model blowing up at exact the same model time!!!why??

General scientific issues regarding ROMS

Moderators: arango, robertson

Post Reply
Message
Author
rongzr
Posts: 35
Joined: Mon Jul 17, 2006 4:03 pm
Location: OUC/UMCES

model blowing up at exact the same model time!!!why??

#1 Post by rongzr » Wed May 11, 2011 1:54 am

I'm running a case in a relatively small domain (10kmX10km, 182X182X5).The boundary condition is from the model output of a larger domain, including (u,v,zeta,ubar,vbar,temp,salt).The model blows up after running 2 days. I tried the following things, but the model just blow up at the same model time(2days):
1) reduce the time step.
2) increase the nudging time from 6day to 30dy.
3) switch off the M3 nudging
4) using wet_dry.
5) smooth the bathymetry
The model just blow up at the same step. So I suspect it's due to the boundary file, maybe nan or extremely large value. But Neither NAN or Inf exists.
Any other reason could cause this,especially blowing up the same time step??

User avatar
kate
Posts: 3780
Joined: Wed Jul 02, 2003 5:29 pm
Location: CFOS/UAF, USA

Re: model blowing up at exact the same model time!!!why??

#2 Post by kate » Wed May 11, 2011 6:41 pm

There's more than one way to make the model blow up and you need to figure out which is it for you since it isn't timestep, etc. ncview is great for a quick and dirty look into what's in your files. Do you have a crash record in your restart file from just before things go to NaN? What does it show you?

rongzr
Posts: 35
Joined: Mon Jul 17, 2006 4:03 pm
Location: OUC/UMCES

Re: model blowing up at exact the same model time!!!why??

#3 Post by rongzr » Thu May 12, 2011 1:15 am

The u,v,w velocity show some sawtooth like things well far from the boundary.

feroda

Re: model blowing up at exact the same model time!!!why??

#4 Post by feroda » Thu May 12, 2011 11:03 am

I would try different vertical mixing schemes. Perhaps, it might also helps to reduce the value of RDRG2/Zob according to your bottom friction method.

User avatar
kate
Posts: 3780
Joined: Wed Jul 02, 2003 5:29 pm
Location: CFOS/UAF, USA

Re: model blowing up at exact the same model time!!!why??

#5 Post by kate » Thu May 12, 2011 4:24 pm

Is this a vertical sawtooth at mid-depth? When I see that, it is sensitive to the timestep.

User avatar
susonic
Posts: 160
Joined: Tue Aug 21, 2007 5:44 pm
Location: Jeju National University
Contact:

Re: model blowing up at exact the same model time!!!why??

#6 Post by susonic » Fri May 13, 2011 1:08 am

Are you doing one-way nesting ?

If so, is your initial file also from larger domain?

rongzr
Posts: 35
Joined: Mon Jul 17, 2006 4:03 pm
Location: OUC/UMCES

Re: model blowing up at exact the same model time!!!why??

#7 Post by rongzr » Fri May 13, 2011 1:34 am

It's horizontal sawtooth which is more apparent at surface but could be seen all the depth.
Yes. I'm doing one-way nesting. Both of them are initialized with climatology at zero time. I will try initialized the nest model later than the large model so that the both the initial and boundary are relatively stable and could match each other.

User avatar
susonic
Posts: 160
Joined: Tue Aug 21, 2007 5:44 pm
Location: Jeju National University
Contact:

Re: model blowing up at exact the same model time!!!why??

#8 Post by susonic » Fri May 13, 2011 1:48 am

I mixed up with your answer. I edited my answer.
So, both of your model(large and nested domain) are initialized with climatology T and S? And the nested domain blows up? Can I see figures of your horizontal and vertical
speed at the blowing time?

Post Reply