[MITgcm-support] [EXTERNAL] SEAICE, very thick ice

Mon May 18 14:55:36 EDT 2020

Hi Stanislav, what boundary conditions are you using?  Where do they come from, what is the space and time resolution at which you apply them?  One issue with sea ice boundary conditions is that even if you apply boundary conditions from an identical but larger-domain simulation, you can still get artifacts at the boundaries, convergence or divergence, if you are not applying full temporal frequency or if there are nonlinearities that cause the regional simulation to diverge relative to the larger-domain simulation.

Clement Bertin and other colleagues, cc-ed, are exploring options for sea ice boundary conditions for a McKenzie River Delta set-up and facing some of the same issues that you are.  Ultimately, we are wondering if some region of relaxation around the edges might be the best practical solution to this problem.

Dimitris

> On May 18, 2020, at 11:19 AM, Yilang Xu <yxu at whoi.edu> wrote:
> 
> Hi Stanislav, 
>  
> In my simulations, the major change is OBCS_SEAICE_AVOID_CONVERGENCE. This option will help to advect sea ice across the boundaries more freely, and reduce the noisy convergence that affects both the boundary and interior values. However, as the code says, the boundary sea ice conditions will be less accurate. I cannot remember my OBCS_SEAICE_COMPUTE_UVICE results, but it is definitely worth a try. 
>  
> Changing SEAICE_no_slip did not affect my results. But my setup is different from yours. You might want to do another test later to confirm it. 
>  
> Best, 
> Yilang 
>  
> From: MITgcm-support <mitgcm-support-bounces at mitgcm.org <mailto:mitgcm-support-bounces at mitgcm.org>> on behalf of Stanislav Martyanov <martyanov.sd at gmail.com <mailto:martyanov.sd at gmail.com>>
> Reply-To: <mitgcm-support at mitgcm.org <mailto:mitgcm-support at mitgcm.org>>
> Date: Monday, May 18, 2020 at 13:38
> To: <mitgcm-support at mitgcm.org <mailto:mitgcm-support at mitgcm.org>>
> Subject: Re: [MITgcm-support] SEAICE, very thick ice
>  
> Hello, Martin and Yilang!
>  
> I thank you for your advices and I will report of the results after I try these options.
>  
> I will start from setting SEAICEpressReplFac=1 and see if it helps. After I will try OBCS_SEAICE_AVOID_CONVERGENCE and OBCS_SEAICE_COMPUTE_UVICE, together or one after another.
>  
> Previously I was thinking about OBCS_SEAICE_AVOID_CONVERGENCE but did not want to use any 'old'  implementations, whatever it means.. Also I was looking at the options SEAICE_no_slip option and SEAICEsimpleRidging..
>  
> Yilang
> 
> 
> 
> 
> Did you use anything besides OBCS_SEAICE_AVOID_CONVERGENCE to handle such situation in your simulations? Did you play with SEAICE_no_slip option? If yes -  does it really affect the results?
>  
> Martin
> >> This is only implemented for OBCS_UVICE_OLD  
> Can't I just use  OBCS_UVICE_OLD in the current version (c67p)?
>  
> In our previous simulations (another model configuration) we did not set uice and vice at the OBs at all, and we also did not use OBCS_SEAICE_COMPUTE_UVICE (checkpoint c65z). May be such approach can help in the current situation, or it does not really matter?
>  
> Stanislav
>  
>  
> пн, 18 мая 2020 г. в 19:41, Martin Losch <Martin.Losch at awi.de <mailto:Martin.Losch at awi.de>>:
>> Hi Stanislav,
>> 
>> the open boundary conditions for sea ice were probably not made for this type of situation (have them in the middle of an active ice region where ice is confined by topography). 
>> 
>> One thing that will always reduce excessive ridging is turning off the default pressure replacement method (set SEAICE_pressReplFac = 0.). With pressure replacement, stagnant ice tends to be pushed and piled up in corners and bays, because the replacement pressure (so the ice strength) reduces dramaticall for small strain rates (when nothing is moving anymore).
>> 
>> other than that, I would probably try the option “OBCS_SEAICE_COMPUTE_UVICE”. This is only implemented for OBCS_UVICE_OLD, but you can easily transfer that implementation of a Neumann boundary conditions (du/dn = 0 across the boundary) from seaice_adjust_uvice.F to seaice_apply_uvice.F In this way, the ice velocities will be determined only by forcing in the interior. 
>> 
>> But you should maybe first check, if you have excessive flux of ice over your boundaries. At least the northwest corner doesn’t look like that in your plots.
>> 
>> Martin
>> 
>> > On 18. May 2020, at 17:38, Stanislav Martyanov <martyanov.sd at gmail.com <mailto:martyanov.sd at gmail.com>> wrote:
>> > 
>> > Hello everyone!
>> >  
>> > I get very thick sea ice at several locations in the high-res Kara Sea model domain (horizontal resolution is 500-1200 m). Interestingly, the general pattern of sea ice concentration and thickness is OK compared to MERCATOR ocean reanalysis, but in some regions the modeled HEFF is reported by the monitor to be about 15 m after 5 months of integration (starting from realistic initial conditions). I have attached the figure to clarify what I mean and where it happens.
>> >  
>> > I suspect the ridging process near the open boundaries to be involved in this situation, but I do not know what can be done with it.
>> >  
>> > Such HEFF also produce problems in vertical layers when useRealFreshWaterFlux is TRUE, but I checked that such too thick ice is also produced when useRealFreshWaterFlux is OFF, so it is not a cause (at least, not the main cause).
>> >  
>> > Unfortunately, there exist too many options to just iterate through all of them.
>> >  
>> > May be you can advise what can be done to mitigate such overestimated HEFF? My preliminary suggestions are as follows, but I am not sure about them:
>> >  
>> > - to switch off the uice and vice at the OBCS?
>> > - to set SEAICE_no_slip=TRUE to reduce the sea ice velocities near the coasts to prevent enhanced ridging?
>> > - to use sponge at the open boundaries?
>> > - there was a CPP key OBCS_SEAICE_AVOID_CONVERGENCE, but it is only for OBCS_UVICE_OLD which, in turn, is undefined by default and even may be not implemented, as stated.
>> >  
>> > Any advice is very welcome!
>> >  
>> > Stanislav
>> >   
>> > PS: I use a SEAICE package configuration which is very close to the default one. What I have changed is:
>> > #define SEAICE_VARIABLE_SALINITY
>> > SEAICEadvScheme = 33,
>> > SEAICE_multDim = 1,
>> > <SEAICE_OPTIONS.h><data><CPP_OPTIONS.h><OBCS_OPTIONS.h><data.obcs><FIGURE HEFF.png>_______________________________________________
>> > MITgcm-support mailing list
>> > MITgcm-support at mitgcm.org <mailto:MITgcm-support at mitgcm.org>
>> > http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support <http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support>
>> 
>> _______________________________________________
>> MITgcm-support mailing list
>> MITgcm-support at mitgcm.org <mailto:MITgcm-support at mitgcm.org>
>> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support <http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support>
> _______________________________________________ MITgcm-support mailing list MITgcm-support at mitgcm.org <mailto:MITgcm-support at mitgcm.org>http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support <http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support>_______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org <mailto:MITgcm-support at mitgcm.org>
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support <http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.mitgcm.org/pipermail/mitgcm-support/attachments/20200518/95a1b14c/attachment-0001.html>