flux_diurnal namelist switch

[mark@atmos_colostate_edu]

Does the flux_diurnal namelist switch which turns on the Large&Caron (2015) diurnal cycle calculation of SST and salinity still considered a viable model option? I attempted a BHIST simulation using cesm2.2.0 with flux_diurnal set to true but it crashes in the 2nd simulated day. The culprit appears to be numerical instability in the semi-implicit solver for salinity although I'm not 100% certain of that claim.

Thanks,
Mark Branson

 

[sacks]

Unfortunately, my understanding is that flux_diurnal is not really used / supported at this point, particularly now that we couple the atmosphere and ocean hourly rather than daily.

 

[mark@atmos_colostate_edu]

Thanks for that info, Bill. I'm curious as to what model version did the coupling change from daily to hourly? Also, when did the the shr_flux_atmOcn_diurnal subroutine first appear in the code? Or is there any way to track that down?

 

[sacks]

Coupling changed from daily to hourly in CESM2. I don't know when shr_flux_atmOcn_diurnal first appeared, but you can use "git blame" on a file to see the history of each line of code.

 

[mark@atmos_colostate_edu]

Hi Bill.

It may be that I'm still misunderstanding something, but where does the difference in the coupling frequency (daily to hourly) actually manifest itself? I was presuming it would be in the coupling variables within env_run.xml, but if I look at NCPL_BASE_PERIOD and ATM_NCPL for both a cesm2.2.0 case as well as from a cesm1.1.1 case, they both have:

<entry id="NCPL_BASE_PERIOD" value="day" />
<entry id="ATM_NCPL" value="48" />

And sure enough, when I put a print statement in seq_flux_mct.F90 to see how often shr_flux_atmocn is called, it gets called 48 times per day in both cases. So I'm not seeing the difference in coupling frequency changing from daily to hourly.

What am I missing?

Thanks,
Mark

 

[sacks]

I'm not 100% sure, but I think what you want to look at is OCN_NCPL: I think you'll see a difference between cesm2.2 vs. cesm1.1 in that value. In both cases, the atmosphere is still coupled every 1/2 hour, and shr_flux_atmocn is called every 1/2 hour as well, but the ocean coupling fields were only updated daily in older model versions (for performance reasons).

 

[mark@atmos_colostate_edu]

In both of my cases (cesm2 versus cesm1) I have this:

<entry id="OCN_NCPL" value="$ATM_NCPL" />

so they seem identical in that regard as well.

 

[sacks]

Interesting. I guess your version used half-hourly coupling of the ocean already. (That wasn't true for all versions.)