Difference between CESM 1.0.4, 1.0.5, and 1.0.6

[shineng_hu@gmail_com]

Hi,I have got a long reference run (B_1850_CN, CESM 1.0.4) on our local machine. Now I am trying to run a set of coupled simulations on Yellowstone with the restarting files from the previous reference run. But I found the earliest version of CESM available on Yellowstone is 1.0.5. I wonder what are the main updates from 1.0.4 to 1.0.5 (and 1.0.6)?Also, does the tropical climatology (both atmosphere and ocean) look similar among those three versions (1.0.4, 1.0.5, and 1.0.6)? I am mainly looking at the tropical mean state and El Niño variability.Thanks a lot in advance!Shineng 

 

[jedwards]

1.0.6 should have the same climate as 1.0.4 and should be able to use the restart files.

 

[shineng_hu@gmail_com]

Thank you!If I understood it correctly, do you mean the climatology doesn't change from 1.0.4 to 1.0.6?Now I have got two simulations (B_1850_CN) starting from exactly the same initial conditions, but with different machines and model versions:Case 1: CESM 1.0.4 on our local parallel machineCase 2: CESM 1.0.6 on YellowstoneBut the outputs differ a lot between them. Are they supposed to have exactly the same outputs if run for two years? If not, is it the model version, machine, or both of them that will affect the integrations?

 

[shineng_hu@gmail_com]

Just to clarify, when I say "the same initial conditions", I mean the branch run uses the same restarting files for all components. :-)

 

[hannay]

The climate should be the same but the solutions are expected to be different because of variability. If you take a 20-year average, you shoudl get similar solutions. 

 

[santos]

"If not, is it the model version, machine, or both of them that will affect the integrations?"It will probably be both. A different compiler or compiler version, or even the same compiler version on different hardware, will produce slightly different results. It is very rare for CESM runs on two different machines to reproduce each others' results exactly. Between CESM 1.0.4 and 1.0.6 we also have likely changed some default compiler flags, and possibly made code changes (for portability or minor bug fixes) that cause the solutions to diverge.But as Cécile said in the above post, none of these changes should affect climatology.