CNDV problem in CLM

[Zhibo Li]

Hi, everyone!

Our group is using CESM1.2.2 to study changes in paleoclimate. We found that turning off CNDV might cause global mean surface temperature to drop by about 2K. The detailed process was that the previous fully-coupled run had come into equilibrium with CNDV on, and then we put results of vegetation from the previous equalized one, mainly from CNDV output, into the surface data and turned off CNDV to restart from the previous run when other parameters were same with the previous one. The result indicated lower surface temperature.

We wonder if it is reasonable to turn off CNDV to restart from a previous one with CNDV on. Is the cooling result because of the seasonal cycle process or other processes in CNDV? Are these two calculation processes in CLM, one with CNDV on, the other with CNDV off, so different that we are not allowed to do the above? For this cooling result induced by turning off CNDV, could you share some possible causes with us? If possible, we still hope to continue run basing on the aforesaid method, could you provide some possible solution?

In addition, we found that the leaf area index and variables related to it in the case with CNDV off were smaller than those in the case with CNDV on, could you share some reasons?

Thanks a lot!!!

 

[crtabor]

Hi Zhibo Li,

I don’t have a solution for you, but I have a few suggestions.

1) You might want to check your vegetation cover in your CNDV run. I have not had much success running CNDV with coupled CESM. The vegetation typically becomes sparse after a few thousand years. How did you configure your CNDV run? However, I would expect less vegetation with CNDV to lead to cooling, not warming, so maybe this is not the problem.

2) Vegetation needs to grow, even with only CN on. Assuming you used CLM4+CN, you might want to run your simulation for a few hundred years to see if the vegetation grows and it warms up.

3) I am not sure how you setup your simulations, but it is possible you have different amount of dust between simulations since erodibility depends on LAI.

As a test, you could try running a simulation using your CNDV outputs with prescribed vegetation (i.e. CN off). That would allow you to check some of these possibilities.

I think the switch from CNDV to CN should work, in theory…Note, I don’t think CNDV was validated with CESM, so it is also possible there is a bug. There are some unusual carbon responses when using CLM4+CNDV.

Hope this helps!
Clay

 

[Zhibo Li]

Hi Zhibo Li,

I don’t have a solution for you, but I have a few suggestions.

1) You might want to check your vegetation cover in your CNDV run. I have not had much success running CNDV with coupled CESM. The vegetation typically becomes sparse after a few thousand years. How did you configure your CNDV run? However, I would expect less vegetation with CNDV to lead to cooling, not warming, so maybe this is not the problem.

2) Vegetation needs to grow, even with only CN on. Assuming you used CLM4+CN, you might want to run your simulation for a few hundred years to see if the vegetation grows and it warms up.

3) I am not sure how you setup your simulations, but it is possible you have different amount of dust between simulations since erodibility depends on LAI.

As a test, you could try running a simulation using your CNDV outputs with prescribed vegetation (i.e. CN off). That would allow you to check some of these possibilities.

I think the switch from CNDV to CN should work, in theory…Note, I don’t think CNDV was validated with CESM, so it is also possible there is a bug. There are some unusual carbon responses when using CLM4+CNDV.

Hope this helps!
Clay

Hi, Clay

Thanks for your good suggestions! I'm going to check it out.

Zhibo