BHIST spin-up for emission-driven runs

[Lakhima]

Hi everyone!

I’m running a simulation starting from 2000 using the BHIST compset to study the impact of satellite-based Modified Combustion Efficiency (MCE)-adjusted fire emissions on CO2, CO, and temperature.
I used year-2000 restart files, and after running 12 years, the model does not seem stable, CO2 and temperature increase unrealistically, and the global-mean RESTOM (FSNT − FLNT) remains strongly positive (~10–15 W/m²).
My goal is to analyze 2012–2024, since we have satellite-based MCE data for those years.

Could anyone please suggest how long I should spin up the BHIST run when emissions are adjusted?
Also, any advice on what I should take care of to improve the model stability for CO2 and temperature in emission-driven runs?

Thanks in advance!
Lakhima

 

[slevis]

Pinging @klindsay and @aswann2 in case they have insights, or they wish to recommend other contacts in the community.

 

[aswann2]

@Lakhima are you using the exact same model version as was used for the original BHIST run that you are using the restart files from? If not then I expect that the model would be out of equilibrium and drift.

If the land is out of equilibrium with the restart files it will need to be spun up before you run a BHIST run.

Can you share the model version for your simulations and for the restart files you are using?

 

[Lakhima]

Thanks for your response!

I'm using CESM v2.1.5 for my current simulation. The restart files I used are from:
/glade/campaign/collections/cmip/CMIP6/restarts/b.e21.BHIST.f09_g17.CMIP6-historical.001/b.e21.BHIST.f09_g17.CMIP6-historical.001.rest.2000-01-01-00000.tar

I'm running a hybrid BHIST case with MOZART chemistry enabled (-chem trop_mozart -co2_cycle), so it’s not the exact same configuration as the original BHIST restart.

Could this version/configuration difference be causing the drift?

Thanks again for your help!
Lakhima

 

[aswann2]

If the CMIP6 version has changes to the land model relative to v2.1.5 that could definitely cause the land model to be out of equilibrium, which would lead to drift. Since you see that you are drifting it likely means that you need to spin up your land before doing your experiments.

The basic procedure is to run an I case in accelerated decomposition mode (AD mode) for several hundred years, then switch back to regular mode for some time. Use the resulting land condition as your initial file for your experiments. See some more info on that in other posts, including this one: Questions on CLM Spinup