[MITgcm-devel] sea ice clean up this week
Martin Losch
Martin.Losch at awi.de
Mon Feb 13 08:32:28 EST 2012
Hi Gael,
I prefer a version with fewer divisions and multiplications, so that I do not have to understand, why (1 - 11.5/12.5)*12.5/11.5 = 1/11.5 ~=1 (o: (I am sure I made as mistake.)
Go ahead and do it. If some of this changes the results at truncation level please consider removing 1/QI etc (see my previous list which will also change the results at truncation level).
M.
On Feb 13, 2012, at 1:50 PM, Gael Forget wrote:
> Martin,
>
> you are right I messed up the untested case. Thanks for catching it early.
>
> The correct code (in the current state of affairs) would be
> elseif ( (SEAICEturbFluxFormula.EQ.4).AND.
> & (AREApreTH(I,J) .GT. 0. _d 0) ) then
> MixedLayerTurbulenceFactor = ONE/MCPHEE_TAPER_FAC
> Correct. May be now that we can switch on this case through a run time parameter
> we should have the second mcphee case tested in a sub-exp of Ian's 1D exp.
>
> Also I went half way there. My intention was to ultimately have the MCPHEE_TAPER_FAC param
> become 11.5/12.5 (in readparms or initfixed normally) and not be 12.5. In growth this would give
> if ( (SEAICEturbFluxFormula.EQ.3).AND.
> & (AREApreTH(I,J) .GT. 0. _d 0) ) then
> MixedLayerTurbulenceFactor = ONE - MCPHEE_TAPER_FAC*AREApreTH(I,J)
> elseif ( (SEAICEturbFluxFormula.EQ.4).AND.
> & (AREApreTH(I,J) .GT. 0. _d 0) ) then
> MixedLayerTurbulenceFactor = ONE - MCPHEE_TAPER_FAC
> Correct? This makes the tapering strategy clearer and has no division in it.
>
> I have a check-in planned today mostly for diags, checks and summary.
> I can do the above at the same time if this works for you. Does it?
>
> Cheers,
> Gael
>
> ps : I still need to send you guys a summary+guidelines regarding fridays' changes...
>
>
> On Feb 13, 2012, at 6:03 AM, Martin Losch wrote:
>
>> Gael,
>>
>> I have not been able to understand the new MCPHEE+GRADIENT_MIXED_LAYER_* logic.
>>
>> - MCPHEE_TAPER_FAC = 12.5
>> - seaice_init_fixed for SEAICEturbFluxFormula.EQ.3 or 4
>> SEAICE_availHeatFrac= MCPHEE_TAPER_FAC * more constants
>> SEAICE_availHeatFracFrz=ZERO
>> - seaice_growth:
>> elseif ( (SEAICEturbFluxFormula.EQ.4).AND.(AREApreTH(I,J) .LE. 0. _d 0) ) then
>> MixedLayerTurbulenceFactor = MCPHEE_TAPER_FAC
>> else
>> [ ... ]
>>
>> This looks to me as if you have MCPHEE_TAPER_FAC^2. It should be MixedLayerTurbulenceFactor=1 in the above case as the default (the final else) 1/MCPHEE_TAPER_FAC. So I suggest that we remove MCPHEE_TAPER_FAC in seaice_init_fixed:
>> SEAICE_availHeatFrac= STANTON_NUMBER *
>> & USTAR_BASE / dRf(kSurface) * SEAICE_deltaTtherm
>> and adjust seaice_growth accordingly. This also avoids a division in a loop. If you don't object. I'll do it. (This options is not tested, because these changes do not change any results).
>>
>> M.
>>
>>
>> On Feb 9, 2012, at 8:16 PM, Gael Forget wrote:
>>
>>> Dear all,
>>>
>>> below is (A) a summary of the last two day clean-up steps (cosmetic changes)
>>> and (B) what it means for known users (based on Patrick's table; attached).
>>> In case you follow testreport, you already know that the modifs have not
>>> changed anything to results. Notice : there is more clean-up coming in the
>>> next couple days, so you may want to wait until doing an update, up to you.
>>>
>>> Cheers,
>>> Gael
>>>
>>> (A)
>>>
>>> retired CPPs :
>>> MCPHEE_OCEAN_ICE_HEAT_FLUX
>>> GRADIENT_MIXED_LAYER_TURBULENCE_FACTOR
>>> SEAICE_OCN_MELT_ACT_ON_AREA, SEAICE_DO_OPEN_WATER_GROWTH
>>> SEAICE_DO_OPEN_WATER_MELT, FENTY_AREA_EXPANSION_CONTRACTION
>>>
>>> added run time params. I include a copy of the doc in SEAICE_PARAMS.h and
>>> tried to trace back where the various options came from (text in parentheses).
>>> SEAICE_doOpenWaterGrowth, SEAICE_doOpenWaterMelt
>>> T/F just like the old CPP versions
>>> T use open water heat flux directly to grow ice (Ian way)
>>> F cool ocean, and grow later if needed (Zhang way)
>>> SEAICEturbFluxFormula :: selects formula for ocean-ice turbulent flux
>>> 1=direct spec. of SEAICE_availHeatFrac/Frz (Dimitri/Ian way)
>>> 2=spec. via SEAICE_gamma_t/_frz (Martin/Ian way)
>>> 3=McPhee with linear tapering (Ian latest way)
>>> 4=McPhee with step function (Ian other latest way)
>>> SEAICE_areaLossFormula :: selects formula for ice cover loss from melt
>>> 1=from all but only melt conributions by ATM and OCN (Ian way)
>>> 2=from net melt-growth>0 by ATM and OCN (Gael/Ian way)
>>> 3=from predicted melt by ATM (Zhang way)
>>> SEAICE_areaGainFormula :: selects formula for ice cover gain from open water growth
>>> 1=from growth by ATM (Ian way)
>>> 2=from predicted growth by ATM (Zhang way)
>>>
>>> as a side note, I put stops in seaice_check.F for combinaisons
>>> that are obviously problematic. Tyipically, if you do NOT use
>>> SEAICE_doOpenWaterGrowth you need to use the Zhang way
>>> for SEAICE_areaLossFormula (3) and SEAICE_areaGainFormula (2),
>>> and this is the only case allowed with SEAICE_GROWTH_LEGACY.
>>>
>>> (B)
>>>
>>> You should double check, but based on Patrick's table,
>>> here is what you will need to add to your namelists :
>>>
>>> Martin : in SEAICE_GROWTH_LEGACY case you should be all set. Otherwise
>>> if TRUE SEAICE_doOpenWaterGrowth then SEAICE_areaLossFormula = 2
>>> if FALSE SEAICE_doOpenWaterGrowth then SEAICE_areaLossFormula = 3, SEAICE_areaGainFormula=2
>>>
>>> Ian et al: SEAICE_doOpenWaterMelt=.TRUE.
>>> SEAICEturbFluxFormula=3
>>> ps: def/undef SEAICE_OCN_MELT_ACT_ON_AREA was doing nothing
>>> since FENTY_AREA_EXPANSION_CONTRACTION was def.
>>>
>>> Me : TBD.
>>>
>>>
>>> <seaice_whodoeswhat_3.pptx>
>>>
>>> _______________________________________________
>>> MITgcm-devel mailing list
>>> MITgcm-devel at mitgcm.org
>>> http://mitgcm.org/mailman/listinfo/mitgcm-devel
>>
>>
>> _______________________________________________
>> MITgcm-devel mailing list
>> MITgcm-devel at mitgcm.org
>> http://mitgcm.org/mailman/listinfo/mitgcm-devel
>
>
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel
More information about the MITgcm-devel
mailing list