Accelerated spinup mode

[bbucho]

Hi all, I'm trying to spinup a fully-coupled emissions-driven version of CESM using an altered parameter set for land. I turned on the accelerated spinup mode (explained below), but I would appreciate any insight into how long I might need to spin up the model or if there is a better process for figuring out when the model has reached equilibrium.

I initially ran the model for 42 years before learning that I can enter the CLM accelerated spinup mode (accelerated decomposition or AD mode). I then followed the method for spinning up CLM5.0-BGC-crop in the CLM5.0 User Guide. I hybrid branched from the end of the 42-year simulation and turned on CLM_ACCELERATED_SPINUP. I ran AD mode for 6 years but it didn't produce land output, so I took the model out of AD mode by running a single-month startup simulation with finidat in the CLM namelist set to the last restart pointer file of the AD mode simulation (following the CLM5.0-BGC-crop documentation). From this, I looked at the total soil organic matter carbon TOTSOMC which was not reaching equilibrium in the initial 42-year simulation. In the figure below, I plotted total global integrated TOTSOMC with the 42 years of simulation in orange and the AD mode in the green dot. It seems like AD mode pushed TOTSOMC in the right direction and I'm currently extending the post- AD mode simulation to check if TOTSOMC is near equilibrium.

- Is there any rough estimate for how long I might need to run (1) in AD mode or (2) after AD mode to reach equilibrium using an altered parameter set for land?
- Are there better methods to check for equilibrium other than looking at the trends of slow carbon pools such as TOTSOMC?

Any help or insight would be appreciated!

CESM version: release-cesm2.1.5-0-g7a6c5b0
switching on AD mode: ./xmlchange CLM_ACCELERATED_SPINUP="on"

 

[oleson]

Generally, it will take 200-300 years in AD mode to reach equilibrium in the deep carbon pools, and then over a thousand years in "normal" mode (see Figures 1.5.3 and 1.5.4 in the technical note). However, maybe you don't care that much about deep carbon in your simulations and you mainly care about land surface properties that are going to affect climate in your coupled simulations, e.g., leaf area index. Leaf area index will generally come into equilibrium within a few decades.

 

[bbucho]

Thanks for the response! That makes sense. Are there any output variables that we can use in AD mode to track the spinup? Or is there a way to add output variables to AD mode using the namelist?

 

[oleson]

By default, there should be this setup in AD mode:

hist_empty_htapes = .true.
hist_fields_list_file = .false.
hist_fincl1 = 'TOTECOSYSC', 'TOTECOSYSN', 'TOTSOMC', 'TOTSOMN', 'TOTVEGC', 'TOTVEGN', 'TLAI', 'GPP', 'CPOOL', 'NPP', 'TWS'
hist_mfilt = 20
hist_nhtfrq = -8760