When I run openifs I use the following command

srun $exe -e ${expid} -f ${fclen} -t ${timestep}


where expid=guki (e.g. the experiment identifier) fclen=d10 (e.g. 10 days), and timestep=1800 (30 minute time step).

I am now running long simulations. Can I specify the run length (fclen) in terms of months or even years?

3 Comments

  1. Glenn Carver

    Hi Victoria,

    The only units the command line option -f understands is days (-f d10), hours (-f h240), and timesteps (-f t480 with 30min timestep).

    Just for completeness, although the length of run can also be specified in the namelist (fort.4), using the NSTOP variable, in this case NSTOP must be the number of steps. If both NSTOP and command line arguments are specified, the command line takes precedence.

    The code that handles the arguments is in src/ifs/setup/suarg.F90. It wouldn't be much of a change to introduce another time unit but I'd caution against it because then the run length calculation will also depend on the type of calendar the model is using (though only gregorian at present).

    Glenn

    1. Victoria Sinclair

      ok, thanks. My main motivation for this question was that I have been trying to work out how many days I need to run for a 20 year simulation and the leap year issue made things a bit more complicated. Hence, I wondered if there was a clever option in the code to deal with this.

      Victoria

      1. Glenn Carver

        I know the problem.

        I've used this website before to work out the days to run (including leap yrs): https://www.timeanddate.com/date/duration.html