Basic question about continue_run

[sur23beeb]

I have what must be a very basic question about the role played by the continue_run flag.

1. I started a branch run from 0265-01 lasting two months after which I had a restart file at 0265-03, consistent with what was in the rpointer.ocn.restart file. For this run, the continue_run flag was set to false. The branch run had a case name that was different from the case corresponding to the restart file at 0265-01.
2. I then made some changes to the tx0.1v3_tavg_contents file by adding more variables to the daily stream.
3. I ran ./case.build --clean-all; ./case.build followed by ./case.submit --skip-preview-namelist.
4. The code continued from 0265-03 (see #1).
5. Midway into the run, I noticed continue_run was still set fo false when I should have changed it to true.
6. My question: how did the code continue running from 0265-03 even when there was no indication it should be a continuation run? (because I forgot to change the flag to true after the first 2-month branch run.)

 

[dbailey]

A true branch run will use the rpointer file as happens in the CONTINUE_RUN. So, this would be the equivalent of a continue run. However, there are sometimes initialization things that happen on a branch that do not on a continue. I would recommend restarting the branch run from 265-01 and copy the rpointer files corresponding to this date.

 

[sur23beeb]

Really appreciate your prompt response. I was thinking about this some more yesterday and thought the reason it continued from 0265-03 even with continue_run=.false is step #3, where I ran ./case.build. I think because I rebuilt the case, cesm probably treated the 0265-03 restart file as the beginning of the branch run, even though the branch began two months earlier at 0265-01. If 0265-03 is treated as the beginning of the branch run, then continue_run=.false. is the proper setting and there are no errors.

 

[sur23beeb]

[The latter part of my reply got chopped off]:

The run that started at 0265-03 ran for a month and I started another run, now ongoing, from 0265-04 (without any new ./case.build) but this time I used continue_run=.true.. It is quite likely (I think) that if I had left continue_run=.false., the continuation run from 0265-04 would have thrown an error.

Does this explanation make sense?

 

[sur23beeb]

A colleague who knows a lot more about cesm than I do just told me the ./case.build thing is not likely to be the reason. So to be safe, I am going to follow the suggestion offered above to just restart from 0265-01. Considering I am only 4 months into this run, it is not too much of a setback.