[MITgcm-support] LLC2160 sea-ice problem
Menemenlis, Dimitris (329D)
Dimitris.Menemenlis at jpl.nasa.gov
Thu Dec 18 19:55:15 EST 2014
Dear Alex, I transfer your question to MITgcm support mailing list.
Maybe others can help.
The llc_2160 version that I use, including all input files and model parameters
are listed and described here:
That version and set of parameters runs fine for at least three years.
Are you trying to run the global llc_2160 simulation or did you
extract the Arctic region only?
What are you doing for boundary conditions?
For llc_270, instructions are here:
It looks like you might have some surface forcing problem for your
ice to melt so quickly. What surface boundary conditions are you using.
> On Dec 18, 2014, at 4:07 PM, Alexander Slavin <alexander.slavin at mcgill.ca> wrote:
> Hello Dimitris,
> Happy upcoming holidays!
> Thank you again for sharing LLC setup with us.
> Can you give me advise on several problems I have with it. Namely:
> I compiled llc_270 and llc_2160 setups (ftp://ecco2.jpl.nasa.gov/data1/llc/). Unfortunately, after several tens/hundreds of time steps I constantly experiencing blowup/NaN output problems.
> 1. For llc_2160 runs lasts for 6 hours of model time maximum (depending on time step). I tried to reduce time step down to 2 sec without success.
> As I found out explosion happened in one place or another. Do you by chance have working version of llc_2160 or llc_4320? Or maybe advise me which grid and initial data I could use?
> llc_270 version also blow up with Extreme temperatures, if you could advise which grid to use with it, it could be good as well, cause I would like to test smaller version first on our cluster.
> 2. With llc_2160 I have strange problem with sea-ice concentration in Arctic after some time steps (see attached files) although initial conditions looks fine.
> I tried to adjust SEAICE_mcPheePiston = dZ/dT (10 m/10 sec time step = 1 for example) without success. I tried different MITgcm versions, different options in data.seaice but problem still exists.
> So I am thinking to start with very beginning again (I already have to many versions where I tried to solve those problems). I would very appreciate if you please could provide llc version which is most stable.
> Thank you,
More information about the MITgcm-support