Preindustrial equilibrium temperature in CESM2.2 vs CESM2.1

[czarakas@uw_edu]

Hi,

I'm running a fully coupled simulation at 2 degree resolution (compset: 1850_CAM60_CLM50%BGC_CICE_POP2%ECO%ABIO-DIC_MOSART_CISM2%NOEVOLVE_WW3_BGC%BDRD, resolution: f19_g17), as a hybrid run branching from year 401 of the 2 degree PI control (b.e21.B1850.f19_g17.CMIP6-piControl-2deg.001). I'm using CESM2.2 with the CLM PPE tag - I'm using this model version instead of CESM2.1 because I'd like to compare results with the CLM5 PPE, and the CESM2.1.z series is incompatible with the CLM version being used by the CLM5 PPE. So far I have run this simulation for 40 years, and it is hotter than the PI control I branched from (global mean temperature is 14.5-14.6, compared with ~14.0 in b.e21.B1850.f19_g17.CMIP6-piControl-2deg.001). Is this warmer PI global temperature consistent with model changes from CESM2.1 to CESM2.2, or does it suggest some problem with how I set up the case?

And if the warmer PI global temperature is expected, could I reasonably start a transient historical run after ~40 years partial spin up, which I think is long enough for the surface ocean and LAI to equilibrate? Or would I need to run the whole fully coupled simulation for long enough for the deep ocean to equilibrate to this warmer base climate (which would not be feasible for this project given timing constraints and our computing allocation)?

The case directory for the simulation above is ~/cesm_cases/coupled_PPE/COUP0000_1850spinup and the run script I used to set up the case is here.

Thanks!

 

[hannay]

There is not a magic number of years before starting a transient simulation. What is going to matter is whether the simulation is equilibrated or not.
You need to make sure TS is not drifting and the net TOA radiation is close to zero.
My recommendation is that you plot a timeseries of global annual average of TS, SST and RESTOM (RESTOM=FSNT-FLNT). We can see from these plots is the simulation is equilibrated.

 

[czarakas@uw_edu]

Thanks very much for getting back to me about this Cecile! I posted the time series of global mean TS and RESTOM here. From these plots, it looks like the simulation is not yet equilibrated. And it seems like this is likely due to running these simulations with CESM2.2 rather than CESM2.1. Since spinning up the simulation on deep ocean timescales isn’t feasible for us, I have a few more questions about how to proceed. Thanks again for helping me with this!

(1) Am I understanding you correctly that if we want to branch to a transient historical run, we need to make sure the simulation is first in equilibrium under PI conditions (i.e. it would not be OK to branch to a transient historical run if there is still a significant TOA radiative imbalance in the PI control)?
(2) Do you know if anyone else has spun up fully coupled CESM2.2 (ideally with 2 degree resolution), and if so would it be possible to access the restart files for those simulations?
(3) Does this initial 40-year temperature trend and radiative imbalance suggest that reaching a new equilibrium for this simulation would require spinning up the deep ocean?

Thanks again!