Spinning up carbon on land

I'm using CESM2.1.5

I'm trying to spin up the land carbon cycle in an I-case (CLM only) to eventually use as an initial condition for land in a B-case in emissions-driven mode.

First I ran an F-case to save coupler history output.
Then I ran 300 years of an I-case with AD mode on, driven by coupler history output. Things looked fairly equilibrated, but I realize that I may not have reached equilibrium with AD mode turned on. (see plots below)

Next I ran a new I-case with AD mode off, using the land restart as finidat (i.e. initial condition for land from the AD mode on case).
The carbon does not look equilibrated after 150 years especially for soil carbon, but curiously is equilibrating in the opposite direction from the way it was going with AD mode on. (i.e. soil carbon was decreasing in AD mode on, increasing with AD mode off). (see plots below) I tried both a hybrid branch and a startup run, they seemed to have equivalent carbon behavior.


Some questions:
1. Since AD mode and non-AD mode are going in opposite directions does running more time in AD mode help solve this problem?

2. How long approximately do you typically have to run with AD mode off after AD mode was on?

2. Is there something else I'm missing about how to do this spin up?

300 years for AD sounds about right. I haven't thought much about the different trends in AD and pAD, but as shown below, it seems typical. Top plot is from a recent AD and bottom from a recent pAD. It does take a long time to reach the strict equilibrium criteria we use, see lower left plot, we use 3% disequilibrium as our threshold. In this case it looks like about 1200 years. The global TOTSOMC comes into equilibrium for pAD after about 400 years, but it takes a lot longer for the 3% disequilibrium to be reached, mostly due to the high latitudes.

@oleson - thanks so much for your response! I have now run my post-AD spinup out more than 1400 years. Here is what I'm getting (using your plotting scripts). My total land area in disequilibrium is very high relative to yours! It seems stuck at 20% of land area, while yours drops steadily lower. Can you let me know what threshold you used as your equilibrium criteria for TOTECOSYSC?

We normally use 3-5%. I'm not sure why yours is plateauing at around 20%. Particularly since your global TOTECOSYSC seems to be in equilbrium. I see that you've selected 20 years for subper in the script (e.g., bottom right plot shows 1421-1401). But it looks like you are repeating coupler history forcing for years 10-30, which is 31 years. So you should set subper to 31 in your script. However, I ran my version of the script on your simulation with subper set to 31 and got something similar to yours, about 26% in disequilibrium.
Your plots are "noisier" than mine are. For example, my GPP is pretty steady at 99.5 while yours is bouncing around by plus/minus 0.6 or so. I wonder if there is something in your forcing that is not repeating exactly every 31 years. If you look at the two bottom right plots, there are regions that seem to flip between negative to positive and vice-versa between 20 year periods, e.g, central africa and Sahel, and southern south america. Normally, those regions should come into equilibrium in a reasonable amount of time. Generally, the last few hundred years of the spinup is to get the cold latitudes into equilibrium. I guess I would check your atm logs to see if you are indeed repeating the forcing every 31 years and if there is anything else that might be inducing variability into the model.

Well, apparently, I can't subtract. You are looping over 21 years, not 31 years. So if you set your subper to 21 instead of 20, you'll see that you are well within equilibrium. Only 1.22% is in disequilibrium.

Thanks for your help @oleson! When I set the subper to 21 years I see that it took 379 years to meet the equilibrium threshold for TOTECOSYSC!