Opened 8 years ago

Closed 7 years ago

#1053 closed help (worksforme)

Model instability

Reported by: ab728 Owned by: luke
Component: UM Model Keywords:
Cc: Platform: <select platform>
UM Version: 7.3

Description

Hi,

I have a job that crashes within UKCA with the following error message:
Resetting column after128 iterations
NO CONVERGENCE 1

The message is then repeated until NO CONVERGENCE 60. The crash occurs after 18 months. I have been having trouble with the sea ice ancillary file for a long time now but crashes due to this has thus far occurred within the first few timesteps. So I was wondering whether anything else could be causing this. I have yet to try halving the UKCA timestep. Would greatly appreciate any help!

Many thanks,
Tara

Change History (3)

comment:1 Changed 8 years ago by ab728

I forgot to say the jobid is xhkqg and the .leave file is here:
/home/n02/n02/ab728/um/umui_out/xhkqg002.xhkqg.d13103.t201550.leave

comment:2 Changed 8 years ago by luke

  • Owner changed from um_support to luke
  • Status changed from new to accepted

Hi Tara,

One of the interesting things here, assuming it is still related to your sea-ice problems, is that the model dies after nearly 6000 timesteps in that jobstep, rather than immediately on starting a new run.

I will look into this further.

Thanks,

Luke

comment:3 Changed 7 years ago by luke

  • Resolution set to worksforme
  • Status changed from accepted to closed

Hi Tara,

I don't think this is related to the sea-ice, but to a model instability that appears in the second-highest level of the atmosphere, which then causes a grid-point storm. It only seems to affect UKCA jobs with interactive ozone (and some form of stratospheric chemistry).

Including the branch 

fcm:um_br/dev/luke/vn7.3_TOA_Temp_Fix/src

at revision 12609 seems to solve this problem.

Thanks,
Luke

Note: See TracTickets for help on using tickets.