[MITgcm-support] Wave damping

[Patrick Rosendahl]

Hello Jean-Michel,

maybe there are problems with non-hs. This says the model:

CONFIG_CHECK: nonHydrostatic NOT SAFE with non-fully implicit Barotropic 
solver
CONFIG_CHECK: To by-pass thisSTOP, comment this test and re-compile 
config_check

...

The model runs into NaNs pretty quick then. Maybe reducing the timestep 
helps, but with smaller timesteps my "old" version works, too.


Thanks for the idea!
Patrick


Jean-Michel Campin wrote:
> Hi Patrick,
> 
> I guess your recent e-mail is related to this one.
> 
> If the waves are in the external mode, the fully implicit
> time-stepping (backward in time) will damp the waves.
> You could try the Crank-Nickelson time stepping:
>  implicSurfPress=0.5,
>  implicDiv2DFlow=0.5,
> (see verification/adjustment.cs-32x32x1/input.nlfs/data)
> that conserve the external mode energy.
> But I am not so sure that it works well with the Non-Hydrostatic
> option (In fact, I've never tried).
> 
> Jean-Michel
> 
> On Thu, Mar 15, 2007 at 02:23:47PM +0100, Patrick Rosendahl wrote:
> 
>>Hello all,
>>
>>I am doing a non-hs simulation, where I generate waves at the one end of 
>>the waterbody. On the other side there is a damping beach.
>>
>>Now the problem. If I use a timestep of 1e-3
>>- 1Hz waves travel with pretty much no damping.
>>- 1.5Hz waves travel with very little damping
>>- 2Hz waves are damped heavily, the profile of the wave peaks looks like 
>>an exponential decay. Setting all the viscosities to 0.0 had no effect, 
>>(including visc4 which is not in the file below.)
>>
>>To remove the damping for 2Hz waves, I have to decrease the timestep to 
>>1e-4.
>>
>>This seems weird to me, because the 2Hz waves travel slower than 1Hz 
>>waves ("external mode"), and the wavemaker generated velocities are the 
>>same (0.02 cm/s maximum vel in internal mode). Hence, the CFL criterion 
>>is met "even better".
>>
>>Is there some setting in the code, that damps short waves, depending on 
>>the timestep? (Shapiro filter is turned off.)
>>
>>Find my settings below.
>>
>>Best regards,
>>Patrick Rosendahl
>>
>>
>>
>>-----------------------------------------------package.conf
>>rw
>>mdsio
>>mnc
>>obcs
>>mom_fluxform
>>generic_advdiff
>>
>>
>>
>>-----------------------------------------------data
>># ====================
>># | Model parameters |
>># ====================
>>#
>># Continuous equation parameters
>> &PARM01
>> tRef=135*20.0,
>># 1cm Salz
>> sRef=1*35.0,134*0.0,
>> viscAh=1.0E-6,
>> viscAz=1.0E-6,
>> no_slip_sides=.FALSE.,
>> no_slip_bottom=.TRUE.,
>> diffKhT=1.0E-7,
>> diffKzT=1.0E-6,
>> diffKhS=5.0E-7,
>> diffKzS=1.0E-6,
>> f0=0.0,
>> sBeta =0.,
>> gravity=9.81,
>> rhoConst=1000.0,
>> rhoNil=1000.0,
>># heatCapacity_Cp=3900.0,
>> rigidLid=.FALSE.,
>> implicitDiffusion=.TRUE.,
>> implicitViscosity=.TRUE.,
>> implicitFreeSurface=.TRUE.,
>> eosType='LINEAR',
>> nonHydrostatic=.TRUE.,
>> readBinaryPrec=32,
>> tempAdvScheme=4,
>> saltAdvScheme=4,
>> exactConserv=.TRUE.,
>># nonlinFreeSurf=3,
>> &
>>
>># Elliptic solver parameters
>> &PARM02
>> cg2dMaxIters=1000,
>> cg2dTargetResidual=1.E-13,
>> cg3dMaxIters=10,
>> cg3dTargetResidual=1.E-9,
>> &
>>
>># Time stepping parameters
>> &PARM03
>> nIter0=0,
>> nTimeSteps=40000,
>> deltaT=0.001,
>> abEps=0.1,
>> pChkptFreq=1.0,
>> chkptFreq=1.0,
>> dumpFreq=0.2,
>> monitorFreq=0.1,
>># outputTypesInclusive=.TRUE.,
>> &
>>
>># Gridding parameters
>> &PARM04
>> usingCartesianGrid=.TRUE.,
>> usingCylindricalGrid=.FALSE.,
>> usingCurvilinearGrid=.FALSE.,
>> delX=800*0.03,
>> dYspacing=0.1,
>> delZ=0.010,40*0.001,20*0.002,20*0.003,10*0.005,23*0.010,10*0.005,11*0.002,
>> &
>>
>># Input datasets
>> &PARM05
>># zonalWindFile='windstressx.const_0.1.bin.real*4',
>># hydrogThetaFile='thetaPol.bin',
>> bathyFile='depth.closedE_0.5.bin.real*4',
>># tCylIn  = 0
>># tCylOut  = 20
>> &
>>_______________________________________________
>>MITgcm-support mailing list
>>MITgcm-support at mitgcm.org
>>http://mitgcm.org/mailman/listinfo/mitgcm-support
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support