[MITgcm-devel] ocean_emissivity consistency

Sun Jan 8 11:09:29 EST 2012

Hi Martin and other,

There has been a checkpoint (63h) since these changes were
made, so I would propose to change the default EXF_OPTIONS.h to have
#define EXF_LWDOWN_WITH_EMISSIVITY, and at the same time,
since it's going to change several results/output*.txt, 
to use the default setting (from EXF) for snow & ice emissivity
in sereval data.seaice.
Might be usefull to keep one (set of) experiment unchanged;
we usually select lab_sea (all 4 fwd + 4 ad) for this purpose, 
but could pick an other one (but preferentially not 
global_ocean.cs32x15). What do you think ? 

I could make those changes once we agree on.

Cheers,
Jean-Michel

On Thu, Dec 22, 2011 at 01:46:42PM +0100, Martin Losch wrote:
> Hi,
> 
> I couldn't help it and had to implement a fix to the emissivity mess:
> now seaice inherits all three emissivities (ocean, snow, ice) from exf by default and applies the accordingly in seaice_solve4temp. This changes a lot of results in the verification experiments, but the can (and are, I keep my fingers crossed that I didn't forget one) all recovered by setting the seaice-parameters to the inconsistent old default (5.5/5.67).
> 
> The problem of hardwired 0.97 in seaice_solve4temp can be removed by setting the EXF_LWDOWN_WITH_EMISSIVITY flag. I think setting this flag should be default, or even moreso, the code that applies when this flag is not set should be removed, because it is wrong (inconsistent leads to lw-sources/sinks at the surface, which will hit us, when coupling to an atmosphere).
> 
> I hope that this does not cause too many hick-ups.
> 
> Martin
> 
> 
> On Dec 19, 2011, at 6:32 PM, Martin Losch wrote:
> 
> > I can only say this much: in thsice_get_exf, the emissivities of snow and ice are correctly taken into account, over the ocean it's not, and in seaice only approximately (only correct if you use an "ocean" emissivity/SEAICE_emissivity of exactly 0.97), what a mess!
> > 
> > M.
> > 
> > 
> >> On Dec 19, 2011, at 5:25 PM, Jean-Michel Campin wrote:
> >> 
> >>> Hi Martin,
> >>> 
> >>> I know this long-wave surface thing can be mis-leading 
> >>> (in AIM, I think there is only one surface emissivity, and 
> >>> when using thsice, not sure that we check if everything is
> >>> consistent, but all can be set as run-time params and made
> >>> consistent).
> >>> Since this has been in place for long time, I propose to wait 
> >>> a little bit for others feedback on this issue.
> >>> But it would be good to fix this at some point.
> >>> 
> >>> Cheers,
> >>> Jean-Michel
> >>> 
> >>> On Mon, Dec 19, 2011 at 05:05:39PM +0100, Martin Losch wrote:
> >>>> Of course, with all of this hard coding, I don't think that it is possible to set parmeters so that old results are reproduced. The default for emissivity in seaice is 5.5e-8/5.67e-8 = 0.9700176366...
> >>>> but the hard coded values are 0.97.
> >>>> 
> >>>> Also, there should be (in principle) 3 different emissivities, for water, snow and ice ( and pkg/exf has those). The seaice_solve4temp legacy could accomodate ice and snow emissivities, the development branch only knows one surface type (which has a hard code emissivity of 0.97). I guess Gael and Ian are as naive about all of this as I am (o:
> >>>> 
> >>>> I think the only way out is to change the code and change the results.
> >>>> 
> >>>> Alternatively, we could close our eyes on this and not change anything, but it will come back later!
> >>>> 
> >>>> Opinions please,
> >>>> Martin
> >>>> 
> >>>> 
