[MITgcm-support] NaNs in the DIC package

Stephanie Dutkiewicz stephd at ocean.mit.edu
Tue Aug 25 14:46:21 EDT 2015


Hi Tobia -
Usually NaN in the DIC fields comes from the pH solver struggling with T 
or S reaching strange values -- so worth check both those fields to see if 
there is something wrong there.

You can try defining DIC_BOUNDS in DIC_OPTIONS.h -- this is a hack to try 
and make sure that the values going into the pH solver are within 
"reasonable bounds" -- but it is a hack -- it is worth checking why the 
fields (T,S or any of the DIC, ALK etc) are going out of bound as this 
might suggest other problems in the simulation.

steph

On Tue, 25 Aug 2015, Tudino, Tobia wrote:

> Dear all,
> I am currently run the MITgcm at a 2.8 resolution with the GCHEM package included. This run is based on a
> previous setting constructed on an idealised forcing field. 
> 
> From this starting point I have changed the forcing fields hoping to reach an equilibrium comparable to
> observations.
> 
> However, for some reasons, after few decades of SPIN-UP runs, the DIC fields disappeared leaving all NaNs.
> 
> Thinking that this could be related to the salinity relaxation time, I have doubled this parameter with a very
> little variation in the final outputs.
> 
> Any additional suggestions?
> 
> Thanks in advance
> 
> Tobia
> 
> Tobia Tudino
> tt282 at exeter.ac.uk
> PhD Student
> Geography
> College of Life and Environmental Science
> Skype: tobia.tudino
> 
> 
> 
> 
>


More information about the MITgcm-support mailing list