[MITgcm-devel] exf wind-stress masking

Mon Jan 30 03:56:49 EST 2012

Jean-Michel,

I think that masking taux/tauy does make sense, so let's do it. Will you?

Martin

On Jan 27, 2012, at 9:24 PM, Jean-Michel Campin wrote:

> Hi Martin and others,
> 
> I think the masking is not "strictly necessary" so that
> not masking fields in exf when using seaice is OK.
> Was conscerned about wind-stress computation (which depends
> on air temp & SST in exf), but it is masked in exf_mapfields.F
> when put on C-grid. And in pkg/seaice wind-stress does not depend on 
> air temperature.
> 
> However, I think it would be better to always masks the air-temp
> (and may be also the air humidity & lwflux) to avoid computing
> fluxes (sensible, latent, net-LW) using SST from a land point.
> 
> And regarding seaice_ocean_stress (in relation with useHB87stressCoupling),
> might be cleaner to make the changes you propose to remove the exf-masking
> dependency; or could leave it as it is (little odd, but more accurate ?).
> 
> Cheers,
> Jean-Michel
> 
> On Tue, Jan 10, 2012 at 02:14:23PM -0500, Jean-Michel Campin wrote:
>> Hi Martin,
>> 
>> I would like to do more tests, since I realized there are some 
>> combinations that are not tested (but not so exotic that
>> they can be used), such as exf-interp + seaice.
>> 
>> I think we agree that we don't want ocean surface temperature on land
>> to have an effect on the solution. And this is what I would like to 
>> check first.
>> 
>> Cheers,
>> Jean-Michel
>> 
>> On Tue, Jan 10, 2012 at 10:56:01AM +0100, Martin Losch wrote:
>>> Hi Jean-Michel,
>>> 
>>> the reason appears to be the unmasked taux/tauy.
>>> 
>>> The are added to FORCEX/Y0 which in turn is added to FORCEX/Y, and these are then added to the rhs, but with masking (seaiceMaskU/V). In seaice_ocean_stress taux/tauy are used, but without masking (and as far as I can see that's also the case in seaice_freedrift).
>>> 
>>> I can fix that by masking taux/y with maskW/S or seaiceMaskU/V if you like and then the dependence on the exf-masking goes away.
>>> 
>>> Martin
>>> 
>>> 
>>> 
>>> On Jan 9, 2012, at 9:20 PM, Jean-Michel Campin wrote:
>>> 
>>>> Hi Martin,
>>>> 
>>>> I did some quick tests, comparing with/without masking of exf variables
>>>> (In the 1rst case, I've just commented out the block:      
>>>>     IF ( useSEAICE ) THEN
>>>> C     Do not filter with pkg/seaice because of B/C-grid interpolation
>>>>      ....
>>>> in pkg/exf/exf_init_fixed.F)
>>>> 
>>>> And I only get different results in 2 exp.:
>>>> a) lab_sea/input.fd   (SEAICEuseFREEDRIFT=.true.)
>>>> b) lab_sea/input.hb87 (EVP & useHB87stressCoupling)
>>>>  (but diff goes away with useHB87stressCoupling=F)
>>>> 
>>>> Regarding SEAICEuseFREEDRIFT, the differences comes from 
>>>> seaice_freedrift.F (I don't get any difference with my free-drift         
>>>> computation within seaice_lsr.F).
>>>> 
>>>> To conclude:
>>>> 1) it's worth to know that these 2 options (SEAICEuseFREEDRIFT
>>>>  and useHB87stressCoupling) are sensitive to masking within EXF.
>>>> 2) I think it might be better to always (whether we use SEAICE or
>>>> not) mask heat-flux & fresh water flux related fields, 
>>>> and to leave only wind & windstress not masked when using SEAICE.
>>>> 
>>>> Cheers,
>>>> Jean-Michel
>>>> 
>>>> On Wed, Jan 04, 2012 at 09:32:35AM +0100, Martin Losch wrote:
>>>>> Hi Jean-Michel,
>>>>> 
>>>>> I am not sure, but I think that for B-grid seaice, the wind (stress) needs to be available a Z-points and this interpolation is done in pkg/seaice/dynsolver.F, so that when masking in exf removes wind values over land, you underestimate the wind over boundary values of seaice.
>>>>> Having said that, the boundary points of B-grid seaice have (uice,vice)=0 as a boundary conditions, so that this masking does not make any difference, so on second thought you are probably right, and this differentiation can be removed safely.
>>>>> 
>>>>> M.
>>>>> 
>>>>> On Jan 3, 2012, at 5:49 PM, Jean-Michel Campin wrote:
>>>>> 
>>>>>> Hi, 
>>>>>> 
>>>>>> I am going to check-in some unfinished changes in exf:
>>>>>> I realized that the masking of the wind-stress is wrong (when not on 
>>>>>> C-grid, which happen when interpolated or read from A-grid file).
>>>>>> I fixed this problem and will check-in this (involves also EXF_PARAMS.h
>>>>>> & exf_init_fixed.F) and update output of global_with_exf exp.
>>>>>> 
>>>>>> I also started to add some params in EXF_PARAMS.h + exf_init_fixed.F
>>>>>> for 2-components vector interpolation (I did this before the stress
>>>>>> mask fix), will be part of the changes but they are not yet used.
>>>>>> 
>>>>>> Regarding masking, I don't see a clear reason why it's only
>>>>>> applied when not using SEAICE. If exf works fine without
>>>>>> masking (when using SEAICE), then why do we care about applying
>>>>>> masks ?
>>>>>> 
>>>>>> Also, noticed that exf_init_fixed.F does not compile with default 
>>>>>> pkg/icefront settings (might have been better to put all this
>>>>>> icefront - exf code in pkg/icefront and still call some utility 
>>>>>> exf & cal S/R).
>>>>>> 
>>>>>> Cheers,
>>>>>> Jean-Michel
>>>>>> 
>>>>>> _______________________________________________
>>>>>> 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
>>> 
>>> 
>>> _______________________________________________
>>> 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