tracer monotonicity check

[littledddna@gmail_com]

Dear all,

I am currently using ICESM to simulate a colder-than-present-day climate.
However, the model stops running after several months and the error seems to related to tracer monotonicity.

new tracer > tmax
i, j, nt = 10 5 29
new tracer = 1.002136127354356E-011
tmax = 0.000000000000000E+000
ice area = 1.496212204915171E-006

The problem seems like to be produced by a round off error. To overcome this issue, I am thinking either close the 'stop' option and let the model continue to run, or dont check tracer monotonicity. Which option is better? (I have already applied smaller time step)

Thanks very much

 

[dbailey]

We don't see this very often. This is happening in the CICE component. How did you change the timestep? You should really be changing ICE_NCPL = ATM_NCPL. However, if your run is already many years in, you won't be able to to this. Instead you can change xndt_dyn (ndtd in newer versions). Moving this to the CICE forum.

 

[littledddna@gmail_com]

We don't see this very often. This is happening in the CICE component. How did you change the timestep? You should really be changing ICE_NCPL = ATM_NCPL. However, if your run is already many years in, you won't be able to to this. Instead you can change xndt_dyn (ndtd in newer versions). Moving this to the CICE forum.

Dear Dave,

The run just started and I increased xndt_dyn to 4 in the user_nl_cice to change the time step.
ICE_NCPL is set to 144, which equals to ATM_NCPL. So, should I set monotonicity to false in the source code?

Many thanks.

 

[littledddna@gmail_com]

problem solved, the atmosphere is 'too warm' compare to the ocean and sea ice. After I changed CAM initial condition, the run continues.

 

[dbailey]

Curious. Is it leading to excessive winds? Normally warm air temperatures lead to thermodynamic problems and not dynamic problems.