> >>>> On Dec 19, 2011, at 4:41 PM, Martin Losch wrote:
> >>>> 
> >>>>> Hi Jean-Michel,
> >>>>> 
> >>>>> obviously I also stumbled over this (again). The default of this parameter is 0.97*stefanBoltzmann.
> >>>>> 
> >>>>> Also there is a "0.97" in various places of the seaice-code, that I intend to replace by the appropriate run time parameter (I think I'll need a new one: effectiveEmissivity = 1.).
> >>>>> 
> >>>>> I just talked to a meteorologist (Gert König-Langlo), and he said this:
> >>>>> 
> >>>>> - lwdown (as computed from some atmospheric model) should not be scaled with anything, unless you parameterize it by some near surface air temperature, then you can multiply is with some (unknown) "effective emissivity", order 0.7. 
> >>>>> 
> >>>>> - if the outgoing lw-radition is parameterized as emissivity*stefanBoltzmann*Tsrf^4, then you need to include a term (1-emissitivity)*lwdown as the reflected "Gegenstrahlung" (incoming longwave), so that the net lwFlux = ocean_emissivity*sigma*SST^4 - lw_down + (1-ocean_emissivity)*lwdown=your expression. But it's sufficient to have emissivity=1. (in line with LY04)
> >>>>> 
> >>>>> I guess that with the current implementation in pkg/exf/exf_radiation.F, we can only have ocean_emissivity=1., for pkg/seaice/seaice_solve4temp and seaice_budget_ocean, it's probably all correct, as long as the ocean/ice_emissivity is =0.97. I'll try to fix the seaice-part, so that other emissivities are possible (and will set wrong parameters to reproduce old results in the verification experiments). What about pkg/exf? It should be easy to add a term (1-e)*lwdown, right?
> >>>>> 
> >>>>> Martin
> >>>>> 
> >>>>> PS. Gert also mentioned that the short wave radiation is likely wrong because we never consider multiple reflection/scattering, but I am not so sure about that (because I do not know what atmospheric models do here).
> >>>>> 
> >>>>> 
> >>>>> On Dec 19, 2011, at 3:02 PM, Jean-Michel Campin wrote:
> >>>>> 
> >>>>>> Hi Martin,
> >>>>>> 
> >>>>>> I don't much how thing goes regading long-wave in pkg/seaice,
> >>>>>> but in pkg/exf, when ocean_emissivity < 1, we should also 
> >>>>>> apply the ocean_emissivity factor to the downward long-wave
> >>>>>> flux when computing the absorbed lw flux:
> >>>>>> LW_net(+=up) = ocean_emissivity*( sigma*SST^4 - lw_down )
> >>>>>> 
> >>>>>> In Large & Yeager 2004, eq.12, they recommand to use 
> >>>>>> ocean_emissivity=1 for this reason. And over sea-ice,
> >>>>>> their equation 23 is clear, with ice_emissivity on both
> >>>>>> side. Do we have the same issue in pkg/seaice ?
> >>>>>> 
> >>>>>> Cheers,
> >>>>>> Jean-Michel
> >>>>>> 
> >>>>>> On Mon, Dec 19, 2011 at 06:49:01AM -0500, Martin Losch wrote:
> >>>>>>> Update of /u/gcmpack/MITgcm/pkg/seaice
> >>>>>>> In directory forge:/tmp/cvs-serv17544/pkg/seaice
> >>>>>>> 
> >>>>>>> Modified Files:
> >>>>>>> 	seaice_readparms.F 
> >>>>>>> Log Message:
> >>>>>>> make ocean_emissivity*stefanBoltzmann the default for
> >>>>>>> SEAICE_emissivity, if useEXF; requires setting a wrong parameter value
> >>>>>>> in global_ocean.cs32x15/input.seaice/data.seaice for backward
> >>>>>>> compatiblity
> >>>>>>> 
> >>>>>>> 
> >>>>>>> _______________________________________________
> >>>>>>> MITgcm-cvs mailing list
> >>>>>>> MITgcm-cvs at mitgcm.org
> >>>>>>> http://mitgcm.org/mailman/listinfo/mitgcm-cvs
> >>>>>> 
> >>>>>> _______________________________________________
> >>>>>> 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
> 
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel