertical grid size is 1m. I have modified the model time step and drive data time to 100s, but I still encounter serious numerical instability.


 I have attached my data file, which contains the configuration parameters of my model setup.


# ====================
# | Model parameters |
# ====================
#
 &PARM01
 tRef= 10*11.,
 sRef= 10*0.27,
 no_slip_sides=.FALSE.,
 no_slip_bottom=.TRUE.,
 viscAh=1.E-4,
 viscAr=1.E-4,
 HeatCapacity_Cp=3986.,
 rhoConstFresh=1000.,
 rhoNil=1000.,
 eosType='JMD95Z',
#
 nonHydrostatic=.TRUE.,
 rigidLid=.TRUE.,
 implicitFreeSurface=.FALSE.,
#
 useCoriolis=.FALSE.,
 use3dCoriolis=.FALSE.,
#
 useSingleCpuIO=.TRUE.,
 readBinaryPrec=32,
 writeBinaryPrec=32,
 &

# Elliptic solver parameters
 &PARM02
 cg2dMaxIters=700,
 cg3dMaxIters=100,
 cg2dTargetResidual=1.E-7,
 cg3dTargetResidual=1.E-7,
 cg2dUseMinResSol=1,
 &

# Time stepping parameters
 &PARM03
 abEps=0.1,
 deltaT=100,
 nIter0=0,
 nTimeSteps=315361,
 dumpFreq=86400.,
 chkPtFreq=2592000.,
 pChkptFreq=8640000.,
 monitorFreq=100.,
 &

# Gridding parameters
 &PARM04
 usingCartesianGrid=.TRUE.,
 dXspacing=500.,
 dYspacing=500.,
 delR=10*1.,
 &
# Input datasets
 &PARM05
 bathyFile='Bathy_500m.bin',
 the_run_name='Ngoring_LakeTest_HH',

 &




I would greatly appreciate any suggestions or guidance you could provide to improve the numerical stability of my simulation. Thank you very much for considering my request. I look forward to your valuable insights and suggestions.


Best regards,


Hong Tang

University of Chinese Academy of Sciences



_______________________________________________
MITgcm-support mailing list
MITgcm-support at mitgcm.org
http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support

_______________________________________________
MITgcm-support mailing list
MITgcm-support at mitgcm.org<mailto:MITgcm-support at mitgcm.org>
http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.mitgcm.org/pipermail/mitgcm-support/attachments/20240801/313aba3c/attachment-0001.html>