[MITgcm-support] surface treatment and resolutiondependentinstability?

[钱钰坤]

Hi Jody,


First of all, thanks for your JGR reference.


Have you tried the combination of rigidLid=.FALSE. with implicitFreeSurface=.FALSE., or combination of rigidLid=.TRUE. with implicitFreeSurface=.TRUE., if they indeed have any sense.


To my opinion, if rigidLid is turn on, 'free surface' option, whether explicit or implicit, seems meaningless as there should be no free surface (eta is zero all the time or at least do not change with time).  I find it a little hard to interpret the rigidLid equation set in the manual (eq. 2.5-2.7), as there is also an eta variable involved.


I remember that implicit numerical schemes (no more details in my head) generally have relaxed CFL requirement or are even unconditional stable as compared to those explicit schemes.  Maybe you can try the above combinations and see what happen then, if those combinations are allowed by the model (although coarsening the resolution only will not vialate the CFL-like stability requirement).


------------------
 Best regards 
 
Yu-Kun Qian (钱钰坤) 
Center for Monsoon and Environment Research 
 Department of Atmospheric Sciences
School of Environmental Science and Engineering 
 Sun Yat-sen University 
No. 135 Xingang West Road, Haizhu District 
Guangzhou, 510275, P.R. China 
Tel; 020-84115227 
Email: qianyk at mail3.sysu.edu.cn      


 
 
 
------------------ 原始邮件 ------------------
发件人: "Jody Klymak"<jklymak at uvic.ca>;
发送时间: 2017年4月27日(星期四) 晚上11:46
收件人: "mitgcm-support"<mitgcm-support at mitgcm.org>; 

主题: Re: [MITgcm-support] surface treatment and resolutiondependentinstability?

 
Hi Yu-Kun Qian,
On 27 Apr 2017, at  0:37 AM, 钱钰坤 <qianyk at mail3.sysu.edu.cn> wrote:

Hi Jody,


One possible reason may be that as your grid gets larger, you need to increase the viscosity to prevent energy accumulating at grid scale.



Yeah, I tried that, and it does allow the runs to go longer.  But its a barotropic problem that is “fixed" at the same viscosities when I switch to 


rigidLid=.FALSE.,
implicitFreeSurface=.TRUE.,


I’m not clear why the free-surface treatment affects the barotropic stability negatively as we go to larger grid cells.




PS: Previous experience tells me that if you turn on rigidLid (rigidLid=.TRUE.), the overall time for the integration will increase significantly (compared to rigidLid=.FALSE.).  I've known idea why this should be the case.



I think it just converges faster.   I *did* find Dukowicz and Smith, (JGR, 1994) that talks about the difference between the two methods, and they indicate faster convergence.  But I admit that I have not fully digested the material yet.


Thanks,   Jody


------------------
 Best regards 
 
Yu-Kun Qian (钱钰坤) 
Center for Monsoon and Environment Research 
 Department of Atmospheric Sciences
School of Environmental Science and Engineering 
 Sun Yat-sen University 
No. 135 Xingang West Road, Haizhu District 
Guangzhou, 510275, P.R. China 
Tel; 020-84115227 
Email: qianyk at mail3.sysu.edu.cn      


 
 
 
------------------ 原始邮件 ------------------
发件人: "Jody Klymak"<jklymak at uvic.ca>;
发送时间: 2017年4月27日(星期四) 上午6:25
收件人: "mitgcm-support"<mitgcm-support at mitgcm.org>; 

主题: [MITgcm-support] surface treatment and resolution dependentinstability?

 

Hi all,

Just trying to educate myself…

I am running a doubly-periodic domain with some the only external forcing a body force applied to keep the flow moving. The domain is 4000 m deep, with dz=10 m.  Hydrostatic simulations...

I had as my surface conditions:

rigidLid=.TRUE.,
implicitFreeSurface=.FALSE.,
exactConserv=.FALSE.

and this ran fine at 100-m horizontal resolution, and 1000-m.  

Without changing anything about the simulations, except moving to 2000-m, the solution would blow up after a few time steps.  In particular I noticed that `dynstat_eta_mean` would start increasing (or decreasing) before any of the CFL criteria seemed to be going bad.   With that clue in mind I found that:

rigidLid=.FALSE.,
implicitFreeSurface=.TRUE.,
exactConserv=.FALSE.

seems stable.

1) Is this instability something thats known?  It seems non-intuitive to me that as dx,dy gets larger the instability grows.
2) What are the consequences of rigid lid versus implicit free surface? Is there a good primer on choosing these?  I’m ashamed to say that I have simply plugged along w/ the same values for years w/o really questioning them.

Thanks a lot,   Jody

--
Jody Klymak    
http://web.uvic.ca/~jklymak/






_______________________________________________
MITgcm-support mailing list
MITgcm-support at mitgcm.org
http://mitgcm.org/mailman/listinfo/mitgcm-support


_______________________________________________
MITgcm-support mailing list
MITgcm-support at mitgcm.org
http://mitgcm.org/mailman/listinfo/mitgcm-support



 --
Jody Klymak    
http://web.uvic.ca/~jklymak/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20170428/0cfc19dc/attachment-0001.htm>