OK yep that makes sense, thanks. Yeah I found the ocean files to be offset by 1 day for the monthly averaging (usually centred on day 16 of month, but now on day 17).
I will probably try to run for the rest of Jan then restart again from February.
Do you happen to know if you can change the date on the restart files so that the data is from day 2 but the model treats it as day 1 (restarts from day 1 with the conditions of day 2)?
Yes you could in theory - it seems risky to me though, because each model stores the restart files differently, and so it might be a fair bit of work to do it. I know cice and mom store them as netcdf, so it might not be too hard, not sure about the UM though.
How hard would be be to start the pacemaker on day 1/2/3/4/5 of the run whilst starting the model at day 0 ?
How hard would be be to start the pacemaker on day 1/2/3/4/5 of the run whilst starting the model at day 0 ?
I’m not sure. The way restoring is turned on is setting read_restore_mask= to true in the mom conf file (app/mom/rose-app.conf). I know there are ways of updating certain files during a run, but the process I am aware of will update a every resubmit (which is typically 6M or yearly, not daily). It might be really complicated to update after 1 day then keep going.
Oh - if how the ensemble is started is not critical, it might be simplest to modify a variable in the day 0 restart files. I think you could add/subtract a tiny amount to all the ocean temperatures (for example) before the run starts to make an ensemble?
Yeah I think that might be simpler in end. There are a few threads perturbing initial conditions in the atmosphere