[MITgcm-devel] non-legacy seaice

Martin Losch Martin.Losch at awi.de
Tue Jan 31 09:14:49 EST 2012 

Hi Gael,

I have to appologize that my tests are not very systematic so far. We are trying to track down a temperature trend in our Arctic configuration and because I am running out of ideas, I just turned on the non-legacy code. But I am willing to do more systematic tests (especially if they only require running the model for a short time, like for these very negative temperatures).

I use the default SEAICE_area_max = 1. (didn't even realize/forgot that this is around).

I'll try with MCPHEE_OCEAN_ICE_HEAT_FLUX undefined with the non-legacy code.

In general (also with legacy code and MCPHEE-stuff undefined) the surface temperatures are rather low compared to the MOM2 code that we use as a "reference". I am fine with temperatures below freezing, but often they are a low as -5degC. I had expected that with a seaice_gamma_t of 3 days (=259200secs), these should go away. Any similar experience?

I also use allowInteriorFreezing=.true., because otherwise the code temperatures are moved away from the surface and then can only be diluted. I realize that this option makes the surface even colder, but I only observe a small effect on SST.

Martin


On Jan 31, 2012, at 2:57 PM, Gael Forget wrote:

> Hi Martin,
> 
> it is nice to see somebody else than me putting those the various evolution branch codes through sensitivity tests. 
> Running such experiments with more than one setup is needed before we can make informed decisions 
> about the evolution branch defaults. I was not planning on putting more work in seaice_growth.F right now. 
> But I guess, since there is some new momentum, now is the time...
> 
> With regard to runaway temperatures, here are my comments. First, as Jean Michel 
> previously mentioned, Jeff and I have seen that using areamax<1 favors
> runaway temperatures. In my set-up, the problem disappeared by setting areamax to 1. 
> 	You are not using areamax<1 are you?
> Second, MCPHEE_OCEAN_ICE_HEAT_FLUX is likely to increase that issue. Indeed 
> theta<freezing means the ocean heat deficit is not used to freeze (see line 885) unlike
> what is done in the other options. 
> 	Have you tried with MCPHEE_OCEAN_ICE_HEAT_FLUX false?
> 
> Cheers,
> Gael
> 
> ps : sorry Jean Michel I did not reply faster. 
> 
> On Jan 31, 2012, at 7:36 AM, Martin Losch wrote:
> 
>> Hi Jean-Michel and others,
>> 
>> thanks for your comments. I did not plan on using the non-legacy code "by default", I just thought that it is interesting that just turning it one is far from safe.
>> 
>> I'll try with Ian's options. Why isn't this combination the default in pkg/seaice/SEAICE_OPTIONS.h?
>> 
>> M.
>> 
>> On Jan 31, 2012, at 6:56 AM, Jean-Michel Campin wrote:
>> 
>>> Hi Martin,
>>> 
>>> Just a short comment: if Gael does not want to answer, you can
>>> always return to SEAICE_GROWTH_LEGACY version, or, as an other
>>> alternative, switch to pkg/thsice thermodynamics as Jeff did.
>>> 
>>> Cheers,
>>> Jean-Michel
>>> 
>>> On Mon, Jan 30, 2012 at 12:52:13PM -0500, Jean-Michel Campin wrote:
>>>> Hi Martin,
>>>> 
>>>> I heard form 2 other users (Jeff Scott and Gael, but both with
>>>> similar options & params) a similar issue regarding unrealistic 
>>>> surface temp.
>>>> So, I think it's a more general problem that just our set-up.
>>>> 
>>>> Cheers,
>>>> Jean-Michel
>>>> 
>>>> On Mon, Jan 30, 2012 at 05:04:48PM +0100, Martin Losch wrote:
>>>>> Dear sea-ice modelers,
>>>>> 
>>>>> just out of curiosity I turned on the non-legacy seaice code (SEAICE_GROWTH_LEGACY undefine and SEAICE_SOLVE4TEMP_LEGACY undefined and set some other flags that go along with it, see below). To further my boldness, I also turned on:
>>>>> #define MCPHEE_OCEAN_ICE_HEAT_FLUX
>>>>> #define GRADIENT_MIXED_LAYER_TURBULENCE_FACTOR
>>>>> 
>>>>> The solutions that I get for an Arctic domain are comparable but different. The non-legacy code gives generally thicker ice in the interior, which I think is generally good, as the legacy code tends to produce ice that's on the "thin side" compared to other solutions.
>>>>> 
>>>>> However (appart from the fact that there was HEFF~=0 over land), I get very cold surface temperatures (down to -21degC) in some areas (Lena Delta), interesting this starts in June/July when ice should be melting. That's clearly not very physical and I am wondering if others have observed this behavior. Is there a solution to this? Before I try it myself, is it connected to the 
>>>>> #define MCPHEE_OCEAN_ICE_HEAT_FLUX
>>>>> #define GRADIENT_MIXED_LAYER_TURBULENCE_FACTOR
>>>>> flags? (Likely, isn't it?) Or  SEAICE_OCN_MELT_ACT_ON_AREA?
>>>>> 
>>>>> Your experience with this is appreciated ...
>>>>> 
>>>>> Martin
>>>>> 
>>>>> PS. Here are the relevant flags:
>>>>> C--   make conservation of freshwater complete
>>>>> #undef SEAICE_ADD_SUBLIMATION_TO_FWBUDGET
>>>>> 
>>>>> C--   Use the Old version of seaice_growth (close to cvs version 1.70)
>>>>> C     otherwise, use the merged version (with some of Ian Fenty s code)
>>>>> #undef SEAICE_GROWTH_LEGACY
>>>>> 
>>>>> C--   options only available in the merged version (from Ian Fenty s code)
>>>>> #ifndef SEAICE_GROWTH_LEGACY
>>>>> C-    to ensure heat conservation in the coupled ocean-seaice system
>>>>> #define SEAICE_HEAT_CONSERV_FIX
>>>>> C-    to switch on/off open-water freezing contribution to thickness tendency:
>>>>> # define SEAICE_DO_OPEN_WATER_GROWTH
>>>>> C-    ifdef SEAICE_DO_OPEN_WATER_GROWTH then define SEAICE_DO_OPEN_WATER_MELT
>>>>> C     to also allow open-water air-sea heat fluxes melt ice
>>>>> # undef SEAICE_DO_OPEN_WATER_MELT
>>>>> C-    to switch on/off ocean heat contribution to seaice cover reduction:
>>>>> # define SEAICE_OCN_MELT_ACT_ON_AREA
>>>>> C-    to preclude infinitesimal ice concentrations:
>>>>> # undef ALLOW_PRECLUDE_INFINITESIMAL_AREA
>>>>> #endif
>>>>> 
>>>>> C-    to use the MCPhee formula in computing ocean/ice fluxes
>>>>> #define MCPHEE_OCEAN_ICE_HEAT_FLUX
>>>>> #define GRADIENT_MIXED_LAYER_TURBULENCE_FACTOR
>>>>> 
>>>>> C--   Use the Old version of seaice_solve4temp (formerly seaice_budget_ice)
>>>>> C     otherwise, use Ian Fenty s version
>>>>> #undef SEAICE_SOLVE4TEMP_LEGACY
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> _______________________________________________
>>>>> 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

More information about the MITgcm-devel mailing list