[MITgcm-support] ploblem with a variable-grid global ocean circulation model

[Martin Losch]

Hi lianzhan,

I am afraid you'll have to provide more detail than "the result is   
noise".
Does the model blow up (stop with "SOLUTION HEADING OUT OF BOUNDS")  
or is the noise "stable", that is, can you continue to run with the  
noisy solution?
Can you make a 2D plot of the "noise", the resolution (dx,dy, or if  
your grid is "isotropic" of cell area), also diagnose VISCAHZ/ 
VISCAHD, so see the actual size of your viscosity, etc. and
NOT send the plots to the list (as they may be large) but put them is  
somewhere on the web?

Martin

On 26 May 2007, at 08:11, lianzhan at fio.org.cn wrote:
>
> Hi Mitgcm,
>
> Yes,I use  delx/dely of 0.1667 to 2 deg.And I set viscAh=0 and  
> viscAhGrid=0.001 as your advice.But the result is still noise.
>
> Is there any expecial request for the setting of grid in the MITgcm ?
>
> Thank you for you help.
>
>> Hi lianzhan,
>
>> do I interpret this correctly: you use delx/dely of 0.1667 to 2 deg?
>> If that's the case then your horizontal viscosity is way too hight
>>> viscAh=1.E4,
>>> viscAhGrid=2,
>> When you look at mom_calc_visc.F you'll see that both viscAh and
>> viscAhGrid contribute to the overall viscosity. viscAh=1e4 is
>> suitable for a 2deg grid and to that you add even more viscosity, so
>> you may be running into a cfl-problem with the viscosities. Courtesy
>> of mom_calc_visc.F:
>> C     harmonic viscosity=
>> C       viscAh (or viscAhD on div pts and viscAhZ on zeta pts)
>> C       +0.25*L**2*viscAhGrid/deltaT
>
>> So in your case 1e4 + 2 * .25*(2*pi/180*6371000)^2/60 = 4.1215e+08 to
>> 2 * .25*(.1667*pi/180*6371000)^2/60 + 1e4 =  2.8733e+06
>> set viscAh=0 and viscAhGrid to a much smaller number so that your cfl
>> will not blow:
>> deltaT * (viscAh/L^2) = deltaT * viscAhGrid < 1
>> Or have a look at
>> http://mitgcm.org/r2_web_testing/latest/online_documents/node85.html
>> and subsections to figure out what you really need.
>
>> Martin
>
>
> On 8 May 2007, at 08:35, lianzhan at fio.org.cn wrote:
>
>> Hi mitgcm,
>>
>> I met a ploblem with a variable-grid global ocean circulation
>> model. There are 546 grids in meridian direction,and 521 grids in
>> latitudinal direction, as following. I seem to get noise in all the
>> domain.
>>
>> Is the grid suitable for the MITgcm? What do I need to modify?
>>
>> That's all.Thanks for your help
>>>
>>> # ====================
>>>
>>> # | Model parameters |
>>>
>>> # ====================
>>>
>>> #
>>>
>>> # Continuous equation parameters
>>>
>>> &PARM01
>>>
>>> tRef= 16.0 , 15.2 , 14.5 , 13.9 , 13.3 ,
>>>
>>>       12.4 , 11.3 ,  9.9 ,  8.4 ,  6.7 ,
>>>
>>>        5.2 ,  3.8 ,  2.9 ,  2.3 ,  1.8 ,
>>>
>>>        1.5 ,  1.1 ,  0.8 ,
>>>
>>> sRef= 34.65, 34.75, 34.82, 34.87, 34.90,
>>>
>>>       34.90, 34.86, 34.78, 34.69, 34.60,
>>>
>>>       34.58, 34.62, 34.68, 34.72, 34.73,
>>>
>>>       34.74, 34.73, 34.73,
>>>
>>> viscAz=1.E-4,
>>>
>>> viscAh=1.E4,
>>>
>>> viscAhGrid=2
>>>
>>> no_slip_sides=.FALSE.,
>>>
>>> no_slip_bottom=.TRUE.,
>>>
>>> diffKhT=1.E3,
>>>
>>> diffKzT=2.E-5,
>>>
>>> diffKhS=1.E3,
>>>
>>> diffKzS=2.E-5,
>>>
>>> tAlpha=2.E-4,
>>>
>>> sBeta =7.4E-4,
>>>
>>> gravity=9.81,
>>>
>>> rigidLid=.FALSE.,
>>>
>>> implicitFreeSurface=.TRUE.,
>>>
>>> useCDscheme=.TRUE.,
>>>
>>> useNHMTerms=.TRUE.,
>>>
>>> eosType='LINEAR',
>>>
>>> readBinaryPrec=32,
>>>
>>> tempStepping=.FALSE.,
>>>
>>> saltStepping=.FALSE.,
>>>
>>> implicitDiffusion=.TRUE.,
>>>
>>> implicitViscosity=.TRUE.,
>>>
>>> &
>>>
>>>
>>>
>>> # Elliptic solver parameters
>>>
>>> &PARM02
>>>
>>> cg2dMaxIters=1000,
>>>
>>> cg2dTargetResidual=1.E-13,
>>>
>>> &
>>>
>>> # Time stepping parameters
>>>
>>> &PARM03
>>>
>>> startTime=0,
>>>
>>>
>>>
>>> endTime=311040000,
>>>
>>> # endTime=864000.,
>>>
>>> deltaTmom=60.0,
>>>
>>> tauCD=321428.,
>>>
>>> deltaTtracer=3600.0,
>>>
>>> deltaTClock =3600.0,
>>>
>>> cAdjFreq=0.,
>>>
>>> abEps=0.1,
>>>
>>> pChkptFreq=311040000.,
>>>
>>> chkptFreq=311040000.,
>>>
>>> dumpFreq=3600.0,
>>>
>>> tauThetaClimRelax=2592000.0,
>>>
>>> tauSaltClimRelax=2592000.0,
>>>
>>> monitorFreq=1.,
>>>
>>> &
>>>
>>>
>>>
>>> # Gridding parameters
>>>
>>> &PARM04
>>>
>>> usingCartesianGrid=.FALSE.,
>>>
>>> usingSphericalPolarGrid=.TRUE.,
>>>
>>> delZ=4.,6.,10.,15.,23.,32.,48.,68.94,
>>>
>>>      88.1964,132.3541,216.3472,331.9537,
>>>
>>>      467.8573, 610.7548, 746.6584, 862.2648,
>>>
>>>      946.2580, 990.4156,
>>>
>>> phiMin=-80.,
>>>
>>> # delY=160*1,
>>>
>>> # delX=360*1,
>>>
>>>  delXfile='dx.bin',
>>>
>>>  delYfile='dy.bin',
>>>
>>> &
>>>
>>>
>>>
>>> # Input datasets
>>>
>>> &PARM05
>>>
>>> hydrogThetaFile='temp.bin',
>>>
>>> hydrogSaltFile='salt.bin',
>>>
>>> bathyFile='biangrid.bin',
>>>
>>> hydrogSaltFile='salt.bin',
>>>
>>> bathyFile='biangrid.bin',
>>>
>>> zonalWindFile='fx.bin',
>>>
>>> meridWindFile='fy.bin',
>>>
>>> thetaClimFile='sst.bin',
>>>
>>> saltClimFile='sss.bin',
>>>
>>> &