[MITgcm-support] The global flow

Martin Losch Martin.Losch at awi.de
Wed Nov 5 10:17:44 EST 2008


Chuncheng,

please don't send emails with LARGE attachments to this list. Just  
image that everyone, who subscribed to the list now gets an 6MB  
email, which is annoying with a close connections. If you want to  
make large figures or data available, please do so by putting it on a  
web/ftp-server and sending just the link.

You can check, whether your forcing is read by looking at the STDOUT;  
look for your fields names, the model (with the default  
"debuglevel=1") tell you when it reads new data. But looking at your  
"data" file, that does not seem to be the problem. Rather your time  
stepping parameters are the problem:
  deltaTmom = 100.,
   deltaTtracer= 360.,
  deltaTClock = 86400.,
  deltaTfreesurf= 3600.,
The model time is defined by deltaTClock (so the actual time is  
startTime+myIter*deltaTClock). That means every 30 time steps you  
read new data, but the time step of the dynamics and thermodynamics  
equations are very small. What happens is this: you compute the  
velocity tendency g(u) and then step the velocity forward in time  
(not really as Euler step, but this does not affect the principle): u 
(n+1)=u(n) + deltaTmom*g(u), so that after 30 time steps (a model  
month) you have only done 30*deltaTmom=3000 seconds of velocity  
integration. The same is true for tracers etc.

I reccommend that you remove all of these parameters and replace them  
with deltaT = 3600 (or less), that way all time steps are the same.  
The different time steps are implemented so that one can do  
accelerated (asynchronous) spin up integration to equilibrate the  
deep tracer distribution  (Bryan, 1973?) . An example of how that is  
done is in global_ocean.90x40x15; basically you keep all time steps  
the same (a long time step) except for deltaTmom, which should be  
short. But even that will reduce your seasonal variability, and it  
will slow down the fast waves in your simulation, see Bryan (1973?),  
Danabasoglu (2004, Ocean modelling, I think) and the model  
documentation for more details, on the distorted physics that you  
will get with this method.

Martin


On 5 Nov 2008, at 13:14, ouc.edu.cn wrote:

>
>  Hi,
>   I'm simulating the global current with the resolution 1 degree(I  
> just customize the global4x4 in the file /verification),but the  
> results show that there are small seasonal variations(See the  
> attachment,which shows that there is little difference of the  
> current between June and December).Besides,the currents in  
> equatorial areas are very strong(see the attachment).
> The forcings like wind and heat flux are downloaded from ecco- 
> group.com with monthly change.As there are small seasonal  
> variations,I doubt that the model only reads one month data rather  
> than a monthly cycle.I mean,only the wind of January forces the  
> model all the time.Can that be possible? Or other reasons?
> Please help me to overcome the problem.Thank you in advance.
> (There is also a 'data' file in the attachment)
>                                                              Yours  
> Sincerely,
>                                                                   
> Chuncheng
>
>
> [广告] 凯旋珠联“利”合分期付款优惠活动进行中 
> <global_flow.rar>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support





More information about the MITgcm-support mailing list